What is the “Status questionis” of the e-nose. The future of ...... (6) Figure 5: G-protein...

1

What is the “Status questionis” of the e-nose. The future of artificial noseJournal of Bioscience & Biomedical Engineering Research Article

Volume 2 | Issue 2J B & Bio Engine; 2021 www.unisciencepub.com

Clarós P1*, Dąbkowska A2, Clarós A1.Portela A3, Pérez R3, Marimon X3, Gabarró MI3 and Gil J.

ISSN 2693-2504

*Correspondence authorsClarós PClarós Clinic. Barcelona, SpainClarós P: orcid.org/0000-0002-7567-0370Clarós A: orcid.org/0000-0001-6084-3470

Submitted : 5 May 2021 ; Published : 20 May 2021

1Clarós Clinic. Barcelona, Spain.

2Scholarship Clarós Foundation. Barcelona, Spain.

3International University of Catalonia. Bioengineering Institute of Technology. Barcelona. Spain.

AbstractAn artificial nose (e-nose) is a multipotential electronic device, based on various sensors with the ability to recognize different odours, in the same way that the human olfactory sense does. An updated e-nose system will allow us to detect different oncological and/or degenerative diseases of the human being that today are diagnosed late. Other options would include providing specific information on the quality and condition of food, analysing and detecting the degree of environmental pollution, analysing perfumes and their essences, determining the composition and characteristics of certain beverages such as wine, tea, oil, cocoa and other products. The application of new technologies, such as artificial intelligence and nanotechnology, make it easier to distinguish many different odours in less time. In this paper, we have made a current investigation of the different types of e-nose existing today.

Aim of the studyCurrently nanotechnology, artificial intelligence and computer science are tools that have revolutionary possibilities for the construction of a new e-nose device and progress over the current scopes. If we manage to unite the advantages provided by each of these new technologies that we have mentioned, we will be able to build a very useful device applicable in various fields such as health, food and beverage industry, perfumes and environment. Therefore, the objectives we have in this study are the following:

• To learn where artificial nose technology stands and what has been developed so far. Based on these findings, ask ourselves the following question: Is it possible to achieve an effective functioning or is it an unreachable project?

• To compile the majority of scientific articles published mainly in the last 10 years with examples of the use that has been made of artificial noses in the measurement of volatile compounds in the different fields mentioned.

• In the same way to gather the information published in the last years in relation to the usefulness, existence in the market and purposes of equipment that can measure the olfaction in the human being, what we will call the Smell-o-meter or olfactometer for human use.

Material and MethodsIn the first part of this research we will gather most of the information existing so far in the international bibliography, as well as the achievements and utilities obtained to date. Following we will analyse all the new concepts related to e-nose devices that exist on sensors, gas chromatography, nanotechnology application, electronic engineering, materials and techniques as preliminary ideas for the development of the devices.

Keywords : artificial nose, e-nose, electric nose, nanotechnology, artificial intelligence

2


AcronymsMOSFET : metal oxide semiconductor field effect tran sistors MOS : metal oxide semiconductors (Taguchi-type sensors) QMB : quartz microbalance SVM : support vector machine SLR : sparse logistic regression ANN : artificial neural networkPCA : principal component analysis CDA : canonical discriminant analysisCRBM : continuous restricted Boltzmann machine LDA : linear discriminant analysis HCA : hierarchical cluster analysis PLS : Partial Least SquaresLCD : linear canonical discriminant analysis CVV : cross validation valueDFA : discriminant factor analysisMLR : multiple linear regressionCA : cluster analysisk-NN : k-Nearest NeighbourBPNN : back propagation neural net-work SVD : single value decomposition2-NM : 2-norm methodMDM : Mahalanobis distance method RBF : radial basis functionSOM : self-organizing mapPNN : probabilistic neural network

Structural Organization

What is an E-Nose?The sense of smell, same way as it occurs with the sense of taste, is a chemical sense, meaning that it can detect chem-ical substances in the environment. “Smelling is a complex process.” Vaporized odour molecules floating in the air reach the roof of the nose cavity, where the olfactory epithelium is located, and they dissolve themselves in the mucus. The spe-cific molecules of each odour are spread out through the mucus or transported by special proteins, attached to hair-like struc-tures known as cilia. The cilia are a part of the receptor olfac-tory cells which contain the molecular machinery of olfactory transduction, including receptors, effector enzymes, and ion channels.

In the human body, it is estimated that there are around 12 mil-lion olfactory receptor cells. They can detect approximately 10,000 odours. The odour molecules bind to the receptor cells themselves which will then generate a signal or impulse that will subsequently send this information to the olfactory bulbs located at the back and roof of the nose. These stimuli are then transmitted to the limbic system structures, known as the most primitive brain centres that influence emotions and memories, or to the neocortex, either the anterior olfactory nucleus or/and the piriform cortex, which are referred to as “higher” centres, with the ability to modify their conscious incorporation and interpret the electrical signals received (Fig.1, 2). The higher the concentration of volatile chemical compounds, the stronger the signal sent by the receptor cells to the olfactory bulb (1).

Figure 1: Anatomy of the human olfactory system

https://www.semanticscholar.org/paper/The-Human-Sense-of-Olfaction-Walliczek-Dworschak-Hummel/1770676b1fe8a281e-4ae4b1fca3b7110f6c23f79

3


Figure 2: Schematic of the nasal respiratory epithelium, olfactory epithelium, and the olfactory bulb. (2)

The rhinoencephalon is the part of the brain related to smell and is included in the limbic system also called the “emotional brain”. In humans, olfaction has regressed significantly, although it is still very important, but with less sensitivity. The human being can capture 7 primary odours: camphor, musk, flowers, mint, ether, pungent and rotten, and correspond to the seven recep-tors located in the mucosa of the nostrils; however, it is estimated that there are more than one hundred primary odour sensations (Fig. 3).

Figure 3: Olfactory system structures in the human being

https://slideplayer.es/slide/14488174/. (3)

4


In 2004 Buck and Axel received the Nobel Prize in Physiology and Medicine for describing the structures and genes involved with the olfactory receptor. An olfactory receptor can be sensitive to different odorants, often from different chemical compounds (e.g. alcohols and aldehydes). At the same time, one odorant can stimulate different olfactory receptors and these respond with different intensities to various volatile compounds. Malnic claims that “an individual odour is encoded by several different tuned olfactory receptors.” He called the phenomenon a “combinatorial odour code” (4), which has been confirmed in different electrophysiological experiments.

The olfactory system of mammals has the ability to discriminate thousands of volatile chemicals with different odours. These are the so-called odour receptors (ORs) that are able to recognize different odours, but at the same time make the same odour recognized by several ORs and also different odours are recognized by multiple ORs and sometimes several types of odours are recognized by different combinations of ORs. This means that the olfactory organ applies a combinatorial OR coding system to encode different odour identities. When slight alterations occur in the composition or concentration of an odour, changes in its “code” can occur, which would condition the perceived quality of the odour. The odour threshold of any person is the concentration of odorant it is able to smell.

Olfactory receptors (OR) belong to the G protein-coupled receptor [G-protein-coupled receptor (GPCR) group] (Figs.4-5). They consist of seven G-protein-coupled receptors (GPCRs), also known as seven-step transmembrane domain receptors. G-protein-coupled receptor (GPCR), also called seven-transmembrane receptor or heptahelical receptors, is a membrane-localized protein of extracellular substances and transmits signals from these to an intracellular molecule called G-protein (guanine nucleotide-binding protein). Some of their elements are highly variable, while others are relatively conserved. Buck described an olfactory receptor similar to an amino acid ligand-binding pocket, formed by the transmembrane helices, which is able to recognize chemically diverse volatile compounds. A ligand (from Latin ligandum, ligand) is a substance, small molecule that sends a signal by binding to the active center of a protein. Proteins or protides are macromolecules formed by linear chains of amino acids. Proteins are made up of amino acids and this sequence is determined by the nucleotide sequence of its corresponding gene. Genetic information determines which proteins a cell, tissue and organism has (5).

Figure 4: Seven-transmembrane receptor.

https://www.nature.com/articles/nrm908 (6)

Figure 5: G-protein Coupled Receptors (GPCR), also also called seven-transmembrane receptor or heptahelical receptor.

ht tps: / /www.sciencedirect .com/science/ar t ic le /pi i /S1877117316300059 (7)

What is interesting, chemically-irritating substances such as those found in hot and spicy foods or ammonia and menthol, stimulate receptors of the trigeminal nerve (V cranial nerve) located in the mucosa of the posterior part of the nasopharynx, and not the olfactory nerve (I par cranial nerve) (5).

The traditional technique for the measurement of odours is based on the olfactometry. This test is performed by diluting a specific sample of an odorous substance in a portion of odourless air to a certain concentration and then varying the concentration. It can be started with high concentrations and

5


lowered until 50% of the people tested cannot distinguish odorous samples from odourless samples. The odour detection threshold is measured in odour units per cubic meter (OUE), starting with 1 OUE/m3, 1 OC or 1D / T (8). The “Odour Unit per cubic meter” (OUE/m3) is the unit of measurement.

Many odour analysis investigations have been carried out using gas chromatographs coupled to mass spectrometers (GC-MS), which separate and analyse molecules in the gas phase (Fig.6). This procedure is widely implemented in many laboratories and is relatively simple to use. By means of this technique, once the components of the sample have been separated, they are ionized by electronic impact and injected into the mass spectrometry analyser, the simplest and most common being the single quadruple. “The single quadrupole is used in waste gas and real-time gas analyser, plasma diagnostics and SIMS surface analysis systems. (9). However, this method requires equipment that is quite expensive, time consuming, and requires expert knowledge to perform the assays, which makes its widespread use difficult. In addition, it should be taken into account that the samples to be analysed are highly complex as they may contain even hundreds of different volatile compounds (CVOs).

Figure 6: The mass spectrometer of a single quadrupole with an electron impact ionizer (Single Quadrupole Mass

Spectrometry).

https://www.waters.com/webassets/cms/library/docs/local_seminar_presentations/GE_Events/GE%20Pharma%20and%20Biopharma%20Forum%202018/Fundamentals%20of%20Single%20Quad%20MS_%20Noud%20van%20der%20Borg_Waters.pdf

Due to the above, the idea of creating a device that could mimic the human sense of smell arose. To this end, several different models of artificial nose have been designed, which correspond to what is known as an electronic nose, commonly referred to as an “e-nose”. The objective is to enable an electronic nose to recognize different odours present in the environment that give very specific information that is defined as an electronic fingerprint (Fig.7). The device offers a wide range of possibilities for odour identification and can be used in a variety of fields, including environmental monitoring, diagnosis of serious diseases whose early detection would be a major therapeutic advance, aspects of public safety, agricultural production, food industry, viticulture, oil processing, cocoa, tea, etc. Persaud and Dodd of the University of Warwick in Great Britain in 1982 were the pioneers in the development of the artificial nose (10). From these bases our projects are to build new and more modern systems.

Figure 7: Chemical digital fingerprint

http://nano.caltech.edu/research/nems-olfaction.html https://www.purdue.edu/uns/x/2008b/080807CooksFingerprints.html

Cook’s Laboratory imatge at Purdue University. Demian R. Ifa et al. Latent Fingerprint Chemical Imaging by Mass Spectrometry. 2008.

Hu described the e-nose as an instrument that contains interactive sensor-arrays that react to compounds on the sensitive materials’ surface accompanying certain actions such as adsorption, desorption and/or reversible reaction, etc. “Desorption is a phenomenon by which a substance is released from or through a surface. The process is the opposite of sorption. This occurs in a system that is in the sorption equilibrium state between the bulk phase and an adsorption surface”. The specific responses detected in the sensor array are recorded and transformed into readable digital values. The information received from the sensors can be stored in a computer and then used to build statistical models based on deep learning to achieve recognition of different odours (11). The comparative schematic model between the mammalian nose and the electronic nose is presented in the graph below in Fig. 8. In Fig. 9, an example of a sensor array is presented, according to Pérez, Marimón and Portéla (UIC group).

Electronic noses are composed of a group of non-selective chemical sensors (Fig 9). This means that each sensor is sensitive to a large amount of volatile compounds, but the sensitivity of each element of the sensor matrix is different (12).

6


Figure 8 : Schematic diagram comparing a mammalian olfactory organ with an electronic nose (according to Pérez R, Marimón X, Portela A. Artificial nose. UIC Barcelona).

Figure 9 : Sensor array from (Pérez R, Marimón X, Portela A. Artificial nose. UIC Barcelona).

Artificial nose devices offer rapid, non-invasive and cost-effective (potentially inexpensive) examinations to interpret odours that employ real-time detection and analysis of volatile compound samples. However, this field of odour detection and analysis research is sparse and faces many technical difficulties, which we hope to be able to resolve shortly. Our research team is going to work to solve these challenges by building an e-nose for common use. One of the problems that we find most difficult to solve is the low concentration of the volatile particles (CVOs) to analyse, or the different environmental conditions in which they are found such as variation of temperature, humidity, etc., as well as the specific conditions of the sensors (duration, selectivity, ability to detect different odours, repeatability of the results) and at the same time, the ability to detect multiple different odours coming from several substances of different chemical groups.

In the literature research there are multiple general descriptions of e-nose. Therefore, before advancing on our own project we wanted to elaborate this review of the latest advances in e-nose, the limitations of the system and search for the usefulness of artificial noses in a large number of areas.

7


How Does An E-Nose Work?The electronic nose is based on the use of a sensor matrix that generates the appropriate pattern after exposure to the odorant, depending on the type of smell. The next step is training the device to interpret and distinguish different odours and odorants and recognizing new patterns of smells. Generally, the electronic nose system is divided into three segments: 1-a sample transporting unit, 2-a detection unit, and 3- a data processing system.

The first of these, which is extremely important in the e-nose, is the transport system, which is designed to transfer volatile particles from the emitting source to the matrix of the sensors themselves located in a chamber under constant temperature and humidity conditions. Alteration in the maintenance of these parameters can lead to adsorption of other odour molecules. The second, the detection unit, contains an electronic converter that transforms a chemical signal into an electrical signal, amplifies and conditions it, and with a digital converter, converts the signal from analogue to digital. Finally, the third one is the one that makes the final calculation and data processing, which is performed in the computer microprocessor, reads the digital signal and performs statistical analysis to recognize or classify the analysed sample (13). An artificial nose is capable of converting various physical odours into measurable systems, using non selective sensors and transforming the electrical signal into a correct response from the examined sample.

So far the multiple electronic noses that have been designed employ various detection systems applying to different volatile substances. Wilson and Baietto used a wide variety of odour sensors (14). The most common sensors in artificial noses are metal oxide type sensors (15), (16), (17); semiconductor polymer sensors (18), (19); conductive electro-active polymer sensors (20), (21), (22), (23); optical sensors (24); surface acoustic wave sensors (24); and electrochemical gas sensors (24). Apart from those previously mentioned for gas detection, gas sensitive - with sensitive field effect transistors - and quartz microbalance sensors (Quartz Microbalance Sensors, QMB) can also be applied. The most promising technologies appear to be Micro-Electro-Mechanical Systems (MEMS) using nanotechnologies.

Tables with types of sensors and usesIn the electrochemical gas sensor, interactions between gas molecules and sensor coating materials modulate the electrical passage through the sensor, which is then detected by a transducer, which converts the modulation into a recordable electronic signal (24). Among the electrochemical sensors are metal oxide gas sensors, metal-oxide semiconductor sensors, field-effect transistors, polymer gas sensors, acoustic wave gas sensors, quartz crystal sensors, microbalance sensors, surface acoustic wave devices, field-effect gas sensors, electrochemical gas sensors, sensors, pellistors and fiber optic gas sensors. “A pellistor is a gas sensor that detects combustible gases and

vapours in air”. The word “pellistor” is a combination of pellet and resistor. While the types of coating sensor materials are classified according to the additive doping materials, the type and nature of chemical interactions, the reversibility of chemical reactions and operating temperature, most of the measurement methods are based on transduction. However there are other sensors that measure mass or temperature changes and heat generation (14), the electrochemical gas sensor is constructed with a material that joins the space between two electrodes or covers a system of interdigitated electrodes. It should be noted that certain disturbances in the environment can cause changes in the electrical resistance of the electrochemical gas sensor, which can alter the measurement. The resistance depends on the amount of gas and changes proportionally.

One of the most common gas sensor used are the metal-oxide gas ones. The advantages are the high sensitivity and reaction to oxidizing compounds and some reducing compounds. The basis of the device is a ceramic support tube with a heater spiral, usually made of platinum. Tin-dioxide (SnO2) is the most often used covering material. To improve its functions, some quantities of catalytic metallic materials are added. The combustion reaction with oxygen on the surface of tin dioxide is caused by the sorption of volatile molecules which modifies the conductivity and its measurement. “Sorption is a physical and chemical process by which one substance adheres to another”. The reaction is shown in Fig.10.

Figure 10 : Scheme presentation of the function of a MOX sensor (Pérez R, Marimón X, Portela Artificial nose. UIC

Barcelona).

Figure 11 : Metal-oxide (MOX) gas sensor elements and arrays for the detection of selected Volatile Organic Compounds

(VOCs). UST Umweltsensortechnik GmbH. (Germany).

8


UST Umweltsensortechnik GmbH is a successful medium sized enterprise, recognised internationally as a leading company for development and production of ceramic sensor technology, successfully acting in the global market (Fig.11). There MOX gas sensor elements are realized in hybrid-technology: ceramic carrier substrate (Al2O3) with a micro-structured platinum thin-film layer, covered with a passivation layer and specific layers for contacts, and gas-sensitive metal-oxide (MOX) layer/s…. The functionality of a MOX gas sensor is based on the conductivity-change of the gas-sensitive MOX-semi-conductor layer/s at gas exposure.

The measurements of conductance changes over a period of time according to the odour signal are shown in Fig. 12. Unfortunately, the reaction requires high temperatures ranging from 300°C to 550°C, which leads to high electricity consumption. At lower temperatures, the reaction is very slow. And at temperatures below 100°C, it is impossible, because the low vapour pressure of water makes oxidative chemical reactions impossible (25). On the other hand, oxide sensors have a lower selectivity and a higher risk of poisoning by some weak acids (26).

Figure 12 : Conductance changes in a period of time according to the type of odour signal (Pérez R, Marimón X, Portela A.

Artificial nose. UIC- Barcelona)

Hu et al. compared different properties of electrochemical gas sensors (7). He described that large surface areas of nanomaterials have enormous sensitivity. Monodispersed SnO2 quantum dots are very sensitive but they work at high temperatures around 350°C. The heterojunction between well-designed dual-phase materials can increase the effectiveness of each material by relying on their synergistic work, for example, the heterojunction between ZnO and SnO2 with a band banding and Te nanowires@SnO2 nanowires hierarchical nanostructure.Hierarchical nanostructures with SnO2 skeletons and ZnO branches are successfully prepared on a large scale by combining the process of transport and vapour deposition (for SnO2 nanowires) and a hydrothermal growth (for ZnO). There is a possibility to increase the effectiveness of metal-oxide gas sensors by inducing the phase change of which the resistances vary. To eliminate the humidity effect on semi-conductive polymer and metal oxide it is possible to cover the surface of sensors with molecular or metal-organic frameworks but

also to use the modified water-detach and float-attach (Water-detach and float-attach, WDFA) method, especially at room temperature.

The bimetallic ZIF-CoZn MOF sheath on ZnO is characterized by a good thermal stability, working temperature, good anti-humidity, ability to recover, and has an enormous thermally catalytic ability on ZnO.

The function of the metal-oxide-semiconductor field-effect transistors (Metal-oxide-semiconductor field-effect transistors, MOSFET) (Fig. 13) is based on the ability of some metals to absorb and dissolve hydrogen (22).

Figure 13 : Metal-Oxide sensors (MOx) (X.Marimon, UIC 2020).

MOSFET is made of three layers. The bottom layer is formed by a single crystal of silicon or germanium-doped silicon. In semiconductor production, doping is the intentional introduction of impurities into an intrinsic semiconductor in order to modulate its electrical, optical and structural properties. The doped material is called an extrinsic semiconductor. This plate is sputtered with a very thin wad of silica or other metal or semi-metal oxide that may be attached. This layer must be continuous, as well as thin as possible. Currently, in the most technologically advanced processors, this layer is as thick as the oxide particles. A very thin steel line is then sprayed from the paper support onto the pressure stress. It appears that the MOSFET sensors are very robust (26), (27), (28).

In conductive or conductive polymer gas sensors, adsorption of gases on the sensor surface causes changes in electrical resistance, which can be measured and provide information about odours. Conducting polymers have many significant advantages. The most important one is the possibility of modifying and giving those specific and desired properties, biocompatibility, high conductivity and flexibility, low specific gravity, high mechanical strength, durability, easy processing, and a low production cost. This possibility is used to improve the long term stability of the sensor as well as high-temperature resistance. Meanwhile, they can operate at room temperature. Conductive polymer gas sensors are built of a substrate, usually silicon, a pair of gold-plated electrodes, and a conducting organic polymer coating as the sensing element (25).

Insertion of carbon nanotubes into the polymer layer improves its electrical and mechanical properties and can improve the response speed and the detection efficiency

9


of low gas concentrations (28). Unfortunately, one of the main disadvantages of conductive polymers is their high susceptibility to environmental humidity. Another type of sensor is the acoustic wave gas sensors. They measure a mechanical (acoustic) wave as it propagates through (bulk acoustic wave, BAW) or over (surface acoustic wave, SAW) the surface of the material covering the sensor. The specific properties of the odours influence their propagation and wave amplitude. The main elements are a piezoelectric substrate and a suitable absorbing material. The waves are typically between 1 and 500 MHz (29).

Gas sensors based on Quartz Crystal Microbalance (QCM) are characterized by changes in a voltage proportional to the mechanical stress which was applied. This phenomenon was discovered by Jacques and Pierre Curie (30). Sauerbrey

described that when voltage is applied to a quartz crystal, it causes oscillation at a specific frequency, the mass on the quartz surface changes proportionally to the voltage (31). The change in QCM frequency determines the mass of analyte adsorbed in ng/cm2. “Analyte is a component of analytical interest in a sample that is separated from the matrix. It is a chemical species whose presence or content is to be known, identifiable and quantifiable, by a chemical measurement process”. It can be applied for elastic subjects, which do not dissipate any energy during oscillation, such as metallic coatings, metal oxides, and thin adsorbed layers. The phenomenon was applied to the artificial nose (32). E-nose based on QCM is able to vary gas samples through the change of the resonant frequency caused by gas absorption/desorption, which provokes changes in the mass of specific materials. Because of low selectivity, QCM is rarely used as a single sensor, it is rather applied as the supportive element of sensor arrays (9) (Table I)

10


The other type of olfactory sensors that demand attention is the optical sensors. They measure the light properties and features of light, such as colour or wavelength, absorbance, fluorescence, polarization, optical layer thickness (Table II).

A few theories were established to overcome the limitations of e-nose. Some manufacturers suggested the application of the e-nose in controlled external conditions for a thorough and accurate evaluation of the collected data (35). Others advised using hybrid systems that contain different types of sensors with the most appropriate features (36). Another group of researchers integrated the e-nose to gas chromatography to achieve better results (37). Son et al. (38) designed a bio-electronic nose for a quicker analysis of examined components. The problem with the size of the device was treated by Wei Xun et al. (39). They proposed a self-developed portable electronic nose equipment (e-nose) to acquire volatile substances non-destructively and achieved a predictive correct response rate of 95.83%.

Qui et al. proposed a promising idea to combine e-nose with e-tongue (36). The fusion of both devices would improve odour detection and classification. E-tongue is an equivalent of e-nose; it includes an array of non-selective chemical sensors with partial specificity for different solution components and an appropriate pattern recognition instrument, capable to recognize quantitative and qualitative compositions of simple and complex solutions.

Aouadi, in his overview, compared different features of various correlative analytical methods used for odour detection, such as sensory analysis, mass spectrometry, chromatography, polymerase chain reaction (PCR), enzyme-linked-immuno-sorbent-assay (ELISA), Dumas, Soxhlet, near-infrared (NIR) spectroscopy, electronic nose (E-nose), electronic tongue (E-tongue) (41).

E-nose is characterized by its good affordability, low detection limit, portability, specificity, fast measurement time, economical maintenance, ability to perform qualitative and quantitative analysis, simple use, no need for sample or reagent preparation. On the downside, its selectivity is poor compared to conventional methods.

E-Noses Available In The MarketThe objective pursued for all the researchers, as well as us, is to manufacture an almost perfect e-nose equipment that can be used for commercial use, able to operate in different uncontrolled, changing environmental conditions and even outdoors.

That is why the sensor array must have a high sensitivity to the specified odours, at the same time it must be resistant to certain varying external parameters, especially air temperature and humidity. This condition is particularly important in the case of agriculture, where external conditions have a great impact on crops. Desired characteristics are the ability to operate at relatively low temperatures, to keep operating costs low, short calibration and training requirements, fast recovery time between duty cycles, relatively short analysis times, and especially high stability and reliability of the sensor array (24). Existing equipment is often too large, too heavy, and not portable. If we were to achieve miniaturization of the devices, artificial noses could be conveniently used in all activities of daily life. However, it must contain integrated recording and analysis components.

Currently on the market, there are a couple of electronic noses already available for commercial use, although the real effectiveness of each of them is not effective for use at the moment (Table III).1. The Aeonose (TheeNoseCompany) for medical use is

used especially in the detection of various diseases using the analysis of exhaled lung air, e.g. colon cancer, lung cancer and tuberculosis.

2. FOODsniffer (FOODsniffer) used to measure the quality of food especially the freshness of raw meat, poultry, and fish; in addition, it can be connected to any smartphone using its application.

3. The HeraclesNeo ( AlphaMOS ) is used to monitor the condition of food and beverages and is a combination of dual electronic nose and fast gas chromatography.

4. The Odotech proposes several types of tools to be used (e-Nose, SulfNose, MultiNose), which are intended for environmental pollution control.

• The e-nose continuously monitors fine fluctuations in odour source concentrations in real time.

• SulfNose is dedicated to measuring the real-time concentration of H2S1 in ambient air. ”Hydrogen sulfide (H2S1) is a gas commonly encountered during the drilling and production of crude oil and natural gas, as well as in wastewater treatment, utility and sewage facilities. The gas is produced as a result of microbial decomposition of organic materials in the absence of oxygen and therefore the smell of rotten eggs”.

11


• MultiNose can measure four different types of gases from a single source, such as VOC, ammonium, methane, carbon dioxide, carbon monoxide, nitrogen dioxide.

5. The AirSense analytics Portable Electronic Nose (PEN) is a small, fast and robust identification system for gases and gas mixtures in environmental applications. The instrument is based on the patented A3 technology, contains ten sensors and can identify ten different compounds or provide a simple answer such as “Good-Bad”, “Yes-No”.

The Company Arybelle propsed certain applications such as Arybell EN and NeOse Pro. “Based on Aryballe’s unique digital olfaction platform, NeOse Pro detects and identifies odours with O-Cell technology to mimic the human sense of smell. Digitized odours are stored in a reference database, allowing new odours to be matched later. This compact device brings the value of objective odours data to the food and beverage and personal care industries, giving them access for the first time to the olfactory fingerprints of their products. NeOse Pro is a universal olfactory sensor capable of detecting, recording and recognizing odours, typically used in R&D, laboratory, quality control and quality assurance applications”. (Table IV).

This device works in this way: first, a digital olfaction occurs with silicon photonics for high volume application, and then odour signals are captured with an array of silicon photonic integrated interferometers. One path is functionalized with Aryballe’s proprietary combination of biosensors, following, odorant VOCs bind to biosensors with varying levels of affinity. Light recombines before the optical sensor, then, optical sensor captures the change in intensity based on the biosensor response. A full array of interferometers captures the odour’s unique pattern of interaction with each biosensor, and a pattern is sent to the server where it is analysed against existing odour database to identify a match. Identified odours are displayed on Arybelle software.

NeOse Pro is a portable device, which can be applied in the food and beverage and personal care industries. Thanks to the connection with the digitalized reference odour’s database, it allows subsequent matching of new odours and answers in real-time https://aryballe.com/solutions/device-solutions/

Cyranose (SenSigents), which contains 32 sensors and is based on Principle component analysis (PCA), K-nearest neighbours (KNNs) and Support vector machines (SVMs), can detect and analyse a wide range of gases. It is still used in several indoor-outdoor monitoring and medical purpose applications(http://www.sensigent.com/products/cyranose.html).

12


E-NOSE Applications Medical-health uses.The possibility of having a device that can diagnose diseases quickly, correctly and non-invasively, would represent a major advance in medicine. The e-nose system offers great hope for the detection and monitoring of certain diseases through exhalation of breath, but also in skin perspiration/sweat, feces, urine, saliva, breast milk and intestinal gases, where volatile organic compounds (VOCs) can be found (42). It is known that some substances present in the blood can penetrate into the alveoli and be exhaled along with the expired air. Moreover, VOCs can arise either as endogenous or exogenous substances present in the blood. They can give information on general metabolic processes, especially on what happens in the lungs. The presence of organic compounds in the exhaled breath does not always indicate the existence of a disease or tumour lesions, but they can provide clues about an organic or tumour process.

The first analysis of the composition of human breath was published by Pauling in 1971 and was performed by using a combination of gas chromatography/mass spectroscopy (GC/MS) (43). This author tried to relate some volatile compounds to certain diseases. Later, Kaji discovered the presence of mercaptans and aliphatics in the breath of patients suffering from liver cirrhosis (44). “A compound with an unpleasant odour, derived from an alcohol in which oxygen is replaced by sulphur. In organic chemistry, a thiol is a compound containing the functional group consisting of a sulfur atom and a hydrogen atom (-SH). Being the sulphur analogue of a hydroxyl group (-OH), this functional group is called a thiol group or sulfhydryl group. Traditionally, thiols are called mercaptans. The adjective aliphatic is used in chemistry to describe those organic compounds that have an open chain as the structure of their molecules. Thus, it is possible to speak of different compounds of the aliphatic type”.

While Simenhoff found dimethyl and trimethylamine in the breath of uremic patients (45). O’Neil in 1998 defined some alkanes (hexane and methylpentane among others) and benzene derivatives such as o-toluidine and aniline as indicators of lung cancer (46). His study was confirmed by Philips, who explored the breath composition of individuals previously diagnosed

with lung cancer (47). All these investigations were carried out using GC/MS. “Methylpentane, known trivially as isohexane, is a branched-chain alkane with the molecular formula C6H14. It is a structural isomer of hexane composed of a methyl group attached to the second carbon atom on a pentane chain”.

The most recent achievements of detection of various VOCs in medicine were published in 2020 by Wilson (48). Recently, the attention of researchers was focused on gastrointestinal and infected wound diseases. Tiele and collaborators performed a screening in patients suffering from inflammatory bowel disease (49), analysing breath samples from a handmade electronic nose (Wolf electronic nose) and with a commercial gas chromatograph and ion mobility spectrometer (GAS BreathSpec GC-IMS). He demonstrated by using GC-IMS that there is a reduced amount of short-chain fatty acids, such as butanoic acid and acetic acid, in breath samples taken from patients with inflammatory bowel disease compared to healthy controls. Using the Wolf eNose, evident differences in the detection of ammonia, sulfur dioxide and nitrogen dioxide. Deianova and cols. analysed with a C-320 electronic nasal device (Sensigent ISS, Baldwin Park, CA, USA) postnatal fecal samples to find differences between preterm, vaginally and cesarean delivered infants (50). No significant variations were detected among preterm births in the groups examined. Only weak potential effects of gestational age on fecal VOCs were found (Table V).

The application of e-nose in medicine is becoming increasingly wide and interesting. This has led to an increasing search for the use of e-nose in various branches of medicine. E-nose can identify bacteria by the analysis of a patient’s breath, especially in the diagnosis of infections, such as pneumonia (51), (52), (53). Referring to Lai, it can distinguish the VOCs patterns of some common respiratory pathogens, such as H. influenza, S. pneumoniae and P. aeruginosa, from control probes (54). Thaler proved that the e-nose can identify bacterial sinusitis in 72% of the patients (56). Moreover, it can classify bacteria in infected diabetic foot ulcers (56). Holmberg was interested in researching urinary tract infection by different types of e-noses. He used an artificial neural network for classification of different species of bacteria present in human urinary tract (57). Pavlou used an electronic nose and a hybrid intelligent odour recognition system in discrimination between Helicobacter pylori and other gastroesophageal isolates (58).

13


Ping, in his analysis of diabetes breath in patients indicated that an electronic nose may be a useful device for diagnosing diabetes (59). Di Natale developed the application of the electronic nose in the rapid detection of blood cells in urine (60) and compounds secreted by the human body through perspiration (57). Another study was carried out in patients suffering from breast or colorectal cancer (62), (63). For breast cancer, the artificial nose called Blue Box has been described, which has sensors that analyses urine samples, although in a preliminary stage, it is expected to be available in 2023.

D’Amico described the use of an electronic nose based on quartz microbalance (QMB) sensors which were coated with different kinds of metalloporphyrins in detecting lung cancer. Metalloporphyrins are one of the best compounds to detect various VOC ligands (63). Ballantine in 1989 explained the action of QMB (64). Adsorption of molecules from the gas phase causes a variation of the oscillating mass of a thin quartz crystal which induces a variation of the oscillation frequency. According to the work of Brunik, the selectivity and sensitivity of sensors based on metalloporphyrins depend on both the

central metal and peripheral substituents of the complex (65). It has been proved that aromatic compounds, such as amines, alcohols, ketones, alkanes, or benzene, may indicate lung cancer and are well detected based on metalloporphyrin receptors. Therefore, metalloorganic compounds are good candidates for VOC sensing. Metalloporphyrins compounds offer wide and different possibilities to change their structure, which depends on the nature of the central metal and peripheral substituents of the porphyrin complex (65). Lung cancer was also a subject of interest of many another authors. For example, Bikov and Dragonieri conducted research on e-noses based on polymer conductive composite sensors (66), (67). Another study with the use of nosensors based on organically functionalised gold nanoparticles and gas chromatography linked to the mass spectrometry technique was carried out on patients suffering from lung, breast, colorectal, prostate cancers (68).

Hu highlights the good ability of sensor arrays made of WO3-type nanofibers (NFs) functionalized with Pt, Pd, Rh and Pt-Y to discriminate between simulated breathing of a halitosis patient and healthy people (11).

14


Food and Beverages The olfactory sense is one of the first senses which react to various types of food, beverages, odours and tastes, etc. It plays an important role in food acceptance by the human being. That is why it has become a key element in the development of a huge potential commercial industry.

The most important features of the artificial nose in the food industry are quality assessment, shelf life and spoilage monitoring of food, detection of hazardous chemicals or existing bacteria, but also the classification of food and beverages. One of the most resistant sensors to high amplitudes of temperature and humidity are the metal oxide sensors that were used in multiple studies presented below.

MeatAs we have mentioned, the study and analysis of food represents a very extensive and useful field. Musatov proved that KAMINA e-nose can determine meat freshness. KAMINA e-nose consists of a metal oxide sensor microarray (MOS) and a linear discriminant analysis (LDA) (69). In Zhang’s study, the e-nose based on six metal oxide sensors determined beef freshness during 6 days of storage at a temperature of 20OC and relative humidity of 60% (70). Haddi classified sheep meat and meat samples stored at 4º C for 15 days by e-nose based on 8 metal oxide sensors with 94.64% accuracy (71) (Table VI).

According to Tian, the e-nose offers the possibility of detecting fraud in the food industry. He performed an e-nose study of ten metal oxide sensors to test samples of minced lamb mixed with 0%, 20%, 40%, 60%, 80% and 100% pork (68). Pattarapon, according to his study based on an e-nose of 14 gas sensors, the recommended storage time at 4ºC for grass carp fillet (herbivorous) to maintain freshness under vacuum at 30 kPa was 10 days but at 50 kPa it can be extended to 12 days (73). Güney, thanks to the e-nose, different fish species (horse mackerel, anchovy and whiting) were classified with 96.18% accuracy (74). Thanks to an E-nose with optical sensors, the researchers were able to detect mesophilic bacteria (Bacteria that decompose organic matter at temperatures ranging from 30 to 40 C) in tilapia fish species, which is an important factor in monitoring fish deterioration.” Tilapia is the generic name for a group of fish of African origin, consisting of several species, some of which are of economic interest, belonging to the genus Oreochromis”. The species of commercial interest are farmed in professional fish farms in various parts of the world (75).

15


Milk and dairy productsE-nose can be used in the dairy industry for the discrimination of Lactobacillus casei strains between milk produced by healthy cows and cows infected with mastitis (76), also in monitoring the ripening of aged cheese (77) and to know the status of cheese samples, their freshness, age, classification, according to Crescenza (78). Today, many producers maintain the quality of raw milk by using antimicrobial agents to reduce the microbial population, some of them may add different types of agents such as detergents. Tohidi studied the composition of raw milk using an e-nose based on eight metal oxide semiconductor sensors to detect the presence of detergent powder. The combination of (Multivariate analysis of variance, MANOVA) and differential baseline correction method as employed in the study. They distinguished different levels of milk adulteration (79) (Table VII).

HoneyThe researchers also determined the origin of the honey using an electronic nose equipped with a mass spectroscopy detector using the Solid Phase Micro-extraction (SPME) sampling technique. They classified the honey according to its botanical origin: acacia, dandelion, chestnut, rapeseed, linden, and spruce (80). Another study was carried out with an e-nose consisting of six

16


semiconductor sensors to classify the different botanical origins of honey. However, in this case, certain conditions had to be met, such as a volumetric flow rate of 15 L/h, a barbotage temperature of 35°C and an acquisition time of 60 seconds (81).

It is also possible to differentiate the botanical and geographical origin of honey. Huang conducted his research with an electronic nose comprising metal oxide (SnO2) sensors. He defined physicochemical parameters such as fructose, glucose, hydroxymethylfurfural (HMF), amylase activity and acidity (83). Faal, using e-nose based on MOS sensors, predicted the moisture, ash content and free acidity of the tested honey samples (79). He also tested an electronic nose with an artificial neural network (ANN) and support vector machine models for the prediction of physicochemical parameters of honey including ash content, free acidity, moisture content, and pH as a function of its aroma (Table VIII).

Coffee More than 800 different compounds have been identified in coffee, each of which plays a role in determining its aroma (84). Giungato examined the degree of roasting of Brazilian, Indian, Vietnamese and Costa Rican coffee beans (85). A similar study was conducted by Marek (86) in which he analysed samples of Brazilian, Guatemalan, Ethiopian and Costa Rican coffees using “principal component analysis” (PCA) plot varieties. “Principal Component Analysis, or PCA, is a dimensionality reduction method often used to reduce the size of large data sets by transforming a large set of variables into a smaller one that still contains most of the information in the large set”. Dong attempted to differentiate Robusta coffee samples by e-nose analysis based on six coffee samples with metal oxide gas. When a complementary test with an e-tongue is added, the final accuracy is improved. When a complementary e-tongue test is added, the final accuracy is improved (87). In another study, coffee samples treated with hot air and heat pump are discriminated in relation to other samples dried under different conditions (88). Brudzewski used an e-nose based on a MOS (Metal Oxide Semiconductor) gas sensor array to detect common fraudulent practices in coffee production. This way he was able to detect that Arabica coffee had been replaced by Robusta coffee, which is cheaper. He also described a detection and classification system for the eleven classes of Arabica blends and Robusta coffees (89) (Figs.14,15). “Arabica coffee has a more delicate and sweet taste, is lighter (contains less caffeine) and is more aromatic. Robusta coffee, on the other hand, has a more intense, stronger flavour (it contains more caffeine) and is less aromatic, but it has more body and allows for the creation of a creamier and frothier espresso”. Pardo performed the classification of the seven different types of coffee using Pico-1 e-nose consisting of thin-film semiconductor sensors. He performed studies using principal component analysis and multilayer perceptrons and compared the results with ratings from trained assessors who performed quantitative measurements and an overall index (called the Hedonic Index, HI). “The multilayer perceptron can be fully or locally connected. In the first case each output of a neuron of layer “i” is input of all neurons of layer “i+1”, while in the second case each neuron of layer “i” is input of a number of neurons (region) of layer “i+1”. The multilayer perceptron is an artificial neural network (ANN) consisting of multiple layers, such that it has the ability to solve problems that are not linearly separable, which is the main limitation of the perceptron (also called simple perceptron). The results were comparable (90) (Table IX)

17


Figure 14: Artificial Nose in the coffee industry.

Figure15: Coffee varieties in radar plots.

https://www.semanticscholar.org/paper/Classification-of-instant-coffee-odors-by-nose-of-ThepudomSricharoenchai/047cebb496d0a195f8d02b8c706c15253113e673https://www.iconfinder.com/icons/2670620/app_coffee_mobile_phone_icon

18


Tea “Tea” is a beverage with a characteristic and unique flavour and aroma. Its manufacturing process involves different stages or processes. These include plucking, withering, rolling, fermentation and finally cooking. The quality of tea is measured according to parameters such as colour, flavour and aroma, characteristics that are acquired during the fermentation process in which polyphenolic compounds are oxidized when exposed to air. “Polyphenolic compounds (PPCs) are secondary plant metabolites that possess at least one aromatic ring in their structure to which one or more hydroxyl groups are attached. CPFs are classified as phenolic acids (PA), flavonoids (FLA) and tannins (TAN)”.

At this stage is when you can better assess the quality of the tea and get the highest quality. The manufacturers know that during the fermentation process there are two different odour peaks, which are difficult for them to assess, so by introducing an e-nose they can detect the optimum level of fermentation by detecting the variation in the aroma level.

Dutta uses the e-nose with four metal oxide sensors to discriminate five tea samples from different conditions of the drying, fermentation, baking and mashing process. According to this study, e-nose can evaluate the five different tea samples with 100% constitutional variants (92). In Sharmilan’s study, an e-nose consisting of an array of semiconductor metal oxide gas sensors is used using different statistical and neural network techniques (SVD, 2-NM, MDM, PCA, SVM, RBF, SOM, PNN, and Elman’s recurrent type) to find the optimum fermentation level of the tea by detecting the variation in aroma level (93) (Table X).

19


Alcoholic beverages In his study, Ragazzo-Sanchez examined with an electronic nose certain alcoholic beverages (beer and wine) contaminated with off-flavours after the dehydration and dealcoholisation procedure. E-nose was able to detect off-flavours in both beer and wine (94). Aleixandre made a comparative study between WiNOSE 2.0 and the evaluation done by experts on the quantification in wines made from binary blends of four white grape varieties and two red wine varieties in different percentages. The process was carried out at temperatures between 200 and 350°C with a matrix composed of four thin layers of nano-crystalline tin oxide. The author demonstrated that an e-nose is faster, simpler and more objective than the effect obtained with human tasters (95). Rodríguez-Méndez reviewed the existing e-nose and e-tongues on the market adapted to the wine industry, concluding that electronic noses have greatly contributed to control the quality of the grapes and the pressing itself (96), (97), the quality and organoleptic characteristics of wines at different stages of production (98), and can even control the fermentation and aging of wines in oak barrels (99), its condition at bottling (100), detection of spoilage, bad odours (101), fraud and adulteration (102) and even to measure chemical properties (103). According to their article, the resistive sensors consisting of doped metal oxides (MOX) and MOSFET are the most common used arrays in wine analysis. A resistive sensor is an electromechanical transducer or device that converts a mechanical change, such as displacement, into an electrical signal that can be monitored after conditioning. They are commonly used in instrumentation. The simplest example of this is a potentiometer. Metals are doped with metal oxide primarily to improve their conductivity for a particular level of gas sample therein increasing the sensitivity of the metal oxide gas sensor. More about doping also increases the selectivity of metal oxide gas sensors towards a particular gas.

The limitation of the sensors is their high sensitivity to water and ethanol, which decreases their ability to detect odours. That is why sampling techniques were developed; one of the most popular in the wine industry is Solid Phase Micro Extraction (SPME). Solid-phase micro-extraction or SPME is a technique used in analytical chemistry to extract chemical compounds for subsequent identification. It was developed in the early 1990s by Dr. Pawliszyn’s team at the University of Waterloo. This technique is widely used in any field of analytical chemistry in the preparative phase. Principal component analysis (PCA) is a widely used pre-processing technique to vary between samples with different organoleptic characteristics. For prediction tasks (e.g. classification or regression) a low-dimensional feature vector is implemented. Classification of unknown samples can be solved by linear discriminant analysis (LDA), Soft Independent Modeling of Class Analogy (SIMCA), Support Vector Machines (SVM) or Artificial Neural Networks (ANN). Partial least square (PLS) is generally used in regression tasks (104). (Fig. 16) (Table XI).

Figure 16: Application in the wine industry. (105)

20


Olive oilCosio developed a classification model for extra virgin olive oils from different regions to distinguish between oils of different geographical origins. He based his study with an e-nose on ten MOSFET sensors and counter-propagation artificial neural networks (CP-ANN). All the models were validated. All models were validated with commercial olive oil samples (106). The combination of an e-nose (10 MOSFET and 12 MOS sensors) and an e-tongue (flow injection analysis with two amperometric detector sensors) is effective in determining the oxidation of extra virgin olive oils under different real storage conditions (107). In his study, Hai attempted to detect adulteration of corn oil with sesame oil using an E-PEN2 terminal (10MOS gas sensors) to assess the quality of corn oil (108). In another study, the same authors examined counterfeiting with camellia seed oil and sesame oil using an electronic nose and Principal component analysis (PCA), Linear discriminant analysis (LDA). Promising results were obtained, with accuracy exceeding 90% for both types of oil (109). Lipid oxidation continuously reduces the nutritional value of fats. Triyana proposed using a portable e-nose that uses a low-cost dynamic headspace and a commercial metal oxide gas sensor array to sort vegetable oils (sunflower, grapeseed) and animal fats (chicken, lamb, pork) (110) (Table XII).

21


Fruits and vegetablesCui diagnosed greenhouse tomato plants infested by aphids in the early stages. New volatile organic compounds (linalool, carveol, and nonane (2,2,4,4,6,8,8-heptamethyl)) indicated aphid invasion on tomato plants (111). Quartz microbalance and mass spectrometry - based electronic noses are used to monitor the aroma profiles of two tomato varieties (Tradiro and Clotilde) during shelf-life conditions. Mass spectrometry-based e-nose showed higher sensitivity, clear distinction between cultivars, and indicated an evident alteration in aroma profiles (112). “A cultivar is a group of plants artificially selected by various methods from a more variable crop, with the purpose of fixing in them characters of importance to the breeder that are maintained after reproduction”. Chen studied the freshness of fresh-cut green bell peppers during 9 days of storage. He used a commercial e-nose instrument consisting of 14 MOS sensors. The best results were obtained by distinguishing between fresh and spoiled vegetables (113). In another study, it was possible to establish differences between fresh broccoli, half-fresh broccoli and other fresh broccoli samples that had deteriorated during storage. The result of this assessment showed that an e-nose of 14 MOS gas sensors is effective if the complex gas chromatography-mass spectrometry method is used (114). In another study, Wang demonstrated that the use of a portable commercial PEN3 E-nose to recognize Golden Delicious apples with Penicillium expansum and Aspergillus niger molds was very useful, with accuracy ranging from 72% to 96.3% (115). Ezhilan conducted a multidimensional approach to assess the freshness of broccoli, using four methods including e-nose, bacterial culture test, gas chromatography-mass spectrometry head sampling method and Fourier transform infrared spectroscopy to detect different species of bacteria. E-nose revealed the presence of Staphylococcus, Salmonella and Shigella in the spoiled samples (116) (Table XIII).

22


Cacao / chocolate Tan tested the fermentation time of cocoa beans (Theobroma cacao, also called cacao tree is a small evergreen tree of the Malvaceae family, native to the deep tropical regions of Mesoamerica. Its seeds, the cocoa beans, are used to make chocolate liquor, cocoa solids, cocoa butter and chocolate) using an in-house designed artificial nose based on nine gas sensors. The artificial neural network (ANN) achieved a misclassification rate of 14.2 % in a short period (7 days). Fermenting cocoa beans generates high water vapour content, which can deteriorate the action of the sensors (117). The same author applied an e-nose based on six machine learning methods (bootstrap, boosted tree, decision tree, artificial neural network (ANN), naïve Bayes, and k-nearest neighbours) to specify the degree of fermentation of cocoa beans. The results show that the bootstrap forest algorithm achieved a misclassification rate of 9.4%, ANN - 12.8%, augmented - 13.6%, other methods did not classify cocoa beans (118).

Valdez presented an evaluation of 26 chocolate bar samples made with an electronic nose based on a metal oxide gas sensor beam with stimulus generated by controlled pressure. This same system was able to differentiate between samples in relation to four characteristics such as, type of chocolate, extra ingredient, sweetener and expiration date. The average classification rate was 81.3% with 0.99% accuracy, 0.86% precision, 0.84% sensitivity and 0.99% specificity, and the average feature recognition rate was 85.36% with 0.96% accuracy, 0.86% precision, 0.85% sensitivity and 0.96% specificity (119). Thanks to a neuro-diffuser-based e-nose, “Neuro-fuzzy systems (NFS) are part of the concept of soft computing. They are a synergistic fusion of fuzzy logic and neural networks with the ability to automate adaptation to training data and knowledge interpretability”, Rahman was able to classify cocoa quality with an accuracy rate of 95.21% (120). Rottiers conducted a study using an e-nose based on ultra-fast gas chromatography (GC) and principal component analysis to rapidly discriminate between fourteen cocoa liquors of various geographical origins and determine their quality. In addition, discriminant factor analysis distinguished between fine and bulk cocoa. “The manufacturers separate cocoa beans into fines and bulk; fines characterize their peculiar notes and are used to produce fine chocolates (Ascrizzi et al., 2017), while bulk beans are used to produce low-quality chocolates and products such as cocoa powder (Vargas Jentzsch et al).(121). Olunloyo also attempted to determine the quality of cocoa beans by using a metal oxide semiconductor sensor array e-nose and an artificial neural network pattern recognition unit, thereby achieving the accuracy rate of 95% (122) (Table XIV).

23


PERFUMESBranca worked on an e-nose applicable to the perfume industry. This author detected the presence of a perfumery note, named “mangone”, in a fragrance at low concentrations. “Notes in perfumery are descriptors of aromas that can be felt with the application of a perfume. Notes are divided into three classes: top / main notes, middle / heart notes and base notes; which denote groups of aromas that can be felt with respect to the time after the application of a perfume”. The results were compared with sensory analysis performed by trained expert analysers (nose men) and by the measurements made by gas chromatography mass spectroscopy (GC-MS). The presence of scent for the concentration 10-2% (w/w) could be identified only by 40% of the trained experts and, at the same time, the identification by GC-MS was very difficult, due to the low concentration of volatile substances examined and the similar retention times obtained for other compounds present in the fragrance. The developed e-nose provided fingerprints for different odours, associated with different samples that were used to generate an odour (123).

In the e-nose, two multivariate data analysis were conducted: unsupervised principal component analysis (PCA) and an artificial neural network (ANN). The e-nose achieved 100% mangone discrimination using a radial basis function (RBF) to an artificial neural network at a concentration of 10-4 % (w/w) (123). Eamsa-ard did a study with gas sensors in an e-nose to evaluate fragrance (specifically Linalool), and human body odour with sweat along with artificial intelligence for body odour tracking. The systems were made of nano-materials. The PVP-NH2 sensor had the highest response to fragrance. The research showed that the e-nose can detect fragrance left on human skin or clothing, which can be adapted to personalized fragrance determination. The conclusions were that an e-nose can be a potentially useful tool to achieve fragrances that are most suitable for each individual (124) (Table XV).

24


Artificial Intelligence Contribution in an Artificial Nose. An important step is to develop an appropriate classification algorithm. The signals generated by the e-nose (collected by the sensor beam) are processed and analysed to provide the desired and satisfactory information. Before the data is analysed, data pre-processing is usually performed to preprocess data before training model, because the quality of the data and their representation directly affects the model ability to learn. In case of artificial nose the most important data preprocessing step is data form transformation e.g. dimensionality reduction or normalization.

Artificial neural networks (ANNs) are currently recognized as the best methods for analysing artificial sensory data, generally because the ANNs can learn high-level features in the data, what allows them to classify data from artificial nose with quite high accuracy. The simplest artificial neural network, containing one neuron, is called a perceptron. The main and most important element of the perceptron is the McCulloch-Pitts neuron, which is a simplified model of a biological nerve cell. The McCulloch-Pitts neuron contains a few or even several dozen entries (inputs) and only one exit (output).

Each input is assigned an appropriate real number (weight). The ANN operation starts with the multiplication of all input values with appropriate weights. In the next step, all obtained products are sent to the summation section. The calculated sum is transmitted to the next section of the neuron: the activation block. At this stage, the weighted sum of the input vectors becomes an argument of the activation function.

Activation is used to determine the value of the output neuron. They are described for many types of activation functions: linear, unipolar threshold, unipolar sigmoidal, bipolar, ReLU, Gaussian and many others. The use of ANN to solve a given problem is possible after collecting a set of training data: examples of input values together with appropriate output parameters defined. The learning of a neural network consists of changing its internal parameters (weights and biases). This is achieved by an appropriate algorithm, usually in the form of supervised teaching. For this purpose, the backward error propagation algorithm is most often used. It updates weights and biases values based on the training data to minimize the loss function that is associated with the errors generated by the network during the training process.

Complex problems are associated with multi-layered artificial neural networks, but require more training data. Their characteristic feature is the presence of several hidden layers of neurons. These layers play an intermediate role in signal transfer between an input node and an output layer. They are constructed in such a way that the output of the neuron in the previous layer meets the input of all neurons in the next layer. The number of neurons in the layers is one of many extremely important parameters for ANN operation. One of the most important advantages of ANN is its ability to generalize information obtained from previously unknown data. In addition, neural networks are characterized by a high tolerance to additional perturbations, lack of continuity and deficiencies in the training set (125). Fig. 17 schematically shows the action of artificial neural networks, which analyse signals, eliminate noise and group information into appropriate clusters.

Figure 17: The graphic that shows the artificial analysis function.

Zahn (126), in his study, introduced a new application to predict the diagnosis of lung cancer with an e-nose. Samples were obtained from 31 patients aged 30 to 80 years with diagnosed lung cancer (adenocarcinoma, small cell carcinoma, squamous cell carcinoma, large cell carcinoma). These patients had not received chemotherapy or radiotherapy during the 6 months prior to the tests and had abstained from smoking for at least 6 months. The control group consisted of 41 healthy volunteers with the same chronological age range. All deep breath samples were collected under the instructions of the investigators within 3-5 minutes. The 72 samples were recollected in Polyethylene Terephthalate aluminized substrate using M3014-4 off-line air collection equipment (ECO MEDICS AG, Duernten, Switzerland) and air filters were used to reduce the environmental effect. “Polyethylene terephthalate is a type of plastic widely used in beverage containers and textiles”.

25


They were stored in a thermostat at 37º C and analysed with an electronic nose made at the State Key Laboratory of Industrial Control Technology in Zhejiang University consisting of 16 metal-oxide semiconductor sensors per hour. All 16 sensors had overlapping specificity. Typical sensor responses to each sample were presented as curves representing one of the 16 sensors.

The sampling time was 0.01s and the data logging period lasted 340 s. All recorded signals were calibrated by eliminating noise. From each signal, 9 commonly used features were extracted for e-nose (a total of 9 features* 16 sensors=144 features). All features were compiled together in a 72* 144 matrix (https://en.wikipedia.org/wiki/Matrix_(mathematics)), a column represented a feature and a row represented a single signal. The columns were normalized by subtracting the minimum and dividing the range of each characteristic. The analysis was done with MATLAB R2017b. “MATLAB (short for MATrix LABoratory) is a numerical computing system that provides an integrated development environment (IDE) with its own programming language (M language). It is available for Unix, Windows, macOS and GNU/Linux platforms”.The models 1NN (CP-1NN) and 3NN (CP-3NN) were trained (XNN stands for X layer neural network). Measurement of nonconformity was based on k-nearest neighbours. “The k nearest neighbors method is a supervised classification method that is used to estimate the density function of the predictors x for each class” The results obtained by the forced conformity prediction in 1NN were 87, 50% and with 3NN 83, 33% and by simple prediction in 1NN were 87,50% and 3NN 81,94%. The leave-one-out cross validation (LOOCV) method was adapted with accuracy of 98.94%.

Van de Goor (127) included in his study 60 patients from the Maastricht University Medical Centre diagnosed with histologically proven primary small cell lung carcinoma and histologically proven non-small cell lung carcinoma and 107 patients visiting the ENT department for benign processes, who were used as healthy controls. Five Aeonoses were used in the study consisting of three metal oxide sensors of micro heating plate with different surface properties (AS-MLV sensors, Applied Sensors GmbH) and a Tenax tube.”Tenax ® TA it is a traditional sorbent (porous polymer) for trapping medium to high boiling point compounds; it is especially useful for low concentrations due to its low background. Tenax TA is hydrophobic and is suitable for use in EPA TO-17 or IP-1B method and other thermal desorption applications “ Exhaled air samples were collected in a fixed scheme: heating plates were periodically heated and cooled between 260 and 340°C in 64 steps for 36 intervals of approximately 20 seconds. During recollection, samples were filtered through charcoal filters from bacterial and viral contamination. Each patient had to exhale through the e-nose for 5 minutes. The total measurement cycle lasted approximately 15 minutes and 36 data were recorded 64 times for five sensors. A Tucker3-like solution for tensor decomposition was used to compress all data.” In mathematics, the Tucker decomposition decomposes a tensor into a set of matrices and a small central tensor”. Some

of the data was pre-labeled as benign or malignant to train the artificial neural networks (ANN). The Leave-10%-Out method was used for cross-validation of the data. ANN was trained on the results of 52 lung cancer patients and 93 healthy controls to separate benign from malignant samples. After training, the ANN model was introduced to count the blinded group of patients (8 lung cancer patients and 14 healthy controls). The model achieved a sensitivity of 83%, specificity of 84%, and an accuracy of 83% for differentiating lung cancer patients from healthy controls.

Mirzaee-Ghaleh (128) based his study using an e-nose, to compare between refrigerated and frozen-thawed chicken meat. The frozen samples were taken 24 h before the fresh samples. The frozen chicken was defrosted for approximately 24 h in the refrigerator at 4° C. The chicken meat was separated from bones and skin, divided into two groups: breasts and other parts then washed with cold water and left in the open air for approximately 20 min. Following, each piece was divided into halves, placed in a vacuum package and kept in a refrigerator at 4°C for 5 days. In the study, the e-nose had to distinguish between fresh and frozen-defrosted chicken meat and also classify four groups into five types according to each 5 days. Thirty samples were taken from the refrigerator each day and analysed. In the end, the researchers achieved 600 records (4 groups × 30 samples × 5 days). An e-nose with an eight metal oxide semiconductor sensor was used for the test. The sample was kept in a cylindrical box in warm water to achieve the appropriate temperature for approximately 180 seconds to mix the air with the sample odour. To achieve the baseline correction step, the sample was inserted into the box for approximately 200 seconds and clean air was injected through the sensors to obtain a stabilized state. Thereafter, the gas was transferred around the sample during 180 seconds and the output voltage variation was measured for each sensor. The chamber and sensors were then flushed with clean air and any residual odour remaining inside the chamber using a pump for approximately 120 seconds. In this study the authors used Fuzzy K-nearest neighbours (F-KNN) algorithm for classification. The F- KNN is mainly used to examine the similarity between the training and test data set. 70% of the data were used to train the device and the remaining 30% for validation. The average accuracy value achieved for the 5-day shelf-life estimation of fresh chicken meat was 95.2% and the shelf-life of frozen-thawed chicken meat was 94.67%.

Future of the E-Nose in the Xxi CenturyIn the proper detection of substances to be analysed with an e-nose, every step is relevant, especially when the amount of volatile compounds to be studied is very small. For example, in the phase of sampling, errors can already begin. Any change in the environment can affect the samples examined. External temperature, pressure, humidity, gas velocity and odour concentration have an impact on the final result, as do volatiles exhaled from the lungs, contaminated by food, or substances present in the environment. The use of an e-nose system in an open medium or on mobile devices encounters many difficulties because the sample preparation must be carefully

26


verified and executed under defined conditions to achieve high repeatability of results and expected measurement accuracy.Often, the most frequent problem is the small size of the samples to be analysed. Most electronic noses require high quality volatile compounds for training and validation. Some types of sensors, e.g. MOS, need to be heated before detecting odours, which prolongs the measurement time and consumes energy. In addition, high humidity limits the performance of the sensors, decreasing their potential to detect different compounds.

Another problem today is the limited ability of today’s artificial noses to detect several odours with different chemical components simultaneously, even if the matrix has several sensors.

Artificial intelligence-based techniques need many different samples to train and validate their neural networks. Therefore large databases of volatile bodies with open access should be developed for faster and better work of electronic noses and to provide users with comparable results. Therefore it is necessary to standardize all research methods.

The instruments to be developed in the future should move in the direction of being devices for personal use and commercially available at low costs. These new devices should meet some special conditions, such as being able to measure a large variety of different types of volatile compounds in a short period of time without the need for large sample preparation, being accurate, proportionally inexpensive, portable and small. The sensors should be characterized for having a long lifetime, high sensitivity and specificity. Another way is to minimize the sample required for the detection process. Sensors should be less sensitive to the environment. This is why new types of sensing materials should be used in electronic noses. The application of nanotechnology provides the opportunity to minimize the size of frequency sensors and perhaps allow the use of a larger number of different sensors in the same array to detect more complex substances and have more universal applications. “Nanotechnology is the study and manipulation of matter in incredibly small sizes, usually between one and 100 nanometers. Nanotechnology encompasses a very wide range of materials, manufacturing processes and technologies that are used to create and improve many products that people use every day” It should also be considered to collect a combination of a limited number of non-specific sensors to identify complex odours. This means that no single sensor is responsible for determining the volatile compounds, but a common work of some of them can give a unique fingerprint of an odour. This idea is called combinatorial selectivity and works successfully in the world of animals (129).

Nano-medicine is the application of nanotechnology to medicine and with it we can treat diseases from inside the human body at the cellular or molecular level. The properties of certain nano-materials allow for early diagnosis of many diseases, whether oncological or neurodegenerative. This is why the future is moving towards creating an artificial nose

at a nano-scale, with electronic sensors capable of detecting nanoparticles or carbon nanotubes with different functions. Taking into account that the presence of cancer inside the body produces the release of volatile biomarkers either in blood, urine, feces, saliva, milk, semen, vaginal secretions or in the expired air itself, the artificial nose in the nano-scale modality will be an easy, non-invasive and effective system to detect early stages of lung, breast and other cancers simply by detecting volatile substances as certain animals can do.

A number of papers had been published in the literature on the ability of trained dogs to detect human diseases such as melanoma (130), bladder cancer (131), and other diseases (132), bladder cancer, breast and lung cancer (133), prostate and colorectal cancer (134).

Guirao and Molins demonstrated that trained dogs are able to detect early stage lung cancer by sniffing exhaled breath samples in a carrier with an accuracy of 95%, a specificity of 98%, a positive predictive value of 95%, a negative predictive value of 98% and an area under the ROC curve of 0.971 (135). Another possibility is to combine the olfactory function with the e-nose and the taste function with an e-tongue. The simultaneous use of both devices works synergistically, increasing the amount of information and influencing sensitivity and specificity.

Conclusion“Artificial noses” are devices that can soon have many promising applications, as they can provide, in a short time, a lot of information about the state of health of people, the quality of food, environmental pollution, analysis of beverages and food products, composition and quality of essences and perfumery products, etc. The development of the materials that make up the artificial nose with its matrix and the design of new sensors with compensatory functions to compensate for deviations caused by contamination are advancing rapidly. The application of nano-medicine and the evaluation system of nano-samples for analysis will enable e-nose to have more selectivity and sensitivity. Another goal is to achieve miniaturization of e-nose to make them portable and robust devices, which can be used as routine on-site analysis tools. To carry out this project it is important to have an extensive team of experts in the different aspects of bioengineering, medicine and artificial intelligence. The possibilities that open up in this regard are many and varied. From the UIC (International University of Catalonia) the whole team will collaborate to achieve very interesting results shortly.

AcknowledgmentsThis work was supported by Mr Salvador Ramón and Clarós Foundation.

Author Contributions Clarós P , Dabkowska A and Clarós A., wrote the paper. All authors listed have also made substantial, direct, and intellectual contribution to the work, and approved it for publication.

27


Conflict of Interest Statement The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References1. Spehr M, Munger SD. Olfactory receptors: GPCRs and

beyond. J Neurochem. 2009;109(6): 1570–1583. 2. Brann D, Tsukahara T, Weinreb C et al. Non-neuronal

expression of SARS-CoV-2 entry genes in the olfactory system suggests mechanisms underlying COVID-19-associated anosmia.Science Advances 31 Jul 2020:Vol. 6, no. 31, eabc5801 DOI: 10.1126/sciadv.abc5801

3. American Academy of Neurology. Neurology 2017;89: 1–32.

4. Malnic B, Hirono J, Sato T, Buck LB. Combinatorial receptor codes for odors. Cell. 1999;96(5):713–23.

5. Buck L, Axel R. A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell. 1991;65(1): 175–187.

6. Pierce, K., Premont, R. & Lefkowitz, R. Seven-transmembrane receptors. Nat Rev Mol Cell Biol 3: 639–650 (2002).

7. P.-Y. Jean-Charles et al. Ubiquitination and Deubiquitination of G Protein-Coupled Receptors. Progress in Molecular Biology and Translational Science,Academic Press, 2016: 141: 1-55.

8. Brattoli M, de Gennaro G, de Pinto V, Loiotile AD, Lovascio S, Penza M. Odour detection methods: olfactometry and chemical sensors. Sensors. 2011 May 16;11(5): 5290–5322.

9. Kondyli A, Schrader W. High-resolution GC/MS studies of a light crude oil fraction. J Mass Spectrom. 2019;54(1): 47–54.

10. Persaud K, Dodd G. Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose. Nature. 1982;299(5881): 352–355.

11. Hu W, Wan L, Jian Y, Ren C, Jin K, Su X, et al. Electronic noses: from advanced materials to sensors aided with data processing. Adv Mater Technol. 2018; 1800488.

12. Kress-Rogers E. Handbook of biosensors and electronic noses: medicine, food, and the environment. CRC Press. 1996.

13. Deshmukh S, Bandyopadhyay R, Bhattacharyya N, Pandey RA, Jana A. Application of electronic nose for industrial odors and gaseous emissions measurement and monitoring-an overview. Talanta. 2015;144: 329–340.

14. Wilson AD, Baietto M. Applications and Advances in Electronic-Nose Technologies. Sensors. 2009;9(7): 5099–5148.

15. Egashira M, Shimizu Y. Odor sensing by semiconductor metal oxides. Sens Actuators B Chem. 1993;13(1): 443–446.

16. Nanto H, Sokooshi H, Kawai T. Aluminum-doped ZnO thin film gas sensor capable of detecting freshness of sea foods. Sens Actuators B Chem. 1993;14(1): 715–7.

17. Shurmer HV, Gardner JW, Chan HT. The application of

discrimination techniques to alcohols and tobacco using tin oxide sensors. Sens Actuat. 1989;18: 359–69.

18. Yim HS, Kibbey CE, Ma SC, Kliza DM, Liu D, Park SB, et al. Polymer membrane-based ion-, gas- and bio-selective potentiometric sensors. Biosens Bioelectron. 1993;8(1): 1–38.

19. Pisanelli A, Qutob A, Travers P, Szyszko S, Persaud K. Applications of multi-array polymer sensors to food industries. Life Chem Rep. 1994;(11): 303–8.

20. Lonergan MC, Severin EJ, Doleman BJ, Beaber SA, Grubbs RH, Lewis NS. Array-based vapor sensing using chemically sensitive, carbon black−polymer resistors. Chem Mater. 1996;8(9):2298–312.

21. Freund MS, Lewis NS. A chemically diverse conducting polymer-based “electronic nose”. Proc Natl Acad Sci. 1995;92(7): 2652–6.

22. Hatfield JV, Neaves P, Hicks P, Persaud K, Travers P. Towards an integrated electronic nose using conducting polymer sensors. Sens Actuators B Chem. 1994;18(1–3): 221–8.

23. Persaud KC, Qutob AA, Travers P, Pisanelli AM, Szyszko S. Odor evaluation of foods using conducting polymer arrays and neural net pattern recognition. In: Kurihara K, Suzuki N, Ogawa H, editors. Olfaction and Taste XI. Tokyo: Springer Japan; 1994. p. 708–10.

24. Staples EJ. Electronic nose simulation of olfactory response containing 500 orthogonal sensors in 10 seconds. In: 1999 IEEE Ultrasonics Symposium Proceedings International Symposium (Cat No99CH37027) [Internet]. Caesars Tahoe, NV: IEEE; 1999 [cited 2021 Jan 28]. p. 417–23 vol.1. Available from: https://ieeexplore.ieee.org/document/849432/

25. Demarne V, Grisel A, Sanjinks R, Rosenfeld D, Lévy F. Electrical transport properties of thin polycrystalline SnO2 film sensors. Sens Actuat B Chem. 1992;7: 704–8.

26. Schaller E, Bosset JO, Escher F. ‘Electronic noses’ and their application to food. LWT - Food Sci Technol. 1998;31(4): 305–16.

27. Lundström I, Shivaraman S, Svensson C, Lundkvist L. A hydrogen−sensitive MOS field−effect transistor. Appl Phys Lett. 1975;26(2): 55–7.

28. Åbom A, Comini E, Sberveglieri G, Hultman L, Eriksson M. Thin oxide films as surface modifiers of MIS field effect gas sensors. Sens Actuat B Chem. 2002;85: 109–19.

29. Drafts B. Acoustic wave technology sensors. IEEE Trans Microw Theory Tech. 2001;49(4): 795–802.

30. Curie J, Curie P. An oscillating quartz crystal mass detector. Rendu. 1880;91:294–7.

31. Sauerbrey G. Use of quartz vibration for weighing thin films on a microbalance. Phys J. 1959;155: 206–12.

32. Dunst K, Jasiński P. Gas sensors based on conducting polymers-recent developments. Zesz Nauk Wydz ETI Politech Gdań Technol Inf. 2018; 73–80.

33. Cui S, Ling P, Zhu H, Keener H. Plant pest detection using an artificial nose system: a review. Sensors. 2018 Jan 28;18(2): 378.

34. Länge K. Bulk and Surface Acoustic Wave Sensor Arrays for Multi-Analyte Detection: A Review. Sensors. 2019

28


Jan;19(24): 5382. 35. Jean-Michel Guillot. E-noses: actual limitations and

perspectives for environmental odour analysis. Chem Eng Trans. 2016 Sep;54: 223–8.

36. Vashist SK, Vashist P. Recent advances in quartz crystal microbalance-based sensors. J Sens. 2011;2011: 1–13.

37. Kiani S, Minaei S, Ghasemi-Varnamkhasti M. Fusion of artificial senses as a robust approach to food quality assessment. J Food Eng. 171: 230–9.

38. Son M, Lee JY, Ko HJ, Park TH. Bioelectronic onse: an emerging tool for odor standardization. Trends Biotechnol. 2017;35(4): 301–7.

39. Xun W, YuChen Z, Di W, ZhenBo W, KunSong C. Rapid and non-destructive detection of decay in peach fruit at the cold environment using a self-developed handheld electronic-nose system. Food Anal Methods. 2018;11(11): 2990–3004.

40. Qiu S, Wang J, Gao L. Qualification and quantisation of processed strawberry juice based on electronic nose and tongue. LWT - Food Sci Technol. 2015;60: 115–23.

41. Aouadi B, Zaukuu J-LZ, Vitális F, Bodor Z, Fehér O, Gillay Z, et al. Historical Evolution and Food Control Achievements of Near Infrared Spectroscopy, Electronic Nose, and Electronic Tongue—Critical Overview. Sensors. 2020;20(19): 5479.

42. Broza YY, Mochalski P, Ruzsanyi V, Amann A, Haick H. Hybrid volatolomics and disease detection. Angew Chem Int Ed Engl. 2015;54(38): 11036–48.

43. Pauling L, Robinson AB, Teranishi R, Cary P. Quantitative analysis of urine vapor and breath by gas-liquid partition chromatography. Proc Natl Acad Sci U S A. 1971;68(10): 2374–6.

44. Hiroshi K, Masaya H, Nariyoshi S, Makoto M. Evaluation of volatile sulfur compounds in the expired alveolar gas in patients with liver cirrhosis. Clin Chim Acta. 1978;85(3): 279–84.

45. Simenhoff ML, Burke JF, Saukkonen JJ, Ordinario AT, Doty R, Dunn S. Biochemical profile of uremic breath. N Engl J Med. 1977;297(3): 132–5.

46. O’Neill HJ, Gordon SM, O’Neill MH, Gibbons RD, Szidon JP. A computerized classification technique for screening for the presence of breath biomarkers in lung cancer. Clin Chem. 1988;34(8): 1613–8.

47. Phillips M, Gleeson K, Hughes JM, Greenberg J, Cataneo RN, Baker L, et al. Volatile organic compounds in breath as markers of lung cancer: a cross-sectional study. Lancet Lond Engl. 1999;353(9168): 1930–3.

48. Wilson AD. Noninvasive early disease diagnosis by electronic-nose and related VOC-detection devices. Biosensors. 2020;10(7): 73.

49. Tiele A, Wicaksono A, Kansara J, Arasaradnam RP, Covington JA. Breath analysis using e-nose and ion mobility technology to diagnose inflammatory bowel disease-a pilot study. Biosensors. 2019;9(2).

50. Deianova N, El Manouni El Hassani S, Niemarkt H, Cossey C, van Kaam A, F J, et al. Fecal volatile organic compound profiles are not influenced by gestational age and mode of delivery: a longitudinal multicenter cohort

study. Biosensors. 2020;10(5). 51. Gardner JW, Craven M, Dow C s, Hines E. Prediction of

bacteria type and culture growth phase by an electronic nose with a multi-layer perceptron network. Meas Sci Technol. 1999;9:120.

52. Sánchez C, Santos J, Lozano J. Use of Electronic Noses for Diagnosis of Digestive and Respiratory Diseases through the Breath. Biosensors. 2019 Feb 28;9(1): 35.

53. Shih-Wen Chiu, Jen-Huo Wang, Kwuang-Han Chang, Ting-Hau Chang, Chia-Min Wang, Chia-Lin Chang, et al. A Fully Integrated Nose-on-a-Chip for Rapid Diagnosis of Ventilator-Associated Pneumonia. IEEE Trans Biomed Circuits Syst. 2014 Dec;8(6): 765–78.

54. Lai SY, Deffenderfer OF, Hanson W, Phillips MP, Thaler ER. Identification of upper respiratory bacterial pathogens with the electronic nose. The Laryngoscope. 2002;112(6): 975–9.

55. Thaler ER, Hanson CW. Use of an electronic nose to diagnose bacterial sinusitis. Am J Rhinol. 2006;20(2): 170–2.

56. Yusuf N, Zakaria A, Omar MI, Shakaff AYM, Masnan MJ, Kamarudin LM, et al. In-vitro diagnosis of single and poly microbial species targeted for diabetic foot infection using e-nose technology. BMC Bioinformatics. 2015;16: 158.

57. Holmberg M, Gustafsson F, Hörnsten EG, Winquist F, Nilsson LE, Ljung L, et al. Bacteria classification based on feature extraction from sensor data. Biotechnol Tech. 1998;12(4): 319–24.

58. Pavlou A, Magan N, Sharp D, Brown J, Barr H, Turner A. An intelligent rapid odour recognition model in discrimination of Helicobacter pylori and other gastroesophageal isolates in vitro. Biosens Bioelectron. 2000;

59. Ping W, Yi T, Haibao X, Farong S. A novel method for diabetes diagnosis based on electronic nose1Paper presented at Biosensors ’96, Bangkock, May 1996.1. Biosens Bioelectron. 1997;12(9): 1031–6.

60. Di Natale C, Mantini A, Macagnano A, Antuzzi D, Paolesse R, D’Amico A. Electronic nose analysis of urine samples containing blood. Phyisiological Meas. 1999;20(4): 377–84.

61. Di Natale C, Macagnano A, Paolesse R, Tarizzo E, Mantini A, D’Amico A. Human skin odor analysis by means of an electronic nose. Sens Actuators B Chem. 2000;65(1): 216–9.

62. Saidi T, Zaim O, Moufid M, El Bari N, Ionescu R, Bouchikhi B. Exhaled breath analysis using electronic nose and gas chromatography–mass spectrometry for non-invasive diagnosis of chronic kidney disease, diabetes mellitus and healthy subjects. Sens Actuators B Chem. 2018;257: 178–88.

63. D’Amico A, Pennazza G, Santonico M, Martinelli E, Roscioni C, Galluccio G, et al. An investigation on electronic nose diagnosis of lung cancer. Lung Cancer Amst Neth. 2010 May;68(2): 170–6.

64. Ballantine D, Wohltjen H. Surface acoustic wave. Devices for chemical analysis. Anal Chem. 1989;61(11).

65. Brunink JAJ, Di Natale C, Bungaro F, Davide FAM, D’Amico A, Paolesse R, et al. The application of

29


metalloporphyrins as coating material for quartz microbalance-based chemical sensors. Anal Chim Acta. 1996;325(1): 53–64.

66. Bikov A, Hernadi M, Korosi BZ, Kunos L, Zsamboki G, Sutto Z, et al. Expiratory flow rate, breath hold and anatomic dead space influence electronic nose ability to detect lung cancer. BMC Pulm Med. 2014 Dec 16;14(1): 202.

67. Dragonieri S, Annema JT, Schot R, van der Schee MPC, Spanevello A, Carratú P, et al. An electronic nose in the discrimination of patients with non-small cell lung cancer and COPD. Lung Cancer Amst Neth. 2009 May;64(2): 166–70.

68. Peng G, Hakim M, Broza YY, Billan S, Abdah-Bortnyak R, Kuten A, et al. Detection of lung, breast, colorectal, and prostate cancers from exhaled breath using a single array of nanosensors. Br J Cancer. 2010 Aug 10;103(4): 542–51.

69. Musatov VYu, Sysoev VV, Sommer M, Kiselev I. Assessment of meat freshness with metal oxide sensor microarray electronic nose: A practical approach. Sens Actuators B Chem. 2010;144(1): 99–103.

70. Zhang Z, Tong J, Chen D, Lan Y. Electronic nose with an air sensor matrix for detecting beef freshness. J Bionic Eng. 2008;5(1): 67–73.

71. Haddi Z, Barbri NE, Tahri K, Bougrini M, Bari NE, Llobet E, et al. Instrumental assessment of red meat origins and their storage time using electronic sensing systems. Anal Methods. 2015;7(12): 5193–203.

72. Tian X, Wang J, Cui S. Analysis of pork adulteration in minced mutton using electronic nose of metal oxide sensors. J Food Eng. 2013;119(4): 744–9.

73. Pattarapon P, Zhang M, Bhandari B, Gao Z. Effect of vacuum storage on the freshness of grass carp (Ctenopharyngodon idella) fillet based on normal and electronic sensory measurement. J Food Process Preserv. 2018;42(2): e13418.

74. Güney S, Atasoy A. Study of fish species discrimination via electronic nose. Comput Electron Agric. 2015;119: 83–91.

75. Semeano ATS, Maffei DF, Palma S, Li RWC, Franco BDGM, Roque ACA, et al. Tilapia fish microbial spoilage monitored by a single optical gas sensor. Food Control. 2018;89: 72–6.

76. Marilley L, Ampuero S, Zesiger T, Casey MG. Screening of aroma-producing lactic acid bacteria with an electronic nose. Int Dairy J. 2004;10(14): 849–56.

77. Trihaas J, Vognsen L, Nielsen PV. Electronic nose: new tool in modelling the ripening of Danish blue cheese. Int Dairy J. 2005;15(6): 679–91.

78. Benedetti S, Sinelli N, Buratti S, Riva M. Shelf life of crescenza cheese as measured by electronic nose. J Dairy Sci. 2005;88(9): 3044–51.

79. Tohidi M, Ghasemi-Varnamkhasti M, Ghafarinia V, Saeid Mohtasebi S, Bonyadian M. Identification of trace amounts of detergent powder in raw milk using a customized low-cost artificial olfactory system: A novel method. Measurement. 2018;124: 120–9.

80. Ampuero S, Bogdanov S, Bosset J-O. Classification of unifloral honeys with an MS-based electronic nose using different sampling modes: SHS, SPME and INDEX. Eur Food Res Technol. 2004;218(2): 198–207.

81. Dymerski T, Gębicki J, Wardencki W, Namieśnik J. Application of an electronic nose instrument to fast classification of Polish honey types. Sensors. 2014;14(6): 10709–24.

82. Huang L, Liu H, Zhang B, Wu D. Application of Electronic Nose with Multivariate Analysis and Sensor Selection for Botanical Origin Identification and Quality Determination of Honey. Food Bioprocess Technol. 2015;8(2): 359–70.

83. Faal S, Loghavi M, Kamgar S. Physicochemical properties of Iranian ziziphus honey and emerging approach for predicting them using electronic nose. Measurement. 2019;148: 106936.

84. Holscher W, Vitzthum OG, Steinhart H. Identification and sensorial evaluation of aroma-impact-compounds in roasted Colombian coffee. Cafe-Cacao-. 1990;34(3): 205–12.

85. Giungato P, Laiola E, Nicolardi V. Evaluation of industrial roasting degree of coffee beans by using an electronic nose and a stepwise backward selection of predictors. Food Anal Methods. 2017;10(10): 3424–33.

86. Marek G, Dobrzański B, Oniszczuk T, Combrzyński M, Ćwikła D, Rusinek R. Detection and differentiation of volatile compound profiles in roasted coffee arabica beans from different countries using an electronic nose and GC-MS. Sensors. 2020;20(7): 2124.

87. Dong W, Zhao J, Hu R, Dong Y, Tan L. Differentiation of Chinese robusta coffees according to species, using a combined electronic nose and tongue, with the aid of chemometrics. Food Chem. 2017;229: 743–51.

88. Dong W, Hu R, Long Y, Li H, Zhang Y, Zhu K, et al. Comparative evaluation of the volatile profiles and taste properties of roasted coffee beans as affected by drying method and detected by electronic nose, electronic tongue, and HS-SPME-GC-MS. Food Chem. 2019;272: 723–31.

89. Brudzewski K, Osowski S, Dwulit A. Recognition of coffee using differential electronic nose. IEEE Trans Instrum Meas. 2012;61(6): 1803–10.

90. Pardo M, Sberveglieri G. Coffee analysis with an electronic nose. IEEE Trans Instrum Meas. 2002 Dec;51(6): 1334–9.

91. Peters Y, Schrauwen RWM, Tan AC, et al.Detection of Barrett’s oesophagus through exhaled breath using an electronic nose Device. Gut 2020;69: 1169-1172.

92. Dutta R, Kashwan KR, Bhuyan M, Hines EL, Gardner JW. Electronic nose based tea quality standardization. Neural Netw. 2003;16(5): 847

93. Sharmilan T, Premarathne I, Wanniarachchi I, Kumari S, Wanniarachchi D. Electronic nose technologies in monitoring black tea manufacturing process. J Sens. 2020;2020: 1–8.

94. Ragazzo-Sanchez JA, Chalier P, Chevalier-Lucia D, Calderon-Santoyo M, Ghommidh C. Off-flavours detection in alcoholic beverages by electronic nose coupled to GC. Sens Actuators B Chem. 2009;140(1): 29–34.

95. Aleixandre M, Cabellos JM, Arroyo T, Horrillo MC.

30


Quantification of wine mixtures with an electronic nose and a human panel. Front Bioeng Biotechnol. 2018 Feb 12;6: 14.

96. Rodríguez-Méndez ML, De Saja JA, González-Antón R, García-Hernández C, Medina-Plaza C, García-Cabezón C, et al. Electronic noses and tongues in wine Industry. Front Bioeng Biotechnol [Internet]. 2016 Oct 25 [cited 2021 Feb 1];4. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5078139/

97. Moreno i Codinachs L, Kloock JP, Schöning MJ, Baldi A, Ipatov A, Bratov A, et al. Electronic integrated multisensor tongue applied to grape juice and wine analysis. The Analyst. 2008;133(10): 1440.

98. Medina-Plaza C, Revilla G, Muñoz R, Fernández-Escudero JA, Barajas E, Medrano G, et al. Electronic tongue formed by sensors and biosensors containing phthalocyanines as electron mediators: Application to the analysis of red grapes. J Porphyr Phthalocyanines. 2014 Jan;18(01n02): 76–86.

99. Peris M, Escuder-Gilabert L. On-line monitoring of food fermentation processes using electronic noses and electronic tongues: A review. Anal Chim Acta. 2013 Dec;804: 29–36.

100. Wei YJ, Yang LL, Liang YP, Li JM. Application of Electronic Nose for Detection of Wine-Aging Methods. Adv Mater Res. 2014 Feb;875–877:2206–13.

101. Prieto N, Gay M, Vidal S, Aagaard O, de Saja JA, Rodriguez-Mendez ML. Analysis of the influence of the type of closure in the organoleptic characteristics of a red wine by using an electronic panel. Food Chem. 2011 Nov;129(2): 589–94.

102. Gil-Sánchez L, Soto J, Martínez-Máñez R, Garcia-Breijo E, Ibáñez J, Llobet E. A novel humid electronic nose combined with an electronic tongue for assessing deterioration of wine. Sens Actuators Phys. 2011 Nov;171(2): 152–8.

103. Penza M, Cassano G. Recognition of adulteration of Italian wines by thin-film multisensor array and artificial neural networks. Anal Chim Acta. 2004 May;509(2):159–77.

104. Santos JP, Lozano J, Aleixandre M, Arroyo T, Cabellos JM, Gil M, et al. Threshold detection of aromatic compounds in wine with an electronic nose and a human sensory panel. Talanta. 2010 Mar 15;80(5): 1899–906.

105. Turner AP, Magan N. Electronic noses and disease diagnostics. Nat Rev Microbiol. 2004;2(2): 160-66.

106. Cosio MS, Ballabio D, Benedetti S, Gigliotti C. Geographical origin and authentication of extra virgin olive oils by an electronic nose in combination with artificial neural networks. Anal Chim Acta. 2006;567(2): 202–10.

107. Cosio MS, Ballabio D, Benedetti S, Gigliotti C. Evaluation of different storage conditions of extra virgin olive oils with an innovative recognition tool built by means of electronic nose and electronic tongue. Food Chem. 2007;101(2): 485–91.

108. Hai Z, Wang J. Electronic nose and data analysis for detection of maize oil adulteration in sesame oil. Sens Actuators B Chem. 2006;119(2): 449–55.

109. Hai Z, Wang J. Detection of adulteration in camellia seed oil and sesame oil using an electronic nose. Eur J Lipid Sci Technol. 2006;108(2): 116–24.

110. Triyana K, Taukhid Subekti M, Aji P, Nur Hidayat S, Rohman A. Development of electronic nose with low-cost dynamic headspace for classifying vegetable oils and animal fats [Internet]. Vol. 771, Applied Mechanics and Materials. Trans Tech Publications Ltd; 2015 [cited 2021 Jan 28]. p. 50–4. Available from: https://www.scientific.net/AMM.771.50

111. Cui S, Inocente EAA, Acosta N, Keener HaroldM, Zhu H, Ling PP. Development of fast e-nose system for early-stage diagnosis of aphid-stressed tomato plants. Sensors. 2019;19(16): 3480.

112. Berna AZ, Lammertyn J, Saevels S, Natale CD, Nicolaı̈ BM. Electronic nose systems to study shelf life and cultivar effect on tomato aroma profile. Sens Actuators B Chem. 2004;2–3(97): 324–33.

113. Chen H-Z, Zhang M, Bhandari B, Guo Z. Evaluation of the freshness of fresh-cut green bell pepper (Capsicum annuum var. grossum) using electronic nose. LWT. 2018;87: 77–84.

114. Chen H, Zhang M, Guo Z. Discrimination of fresh-cut broccoli freshness by volatiles using electronic nose and gas chromatography-mass spectrometry. Postharvest Biol Technol. 2019;148: 168–75.

115. Wang J, Liang G, Tian H, Sun J, Wan C. Electronic nose-based technique for rapid detection and recognition of moldy apples. 2019 [cited 2021 Jan 28]; Available from: https://www.preprints.org/manuscript/201903.0008/v2

116. Ezhilan M, Nesakumar N, Babu KJ, Srinandan CS, Rayappan JBB. Freshness assessment of broccoli using electronic nose. 2019;

117. Tan J, Balasubramanian B, Sukha D, Ramkissoon S. Sensing cocoa (Theobroma cacao L.) beans fermentation by electronic nose system. [cited 2021 Jan 29]; Available from: https://www.semanticscholar.org/paper/Sensing-Cocoa-(Theobroma-cacao-L.)-Beans-by-Nose-Tan-Balasubramanian/0115e2dfecc667dfbff0ae7776d9adbdb0855661

118. Tan J, Balasubramanian B, Sukha D, Ramkissoon S, Umaharan P. Sensing fermentation degree of cocoa (Theobroma cacao L.) beans by machine learning classification models based electronic nose system. J Food Process Eng. 2019;42(6): e13175.

119. Valdez LF, Gutiérrez JM. Chocolate classification by an electronic nose with pressure controlled generated stimulation. Sensors. 2016 Oct;16(10): 1745.

120. Rahman IN, Lelono D, Triyana K. Klasifikasi kakao berbasis e-nose dengan metode neuro fuzzy. IJEIS Indones J Electron Instrum Syst. 2018;8(1):undefined-undefined.

121. Rottiers H, Tzompa Sosa DA, Van de Vyver L, Hinneh M, Everaert H, De Wever J, et al. Discrimination of cocoa liquors based on their odor fingerprint: a fast GC electronic nose suitability study. Food Anal Methods. 2019 Feb 1;12(2): 475–88.

122. Olu loyo V, Dinrifo R. Neural network-based electronic nose for cocoa beans quality assessment. Agric Eng Int

31


CIGR J. 2011 Jan 1;13. 123. Branca A, Simonian P, Ferrante M, Novas E, Negri RM.

Electronic nose based discrimination of a perfumery compound in a fragrance. Sens Actuators B Chem. 2003 Jul 1;92(1): 222–7.

124. Eamsa-ard T, Swe MM, Seesaard T, Kerdcharoen T. Development of electronic nose for evaluation of fragrance and human body odor in the cosmetic industry. In: 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE) [Internet]. Nara, Japan: IEEE; 2018 [cited 2021 Feb 2]. p. 363–4. Available from: https://ieeexplore.ieee.org/document/8574831/

125. Gebicki J. Application of electrochemical sensors and sensor matrixes for measurement of odorous chemical compounds. TrAC Trends Anal Chem. 2016;77: 1–13.

126. Zhan X, Wang Z, Yang M, Luo Z, Wang Y, Li G. An electronic nose-based assistive diagnostic prototype for lung cancer detection with conformal prediction. Measurement. 2020;158: 107588.

127. van de Goor R, van Hooren M, Dingemans A-M, Kremer B, Kross K. Training and validating a portable electronic nose for lung cancer screening. J Thorac Oncol. 2018;13(5): 676–81.

128. Mirzaee-Ghaleh E, Taheri-Garavand A, Ayari F, Lozano J. Identification of Fresh-Chilled and Frozen-Thawed Chicken Meat and Estimation of their Shelf Life Using an E-Nose Machine Coupled Fuzzy KNN. Food Anal Methods. 2020 Mar 1;13(3): 678–89.

129. Grabe V, Sachse S. Fundamental principles of the olfactory code. Biosystems. 2018 Feb 1;164: 94–101.

130. Church J, Williams H. Another sniffer dog for the clinic? The Lancet. 2001 Sep 15;358(9285): 930.

131. Willis CM, Church SM, Guest CM, Cook WA, McCarthy N, Bransbury AJ, et al. Olfactory detection of human bladder cancer by dogs: proof of principle study. BMJ. 2004 Sep 25;329(7468): 712.

132. McCulloch M, Jezierski T, Broffman M, Hubbard A, Turner K, Janecki T. Diagnostic accuracy of canine scent detection in early- and late-stage lung and breast cancers. Integr Cancer Ther. 2006 Mar;5(1): 30–9.

133. Cornu J-N, Cancel-Tassin G, Ondet V, Girardet C, Cussenot O. Olfactory detection of prostate cancer by dogs sniffing urine: a step forward in early diagnosis. Eur Urol. 2011 Feb;59(2): 197–201.

134. Sonoda H, Kohnoe S, Yamazato T, Satoh Y, Morizono G, Shikata K, et al. Colorectal cancer screening with odour material by canine scent detection. Gut. 2011 Jun;60(6): 814–9.

135. Guirao Montes Á, Molins López-Rodó L, Ramón Rodríguez I, Sunyer Dequigiovanni G, Viñolas Segarra N, Marrades Sicart RM, et al. Lung cancer diagnosis by trained dogs. Eur J Cardio-Thorac Surg Off J Eur Assoc Cardio-Thorac Surg. 2017 Dec 1;52(6): 1206–10.

Copyright: ©2021 Clarós P. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in anymedium, provided the original author and source are credited.

What is the “Status questionis” of the e-nose. The future of ...... (6) Figure 5: G-protein...

Documents

Transcript of What is the “Status questionis” of the e-nose. The future of ...... (6) Figure 5: G-protein...