10 IEEE Instrumentation & Measurement Magazine April 20181094-6969/18/$25.00©2018IEEE

M echanical removal of blood clots is a promis-ing approach towards the treatment of vascular diseases caused by pathological clot formation

in the circulatory system. These clots can form and travel to deep seated regions in the circulatory system and result in sig-nificant problems as blood flow past the clot is obstructed. A microscopically small helical microrobot (Fig. 1a) offers great promise in the minimally-invasive removal of these clots. The simple design of the microrobot, which was originally presented at Tohoku University by Ishiyama et al., enables fab-rication at micro and nano scales [1]. The incorporation of a magnetic material to the helical microrobots allows them to controllably navigate using an external source of magnetic field and feedback control [2]. The external source of magnetic field has been provided using a configuration of electromag-netic coils or rotating permanent magnets, while several research groups have relied on visual feedback to design feed-back control systems.

The greatest power of these microrobots in biomedical ap-plications has emerged when clinical imaging modalities were incorporated to provide feedback and enable accurate in vivo tracking and control towards a desired position in three-di-mensional space. Sylvain Martel's and Brad Nelson's groups have succeeded in achieving control of untethered microro-bots, magnetotactic bacteria [3], and artificial bacterial flagella [4] using magnetic resonance imaging and fluorescence-based in vivo imaging, respectively. These ongoing in vivo attempts indicate that we will be able to use microrobots to achieve new minimally-invasive diagnosis and therapeutic procedures. For example, the microrobot that we are using consists of a helical body and a magnetic head (Fig. 1b), with a magnetization vec-tor perpendicular to the long axis of its helical body, and can swim controllably in a vessel using a rotating magnetic field in milliTesla range. This level of control enables the microro-bots to swim controllably and selectively target a blood clot, as shown in Fig. 1c and Fig. 1d. The formation of blood clots

starts with the activation and aggregation of platelets followed by the activation of blood coagulation factors. These factors generate strands of fibrin surrounding the platelets, and more platelets and blood cells are entrapped inside the fibrin net-work. The continuous interaction between the tip of the helical microrobot and the fibrin network of the clot has proven to be efficient in decreasing the size of the clot in vitro [5]. However, the translation of this concept into in vivo trials requires several challenges to be overcome. Among these challenges is the need to achieve real-time tracking and motion control of the helical microrobot using feedback provided by a clinical imaging sys-tem. In addition, the accurate estimation of the blood clot size during the interaction of the helical microrobot is also neces-sary to study the influence of the microrobot on the clot.

Our in vitro experimental model is shown in Fig. 2. The sys-tem integrates several modules to control the motion of the helical microrobots, localize the microrobot using ultrasound feedback (Fig. 2a), and analyze the volume and composition of the clot during interaction with the microrobot. The first module is a permanent magnet-based robotic system (Fig. 2b) that consists of two rotating dipole fields. Single rotating dipole field can also be used to externally actuate the micro-robot, however, a magnetic force will pull the microrobot along its lateral direction towards the inner wall of the ves-sel. This pulling magnetic force is decreased when the two synchronized rotating dipole fields are used and the microro-bot is positioned in their common center, as shown in Fig. 2b. Rotation of the dipole fields enables the microrobot to swim through the medium inside a catheter segment that contains the blood clots and a flowing stream of phosphate buffered sa-line (PBS) (Fig. 2c). Blood clots with volume of 94.24 mm3 are prepared and inserted inside the catheter segment, and PBS is injected at a flow rate of 10 ml/hr against the direction of the swimming microrobot. The second module of the in vitro model provides imaging of the microrobot and the clot. An ul-trasound system (HD 5 Diagnostic Ultrasound System, Philips

This work was supported by funds from the German University in Cairo and the DAAD-BMBF funding project. The authors also acknowledge funding from the Science and Technology Development Fund in Egypt (No. 23016).

Ultrasound-Guided Minimally Invasive Grinding for Clearing

Blood Clots: Promises and Challenges

Dalia Mahdy, Ramez Reda, Nabila Hamdi, and Islam S. M. Khalil

April 2018 IEEE Instrumentation & Measurement Magazine 11

and Neusoft Medical Systems, Amsterdam, The Netherlands) and two digital cameras are used to localize the microrobot and visualize the clot in each trial, respectively. The third mod-ule of the in vitro model provides quantitative analysis on the pre- and post-conditions of the blood clots and also provides physical information during interaction. The three modules of the in vitro model are utilized on approximately 1-hour old blood clots, corresponding to a clinically relevant time re-garding the treatment's effectiveness of several conditions of vascular obstruction. The aim of their integration is threefold: first, to actuate the helical microrobot at a distance; second, to control the motion of the helical microrobot towards the clot under ultrasound guidance; and third, to analyze the pre- and post-conditions of the clot to verify the effectiveness of this minimally-invasive approach.

Our approach begins by localizing the microrobot using ultrasound feedback (Fig. 2a). The transducer is fixed with respect to the catheter segment which is embedded into gel-atin to achieve air-free coupling. Closed-loop motion control of the helical microrobot is implemented upon detecting its ultrasound feedback (Fig. 3). The position tracking error is

calculated based on the positions of the blood clot and the microrobot, and the angular velocity of the rotating dipole fields is adjusted such that the position error decreases as-ymptotically (Fig. 2d). This closed-loop control is followed by continuous grinding of the blood clot. Two cameras are used to simultaneously capture the morphology of the clot during each trial for off-line calculation of the clot volume. In addition, the PBS past the blood clot is collected and the number of blood cells is calculated. The volume of the clot is constructed using the visual feedback and provides overall information pertain-ing to the morphology of the clot, whereas the number of blood cells provides a measure of the influence of the helical microro-bot on the composition of the blood clot.

Fig. 2e shows experimental results conducted using 1-hour old blood clot samples. The samples are prepared using blood taken from healthy donors. This experiment is approved by Institutional Review Board, and donors' written informed con-sents are obtained. We use 4-mm-long helical microrobots with diameter of 364 μm. The swimming speed of these microrobots depends on the frequency of the rotating magnetic field. For instance, average swimming speeds of 10 mm/s and 26 mm/s

Fig. 1. Clearing of blood clot is achieved using a helical microrobot and rotating magnetic field. (a) The microrobot consists of a magnetic head and helical tail. (b) The helical tail is attached to a cylindrical magnet with axial magnetization (red arrow). (c) The tip of the helical microrobot tears the three-dimensional fibrin network of a blood clot and enables blood cells to break free. (d) An externally actuated helical microrobot decreases the size of blood clot via mechanical grinding under the influence of rotating magnetic field. The straight and curved arrows indicate the direction of swimming and rotation, respectively.

12 IEEE Instrumentation & Measurement Magazine April 2018

are observed under the influence of rotating magnetic field of 5.5 mT at frequencies of 20 Hz and 45 Hz, respectively. How-ever, under the influence of ultrasound feedback, the speed of the microrobot varies (proportional to the position error between the clot and the microrobot) and decreases as the mi-crorobot approaches the clot. The average speed of the helical microrobot is 5.32 mm/s during the closed-loop control trials, and mechanical grinding of the clot is achieved upon contact. Ischemic tissue, deprived of blood because of the clot, will not survive. Therefore, the microrobot must achieve enough grinding and dissolution of the thrombus, and hence recanali-zation of the blocked artery, within a biologically meaningful time window which imposes a significant limitation on the grinding time [6]. Hence, the grinding time is limited to 40 minutes in all of our in vitro trials. This limitation can be over-come by controlling the removal rate of the clot using the frequency of the rotating magnetic field. The removal rate of the blood clots is proportional to the rotation frequency of the

microrobot. For instance, negligible removal is observed be-low actuation frequency of 30 Hz. Removal rate of -0.23 mm3/min (n=6), -0.885 mm3/min (n=6), and -0.315 mm3/min (n=6) are measured under the influence of rotation frequency of 30 Hz, 35 Hz, and 40 Hz, respectively. Again, we observe a de-crease in the removal rate above actuation frequency of 40 Hz, owing to the increased damping at relatively high frequen-cies and due to the step-out frequency of the microrobot, i.e., frequency above which the microrobot no longer aligns with the applied magnetic field [5]. Another method of assessment of our minimally-invasive technique is the analysis of the de-composition of the collected PBS past the grinding. We count the blood cells past the clot and also observe that their number is directly proportional to the actuation frequency of the mi-crorobot. In the absence of interaction between the clot and the microrobot (Fig. 2f), the cell count is 162.75±81.25×104 cell/ml (n=6), whereas the count increases to 832.91±206.6×104 (n=6) cell/ml for grinding at frequency of 35 Hz.

Fig. 2. Motion control of a helical microrobot towards a blood clot is achieved under ultrasound guidance. (a) The microrobot is localized and its position is feedback to a closed-loop control system. (b) A permanent magnet-based robotic system provides rotating magnetic field to actuate the microrobot. (c) The microrobot swims under the influence of the rotating magnetic fields towards the clot. (d) Motion control is achieved towards a reference position (black dashed line) that represents the position of the clot. (e) The microrobot grinds the clot and decreases its size with time. The volume ratio represents the current estimated volume of the clot to its original volume before grinding. (f) The count of the blood cell indicates that the helical microrobot tears the fibrin network and allows cells to break free from the clot.

April 2018 IEEE Instrumentation & Measurement Magazine 13

The continuous interaction between the tip of the helical microrobot and the fibrin network of the clot has proven to be efficient in decreasing the size of the clot in vitro and allowing blood cells to break free from the fibrin network of the clot. The hope of translating this concept into in vivo trials one day im-plicates that we need to overcome several challenges. The first challenge is the optimal control of the microrobot in the real cir-culatory system while being exposed to the natural blood flow, which varies according to the type of blood vessel in question,

its cross sectional area, and the blood velocity in a specific re-gion of the body. This challenge necessitates an optimization of the size of the microrobot as well as the respective mechan-ical properties in function of type and location of the clot. As real-time tracking of the helical microrobot in vivo is essential, we are using ultrasound feedback as a non-invasive and safe clinical imaging system. Another challenge is to find a suitable method to assess the mechanical grinding's effect on the blood clot in vivo, especially since the echogenic signal of fresh blood

Fig. 3. Mechanical grinding of blood clots is achieved using two synchronized rotating dipole fields and ultrasound feedback. Position of the microrobot is determined using the ultrasound scans and is used to generate position error based on the position of the clot. The angular positions and velocities of the two rotating dipole fields are used to synchronize their motion to exert magnetic torque on the dipole of the microrobot and mitigate the magnetic force along its transverse direction. The red and green dashed curves and arrows indicate the original ultrasound waves and the reflected waves via the helical microrobot only, respectively.

14 IEEE Instrumentation & Measurement Magazine April 2018

clots is relatively weak. Would it be possible to accurately an-alyze the changes of a blood clot's size during rubbing? Or should we rely on other parameters rather than the size and the number of blood cells past the clot? In addition, as much as using biodegradable material sounds relieving, the fate of the microrobot in vivo, once the clot is removed, needs to be studied. It is also essential to mitigate the interaction of the microrobot with the normal vessel wall in vivo and inhibit the coagulation around the microrobot itself. Therefore, it may be possible to coat the microrobot using a material similar to those used in vascular stents and catheters. Another promising aspect of using microrobots is the possibility to use them as ve-hicles for in situ delivery of thrombolytic drugs in addition to the mechanical grinding; in this case the combination therapy could enhance the local effect on the blood clot while prevent-ing the severe complications that can be related to the systemic administration of thrombolytic drugs. In this context, an in vivo comparative study between the pure mechanical grinding and the combination with in situ chemical lysis requires fur-ther investigation.

AcknowledgmentThe authors would like to thank Dr. A. M. R. Abuelatta from the Diagnostic and Interventional Radiology of the Na-tional Cancer Institute for assistance with the preparation of the in vitro model. They would also like to thank Mr. A. El Sharkawy and Ms. S. Hesham for assistance with the exper-imental work.

References[1] K. Ishiyama, M. Sendoh, and K. I. Arai, “Magnetic micromachines

for medical applications,” J. Magnetism and Magnetic Materials,

vol. 242-245, no. 1, pp. 41-46, Apr. 2002.

[2] B. J. Nelson, I. K. Kaliakatsos, and J. J. Abbott, “Microrobots for

minimally invasive medicine,” Annual Review of Biomedical Eng.,

vol. 12, pp. 55-85, Apr. 2010.

[3] O. Felfoul, M. Mohammadi, S. Taherkhani, D. de Lanauze,

Y. Z. Xu, et al., “Magneto-aerotactic bacteria deliver drug

containing nanoliposomes to tumour hypoxic regions,” Nature

Nanotechnology, Aug. 2016.

[4] A. Servant, F. Qiu, M. Mazza, K. Kostarelos, and B. J. Nelson,

“Controlled in vivo swimming of a swarm of bacteria-like

microrobotic flagella,” Advanced Materials, vol. 27, no. 19, pp.

2981-2988, Apr. 2015.

[5] I. S. M. Khalil, A. F. Tabak, K. Sadek, D. Mahdy, N. Hamdi, and

M. Sitti, “Rubbing against blood clots using helical robots:

modeling and in vitro experimental validation,” IEEE Robotics and

Automation Letters, vol. 2, no. 2, pp. 927-934, Apr. 2017.

[6] S.-I. Ando, “What does a perfect blood pressure meter look like

from a clinician point of view?” IEEE Instrum. Meas. Mag., vol. 17,

no. 3, pp. 15-20, Jun. 2014.

Dalia Mahdy is currently a Research Associate with the Med-ical Micro and Nano Robotics of the German University in Cairo. She received her Bachelor's degree in computer engi-neering and is currently working toward the M.Sc. degree in computer engineering. Her research interests include design and development of minimally-invasive techniques for tar-geted drug delivery using microrobots.

Ramez Reda is currently an Assistant Professor of Neurology at Ain Shams University. He held the position of Director of the Neurology Unit at Ain Shams University until 2012 and serves as a Consultant Neurologist at several major hospitals across Egypt, including Nozha International Hospital and Cleopa-tra Hospital.

Nabila Hamdi is currently an Assistant Professor and the Head of the Molecular Pathology Unit of the German Uni-versity in Cairo. Dr. Hamdi graduated from the Medical University in Sfax, Tunisia and spent the practical years of her medical studies in Munich, Germany. She received her Ph.D. degree in molecular genetics pathology in 2009 from the Ger-man University in Cairo.

Islam S. M. Khalil (islam.shoukry@guc.edu.eg) is currently an Assistant Professor with the German University in Cairo and the Director of the Medical Micro and Nano Robotics Laboratory. He received the Master's and Ph.D. degrees in mechatronics engineering from Sabanci University, Istanbul, Turkey. For two years, he has been a Postdoctoral Research As-sociate with the Robotics and Mechatronics Research Group and MIRA-Institute for Biomedical Technology and Technical Medicine, University of Twente, the Netherlands.