Publications Vol. 1

IQMIntegral Quality Monitor

Publications Vol. 1

Ladies and Gentlemen,

Welcome to Volume 1 of the IQM Publications series.

The first volume of the series is dedicated to the research paper titled: „An integral quality monitoring system for real-time verification of intensity modulated radiation therapy“published in Medical Physics, Vol. 36, No. 12, December 2009 by Dr. Mohammad Islam, et al.

This article represents the original proof of concept for the Integral Quality Monitor System. This proof of con-cept convinced us to get involved with the IQM project by signing a co-development agreement with UHN Toronto and ultimately initiated the founding of iRT Systems.

It is a really enlightening publication and we hope that you will enjoy it as much as we did.

With best regards,

Jürgen Oellig Managing Director

iRT Systems

Mohammad K. Islam, Bernhard D. Norrlinger, Jason R. Smale, Robert K. Heaton,

Duncan Galbraith, Cary Fan, David A. Jaffray

An integral quality monitoring system for real-time verification of intensity modulated radiation therapy

Medical Physics, Vol. 36, No. 12,

December 2009

Dr. Mary Gospodarowicz

Medical Director of Princess Margaret Cancer Centre

Dr. David Jaffray

Head of Radiation Physics

Dr. Fei-Fei Liu

Head of Radiation Medicine Program

16x Linear Accelerators (Elekta and Varian)

2x Elekta Gamma Knife Perfexion®

1 MRgRT unit based on MR on rail and a Varian TrueBeam

Philips Pinnacle Treatment Planning System,

Raysearch Raystation Treatment Planning System

Elekta MOSAIQ (Radiation Oncology Information system)

Radiation Therapy Facilities

Princess Margaret Cancer Centre, Toronto

Radiation Therapy Center

Abstract

Purpose: To develop an independent and on-line beam monitoring system, which can validate the accuracy of segment-by-segment energy fluence delivery for each treatment field. The system is also intended to be utilized for pretreat-ment dosimetric quality assurance of intensity modulated radiation therapy IMRT, on-line image-guided adaptive radiation therapy, and volumetric modulated arc therapy.

Methods: The system, referred to as the integral quality monitor (IQM), utilizes an area integrating energy fluence monitoring sensor (AIMS) positioned between the final beam shaping device i.e. multileaf collimator (MLC) and the patient. The prototype AIMS consists of a novel spatially sensitive large area ionization chamber with a gradient along the direction of the MLC motion. The signal from the AIMS provides a simple output for each beam segment, which is compared in real time to the expected value. The prototype ionization chamber, with a physical area of 22x22 cm2, has been constructed out of aluminum with the electrode separations varying linearly from 2 to 20 mm. A calculation method has been developed to predict AIMS signals based on an elementwise integration technique, which takes into account various pre-determined factors, including the spatial response function of the chamber, MLC characteristics, beam transmission through the secondary jaws, and field size factors. The influence of the ionization chamber on the beam has been evaluated in terms of transmission, surface dose, beam profiles, and depth dose. The sensitivity of the system was tested by introducing small de-viations in leaf positions. A small set of IMRT fields for prostate and head and neck plans was used to evaluate the system. The ionization chamber and the data acquisition software systems were interfaced to two different types of linear accelerators: Elekta Synergy and Varian iX.

Results: For a 10x10 cm2 field, the chamber attenuates the beam intensity by 7% and 5% for 6 and 18 MV beams, respectively, without significantly changing the depth dose, surface dose, and dose profile characteristics. An MLC bank calibration error of 1 mm causes the IQM signal of a 3x3 cm2 aperture to change by 3%. A positioning error in a single 5 mm wide leaf by 3 mm in 3x3 cm2 aperture causes a signal difference of 2%. Initial results for prostate and head and neck IMRT fields show an average agreement between cal-culation and measurement to within 1%, with a maximum deviation for each of the smallest beam segments to within 5%. When the beam segments of a prostate IMRT field were shifted by 3 mm from their original position, along the direction of the MLC motion, the IQM signals varied, on average, by 2.5%.

Conclusions: The prototype IQM system can validate the accuracy of beam delivery in real time by comparing precalculated and measured AIMS signals. The sys-tem is capable of capturing errors in MLC leaf calibration or malfunctions in the positioning of an individual leaf. The AIMS does not significantly alter the beam quality and therefore could be implemented without requiring recom-missioning measurements.

Islam, Norrlinger, et al: An integral quality monitoring system for real-time verification of intensity modulated radiation therapy

6

I. Introduction

In recent years the practice of radiation therapy (RT) has been going through rapid advancements with the potential of improving treatment outcomes. These advances can be attributed to innovations in treatment planning and delivery such as intensity modulated radiation therapy (IMRT)1,2, the availabi-lity of high quality imaging modalities 3,4 for both target volume definition (com-puted tomography, positron emission tomography, and magnetic resonance imaging) and treatment setup verification (electronic portal imaging devices and kilovolt cone beam computed tomography). These innovations, cou-pled with the rapid progress in information technologies, have revolutionized the RT field. However, these developments have led to a more complex and less intuitive planning and treatment delivery process: Multidisciplinary teams now develop plans utilizing multiple imaging modalities to define target volumes in three and four dimensions (including time evolution); computer-as- sisted optimization software devises intensity modulated beams; and the synchronization of dynamic multileaf collimator (MLC) motion, variable dose rate, and gantry angle (the recently introduced volumetric modulated arc therapy (VMAT)) form the actual treatment delivery. Moreover, the entire process of RT utilizes multiple software systems, often provided by different vendors. The increased complexities in treatment planning, delivery, and the overall process have created enormous quality assurance (QA) challenges for modern radiation therapy. Presently, pretreatment IMRT QA tasks are performed by employing conventional dosimetry tools in a fragmentary manner, involving many staff and machine hours. 5-8 The QA is performed only prior to the first of many (30–40) treatment sessions; treatment delivery errors that may be introduced in subsequent sessions could go undetected. Common mistreatments may involve human errors as well as errors in software and hardware due to malfunctions or as a result of system upgrades. No compre-hensive QA solution is available to meet the complexities associated with the modern radiation therapy process and challenges associated with the emer-ging technologies such as on-line image-guided adaptive radiation therapy (IGART) and VMAT. An efficient and independent on-line beam monitoring system could play an important role in meeting the needs of modern and up-coming RT QA practice. The concept of an on-line and independent beam

delivery monitor was first introduced by Paliwal et al.9 for non-computerized linear accelerators (LinAc‘s). The authors demonstrated the usefulness of a large area monitor chamber, mounted on the shielding tray, for checking the daily constancy of treatment delivery. Poppe et al.10 described a novel method of monitoring the constancy of IMRT beam delivery by using a mul-ti-wire large area ionization chamber, mounted on the shielding tray, which func-tioned like an array of line detectors intercepting radiation fluence along the length of each MLC leaf pair. An on-line dosimetry system, COM-PASS®, has been recently released commercially by IBA Dosimetry, Uppsala, Sweden. The system consists of a matrix of 40x40 small ionization chambers, mounted just below the collimator, to capture the beam fluence map. The fluence maps are utilized to calculate the actual dose delivered to the pa-tient’s anatomy using the system’s treatment planning software. The COM-PASS® system can be used for pretreatment plan QA, as well as for on-line verification of treatment delivery.

We have developed a simple alternative energy fluence verification system, which can verify the accuracy and consistency of beam delivery during each treatment session.


7


8

FIG. 1. Schematic diagram showing the process flow for the IQM system. The system consists of a large area energy fluence monitoring sensor (AIMS) and a calculation program, IQM_CALC. The IQM MANAGER (software system) compares the signal from the detector with the precalculated expected value on a segment-by-segment basis. The eavesdrop data tap

allows the IQM system to automatically capture the selected patient’s name and ID during the transfer of patient data from the record-and-verify system to the linac.

record-and-verify system and the linac control system. Theprocess flow of the IQM system is illustrated in Fig. 1.

II.A. Area integrating energy fluence monitoringsensor

The AIMS is composed of a spatially sensitive large areaionization chamber and a dosimetry system incorporating awide dynamic range electrometer. For the prototype system,a one-dimensional spatial variation in the chamber sensitivitywas investigated to capture aperture errors due to errors inMLC leaf positions. By angling the polarizing electrodesrelative to the collector, a position varying electrode separa-tion across the chamber was created, which consequentlyprovided a spatially variable sensitivity �Fig. 2�a��. Thechamber was mounted such that the slope of ion chamberseparation was parallel to the direction of MLC leaf motion.

The prototype ion chamber was designed for both me-chanical and dosimetric stability to monitor an area largeenough to fully encompass typical IMRT beams and to gen-erate a clinically useful spatially sensitive signal through areasonable selection of polarizing electrode slope. Chambercomponents were constructed from aluminum �alloy 6061�with a sensitive area of 22�22 cm2; when mounted at thelinac collimator face, the chamber can monitor a radiationfield that projects to a size of approximately 34�34 cm2 atthe isocenter. The outer polarizing plates were 3.18 mm �1/8in.� thick, while the inner collector plate was 1.59 mm �1/16in.� thick, with the larger outer plate thickness chosen forstructural rigidity. During the initial design, it was assumedthat the spatial sensitivity of the chamber would be propor-

tional to the electrode plate separation �assuming a constantcollection efficiency� and thus proportional to the plate slope.Although a higher plate slope would produce a larger spatialsensitivity, a trade-off was necessary to keep the overallchamber thickness to a practical size. The slope of the plateseparation was chosen to produce a change in response ofapproximately 0.5% mm−1 near the center of the chamberalong the gradient. The total separation between the highvoltage polarizing electrodes varied linearly from 2 to 20

FIG. 1. Schematic diagram showing the process flow for the IQM system. The system consists of a large area energy fluence monitoring sensor �AIMS� anda calculation program, IQM_CALC. The IQM MANAGER �software system� compares the signal from the detector with the precalculated expected value on asegment-by-segment basis. The eavesdrop data tap allows the IQM system to automatically capture the selected patient’s name and ID during the transfer ofpatient data from the record-and-verify system to the linac.

Aluminum

PMMA

InsulatorGuardElectrode

CollectorElectrode

PolarizingElectrode(a)

Integrator B

Integrator AADC A

ADC B

Microprocessor

cMU Counter(optional)

16:1 Mux B

3 MSB toNOR B

16:1 Mux A

3 MSB toNOR AIon

ChamberInput

(b)

FIG. 2. �a� Schematic diagram of the prototype spatially sensitive large area�22�22 cm2� ionization chamber; outer electrode �polarizing electrode�separations vary linearly from 2 to 20 mm. �b� Functional block diagram ofthe dosimetry system, which includes a wide dynamic range dual integratorelectrometer and a microprocessor. The microprocessor controls the integra-tion and readout of signals in synchronization with beam segments.

5422 Islam et al.: Real-time verification of IMRT 5422

Medical Physics, Vol. 36, No. 12, December 2009

II. Materials and Methods

The goal of this work was to develop an independent dosimetry system, which can validate the accuracy of energy fluence in real time with minimal user interaction. This requires a beam monitoring detector that encompasses the entire radiation field, has robust and stable performance, and provides a spatially sensitive signal in terms of a simple output.

To fulfill these criteria, a large area radiation detector was designed to be mounted below the MLC. The detector generates a spatially sensitive integral signal, or more precisely, a spatially sensitive dose-area product, for each beam segment. Based on precalculated or reference measurement values, the system is capable of validating the accuracy of the beam delivery in real time.

The prototype verification system, referred to as the integral quality monitor (IQM), consists of two key components: An area integrating energy fluence monitoring sensor (AIMS) and a calculation module, IQM_CALC. The measured signal from the AIMS for each beam segment is compared on-line to the precalculated value to verify the accuracy of the treatment delivery. The expected signal is calculated by IQM_CALC based on the field information derived directly from the treatment planning system (TPS) and is independent of the treatment delivery systems, i.e., the record-and-verify system and the linac control system. The process flow of the IQM system is illustrated in Fig. 1.







Aluminum

PMMA


CollectorElectrode


Integrator B

Integrator AADC A

ADC B

Microprocessor


16:1 Mux B

3 MSB toNOR B

16:1 Mux A

3 MSB toNOR AIon

ChamberInput

(b)




Treatment process flowIQM process flowPatient/field ID eavesdrop data tap

II.A. Area integrating energy fluence monitoring sensor

The AIMS is composed of a spatially sensitive large area ionization chamber and a dosimetry system incorporating a wide dynamic range electrometer. For the prototype system, a one-dimensional spatial variation in the chamber sensitivity was investigated to capture aperture errors due to errors in MLC leaf positions. By angling the polarizing electrodes relative to the collector, a position varying electrode separation across the chamber was created, which consequently provided a spatially variable sensitivity (Fig. 2a).

The chamber was mounted such that the slope of ion chamber separation was parallel to the direction of MLC leaf motion. The prototype ion chamber was designed for both mechanical and dosimetric stability to monitor an area large enough to fully encompass typical IMRT beams and to generate a clinically useful spatially sensitive signal through a reasonable selection of polarizing electrode slope. Chamber components were constructed from aluminum (alloy 6061) with a sensitive area of 22x22 cm2; when mounted at the linac collimator face, the chamber can monitor a radiation field that projects to a size of approximately 34x34 cm2 at the isocenter. The outer polarizing plates were 3.18 mm (1/8 in.) thick, while the inner collector plate was 1.59 mm (1/16 in.) thick, with the larger outer plate thickness chosen for struc-tural rigidity. During the initial design, it was assumed that the spatial sensitivity

of the chamber would be proportional to the electrode plate separation (assuming a constant collection efficiency) and thus proportional to the plate slope. Although a higher plate slope would produce a larger spatial sensitivity, a trade-off was necessary to keep the overall chamber thickness to a practical size. The slope of the plate separation was chosen to produce a change in response of approximately 0.5% mm-1 near the center of the chamber along the gradient. The total separation between the high voltage polarizing electrodes varied linearly from 2 to 20 mm across the chamber, as shown in Fig. 2a. The sensitive volume of the ion chamber was approximately 530 cm3. Electrode plate separation and orientation were maintained by an insulating frame of machined polymethyl methacrylate (PMMA). A guard electrode at the same potential as the collector was formed from a continuous aluminum channel embedded within the PMMA frame, which supported the collector electrode and intercepted leakage currents between the high voltage and the collector electrodes. The guard and collector electrodes were maintained at ground potential with a polarizing potential of 500 V applied to the high voltage electrodes during normal operation. The overall thickness of the prototype ion chamber was approximately 4 cm. The cham-ber was mounted at the shielding tray of an Elekta linac and at the wedge tray of Varian linac. The distance between the face of the collimator and the bottom of the chamber was 7.5 cm for the Elekta linac, while a distance of 6.5 cm was achieved between the wedge tray slot and the bottom of the chamber for the Varian linac. Standard commercially available electro-meters are typically unable to integrate the charge from a large volume ion chamber without saturating.


9

FIG. 2a. Schematic diagram of the prototype spatially sensitive large area (22x22 cm2) ionization chamber; outer electrode (polarizing electrode) separations vary linearly from 2 to 20 mm.







Aluminum

PMMA


CollectorElectrode


Integrator B

Integrator AADC A

ADC B

Microprocessor


16:1 Mux B

3 MSB toNOR B

16:1 Mux A

3 MSB toNOR AIon

ChamberInput

(b)




10

While it is possible to design an electrometer that can integrate a large charge, readout accuracy and resolution for small charge readings are often compromised. We have developed a unique wide dynamic range electro-meter based dosimetry system that overcomes these problems by using dual electrometers operating in a switching configuration (Fig. 2b). Each electrometer is of the familiar integrating capacitor type. The dual integrator input design allows the electrometer to accept a wide range of currents (5 nA–0.5 mA). While one integrator is collecting charge, the second integrator is held in reset mode. Once the active integrator approaches saturation, the micro-processor switches to the second integrator, allowing the previously active integrator to be read out and reset. The analog to digital converter converts the integrator’s charge into a digital value, which is continuously reported to the microprocessor through the 16:1 multiplexer. The NOR gate provides the signal to the microprocessor that the integrator is fully charged.

The microprocessor also controls the integration and readout of signals in synchronization with beam segments. In addition, the micro-controller main-tains communications with PC-based software, the IQM MANAGER. Using this system, large charge measurements are possible while maintaining the accuracy and resolution of low charge measurements. The maximum mea-surable charge is limited only by the firmware. While the individual integrators can be switched at a maximum frequency of 160 Hz to avoid saturation of signals, the electrometer has been designed to report the integral charge at a maximum rate of five times per second.


FIG. 2b. Functional block diagram of the dosimetry system, which includes a wide dynamic range dual integrator electrometer and a microprocessor. The microprocessor controls the integration and readout of signals in synchronization with beam segments.







Aluminum

PMMA


CollectorElectrode


Integrator B

Integrator AADC A

ADC B

Microprocessor


16:1 Mux B

3 MSB toNOR B

16:1 Mux A

3 MSB toNOR AIon

ChamberInput

(b)




11

II.B. Calculation module „IQM_CALC“

The predicated signal is calculated by the IQM_CALC program based on the field description which includes the jaw settings, beam segment shapes, and monitor units (MUs). The signal is calculated using the geometry and elements illustrated in Fig. 3 through an elementwise integration technique, which incorporates MLC dosimetric parameters and the spatial response of the chamber.

The predicted AIMS signal SCalc is given by

where K is the system constant, MU is the monitor unit, AOF (X,Y) is the area integrated output factor for the jaw (X,Y) settings, and F (x,y) chamber res-ponse function. The function F represents the fluence distribution, including the penumbra associated with MLC leaves and the jaws. The elements of F (x,y) in the open field region of step-and-shoot field aperture are set to unity except across the aperture boundaries. For the prototype system, the value of F at the boundary was assumed to be 0.5 and a simple linear penumbra was assumed over a 5 mm distance on either side of the boundary. The limits of the integral A1 and A refer to the effective regions of the aperture defined by the MLC and regions defined by the jaw, as shown in Fig. 3. The third term in Eq. 1 accounts for the detector signal component due to leakage through the jaws and/or jaws and MLC to the ionization chamber (area R). The term T(x,y) defines the corresponding transmission factors through the jaws and MLC leaves. For simplicity, an average transmission factor for the jaws (TJaw) as well as for the MLC leaf (TMLC) was used in the calculation. The spatial response function of the chamber o(x,y) defines the relative chamber response for an elementary beamlet at position (x,y) with the chamber mounted on the collimator assembly.


FIG. 3. Illustration of the IQM calculation geometry and parameters (example shown is for a Varian linear accelerator configuration). The figure on the left shows the mounting geometry of IQM chamber. The figure on the right shows the beam’s eye view to the chamber and various parameters included in the calculation: Irradiated aperture defined by the MLC is shown in gray shades, rectangular field defined by the jaws outlined by the thick lines.mm across the chamber, as shown in Fig. 2�a�. The sensitive

volume of the ion chamber was approximately 530 cm3.Electrode plate separation and orientation were main-

tained by an insulating frame of machined polymethyl meth-acrylate �PMMA�. A guard electrode at the same potential asthe collector was formed from a continuous aluminum chan-nel embedded within the PMMA frame, which supported thecollector electrode and intercepted leakage currents betweenthe high voltage and the collector electrodes. The guard andcollector electrodes were maintained at ground potential witha polarizing potential of 500 V applied to the high voltageelectrodes during normal operation.

The overall thickness of the prototype ion chamber wasapproximately 4 cm. The chamber was mounted at theshielding tray of an Elekta linac and at the wedge tray ofVarian linac. The distance between the face of the collimatorand the bottom of the chamber was 7.5 cm for the Elektalinac, while a distance of 6.5 cm was achieved between thewedge tray slot and the bottom of the chamber for the Varianlinac.

Standard commercially available electrometers are typi-cally unable to integrate the charge from a large volume ionchamber without saturating. While it is possible to design anelectrometer that can integrate a large charge, readout accu-racy and resolution for small charge readings are often com-promised. We have developed a unique wide dynamic rangeelectrometer based dosimetry system that overcomes theseproblems by using dual electrometers operating in a switch-ing configuration �Fig. 2�b��. Each electrometer is of the fa-miliar integrating capacitor type. The dual integrator inputdesign allows the electrometer to accept a wide range of

currents �5 nA–0.5 mA�. While one integrator is collectingcharge, the second integrator is held in reset mode. Once theactive integrator approaches saturation, the microprocessorswitches to the second integrator, allowing the previouslyactive integrator to be read out and reset. The analog to digi-tal converter converts the integrator’s charge into a digitalvalue, which is continuously reported to the microprocessorthrough the 16:1 multiplexer. The NOR gate provides thesignal to the microprocessor that the integrator is fullycharged. The microprocessor also controls the integrationand readout of signals in synchronization with beam seg-ments. In addition, the microcontroller maintains communi-cations with PC-based software, the IQM MANAGER. Usingthis system, large charge measurements are possible whilemaintaining the accuracy and resolution of low charge mea-surements. The maximum measurable charge is limited onlyby the firmware. While the individual integrators can beswitched at a maximum frequency of 160 Hz to avoid satu-ration of signals, the electrometer has been designed to re-port the integral charge at a maximum rate of five times persecond.

II.B. Calculation module „IQM_CALC…The predicated signal is calculated by the IQM_CALC pro-

gram based on the field description which includes the jawsettings, beam segment shapes, and monitor units �MUs�.The signal is calculated using the geometry and elementsillustrated in Fig. 3 through an elementwise integration tech-nique, which incorporates MLC dosimetric parameters and

PrimaryCollimator

SecondaryCollimator

MultileafCollimator

IQMChamber

Irradiated Aperture

Jaw Attenuated ApertureMLC Leaf Primary Collimator Edge

Secondary Collimator Edge

IQM Ion Chamber

FIG. 3. Illustration of the IQM calculation geometry and parameters �example shown is for a Varian linear accelerator configuration�. The figure on the leftshows the mounting geometry of IQM chamber. The figure on the right shows the beam’s eye view to the chamber and various parameters included in thecalculation: Irradiated aperture defined by the MLC is shown in gray shades, rectangular field defined by the jaws outlined by the thick lines.



the spatial response of the chamber. The predicted AIMSsignal SCalc is given by

SCalc = MU · K · AOF�X,Y� · ��A1

F�x,y��x,y�dxdy

+ �A−A1

TMLC�x,y�F�x,y��x,y�dxdy

+ �R−A

TJaw�x,y�TMLC�x,y�F�x,y��x,y�dxdy� ,

�1�

where K is the system constant, MU is the monitor unit,AOF�X ,Y� is the area integrated output factor for the jaw�X ,Y� settings, and F�x ,y� and ��x ,y� are the correspondingfluence distribution and spatial chamber response function.The function F represents the fluence distribution, includingthe penumbra associated with MLC leaves and the jaws. Theelements of F�x ,y� in the open field region of step-and-shootfield aperture are set to unity except across the apertureboundaries. For the prototype system, the value of F at theboundary was assumed to be 0.5 and a simple linear penum-bra was assumed over a 5 mm distance on either side of theboundary. The limits of the integral A1 and A refer to theeffective regions of the aperture defined by the MLC andregions defined by the jaw, as shown in Fig. 3. The third termin Eq. �1� accounts for the detector signal component due toleakage through the jaws and/or jaws and MLC to the ion-ization chamber �area R�. The term T�x ,y� defines the corre-sponding transmission factors through the jaws and MLCleaves. For simplicity, an average transmission factor for thejaws �TJaw� as well as for the MLC leaf �TMLC� was used inthe calculation. The spatial response function of the chamber��x ,y� defines the relative chamber response for an elemen-tary beamlet at position �x ,y� with the chamber mounted onthe collimator assembly. The response function includes boththe chamber sensitivity and off-axis beam intensity variation.In our initial investigation on a Varian accelerator, the re-sponse function was determined for a 14�24 cm2 field, de-fined by the jaws. The beamlets were formed by MLC leaveswhile keeping the jaws fixed. To eliminate the contributionof transmission through the jaws and MLC leaves in themeasurement, the response of a beamlet �1�1 cm2� wasdetermined by a subtraction technique: The signal of1�1 cm2 segment was subtracted from that of 2�1 cm2

segment, as shown in Fig. 4. This method underestimates thecontribution of the primary fluence in the beamlet by theMLC leakage through the same beamlet; however, for rela-tive response measurement, this effect was assumed to benegligible and no further corrections were applied.

The AOF values were determined semiempirically byequating the measured and calculated values. To illustrate,Eq. �1� can be written for rectangular open fields, defined bythe jaws and backed by the MLC leaves as

SCalc�X,Y� = SCalc� �X,Y� · AOF�X,Y� , �2�

where

SCalc� �X,Y� = MU · K · ��A


+ �R−A

TJaw�x,y�TMLC�x,y�

�F�x,y��x,y�dxdy� . �3�

The limit of integral A corresponds to the area defined by thejaw settings X and Y. By equating measured signals withthose of the calculated values, Eq. �2� can be written as

SMeas�X,Y� = SCalc�X,Y� = SCalc� �X,Y� · AOF�X,Y� . �4�

Consequently, the values of AOF can be determined from thefollowing equation:

AOF�X,Y� = SMeas�X,Y�/SCalc� �X,Y� . �5�

First, relative measurements were performed �normalizedwith respect to a 10�10 cm2 field� for a series of squarefields �up to 14�14 cm2� and for some rectangular fields�from 14�16 to 14�24 cm2� defined by the jaws andbacked by the MLC leaves. Calculations corresponding tothe measured fields were then made using Eq. �3�. Subse-quently, the AOF values were determined using Eq. �5�.

II.C. Performance tests

II.C.1. System characteristics

The short term reproducibility measurements of the AIMSwere made by exposing the chamber to a cobalt-60 tele-therapy beam, with a field size of 10�10 cm2 at a source-to-surface distance of 80 cm. The linearity and dose ratedependence of the AIMS response was measured for variousdose rate settings on a linear accelerator.

II.C.2. Validation of IMRT delivery „measurement vscalculation…

Calculations and corresponding measurements were madefor step-and-shoot IMRT fields for clinical prostate and headand neck plans, generated by Pinnacle V.7.6C �Philips Medi-cal System, Fitchburg, WI�. To compensate for the

.

- =

Fluence

Intensity

FIG. 4. Illustration of the chamber sensitivity measurements by a beamletsubtraction technique. The signal from the beam arrangement on the right issubtracted from that of the left to obtain a relative response due to a beamletof size 1�1 cm2; the top row shows MLC apertures, while the bottom rowshows corresponding fluence profile.






+ �A−A1


+ �R−A


�1�





where



+ �R−A


�F�x,y��x,y�dxdy� . �3�











.

- =

Fluence

Intensity




The response function includes both the chamber sensitivity and off-axis beam intensity variation. In our initial investigation on a Varian accelerator, the response function was determined for a 14x24 cm2 field, defined by the jaws. The beamlets were formed by MLC leaves while keeping the jaws fixed. To eliminate the contribution of transmission through the jaws and MLC leaves in the measurement, the response of a beamlet (1x1 cm2) was determined by a subtraction technique: The signal of 1x1 cm2 segment was subtracted from that of 2x1 cm2 segment, as shown in Fig. 4.

FIG. 4. Illustration of the chamber sensitivity measurements by a beamlet subtraction technique. The signal from the beam arrangement on the right is subtracted from that of the left to obtain a relative response due to a beamlet of size 1x1 cm2; the top row shows MLC apertures, while the bottom row shows corresponding fluence profile.

This method underestimates the contribution of the primary fluence in the beamlet by the MLC leakage through the same beamlet; however, for relative response measurement, this effect was assumed to be negligible and no further corrections were applied.

The AOF values were determined semiempirically by equating the measured and calculated values. To illustrate, Eq. 1 can be written for rectangular open fields, defined by the jaws and backed by the MLC leaves as

where

The limit of integral A corresponds to the area defined by the jaw settings X and Y. By equating measured signals with those of the calculated values, Eq. 2 can be written as

Consequently, the values of AOF can be determined from the following equation:

First, relative measurements were performed (normalized with respect to a 10x10 cm2 field) for a series of square fields (up to 14x14 cm2) and for some rectangular fields (from 14x16 to 14x24 cm2) defined by the jaws and backed by the MLC leaves. Calculations corresponding to the measured fields were then made using Eq. 3. Subsequently, the AOF values were determined using Eq. 5.


12




+ �A−A1


+ �R−A


�1�





where



+ �R−A


�F�x,y��x,y�dxdy� . �3�











.

- =

Fluence

Intensity







+ �A−A1


+ �R−A


�1�





where



+ �R−A


�F�x,y��x,y�dxdy� . �3�











.

- =

Fluence

Intensity







+ �A−A1


+ �R−A


�1�





where



+ �R−A


�F�x,y��x,y�dxdy� . �3�











.

- =

Fluence

Intensity







+ �A−A1


+ �R−A


�1�





where



+ �R−A


�F�x,y��x,y�dxdy� . �3�











.

- =

Fluence

Intensity






The short term reproducibility measurements of the AIMS were made by ex-posing the chamber to a cobalt-60 teletherapy beam, with a field size of 10x10 cm2 at a sourceto-surface distance of 80 cm. The linearity and dose rate dependence of the AIMS response was measured for various dose rate settings on a linear accelerator.

II.C.2. Validation of IMRT delivery (measurement vs calculation)

Calculations and corresponding measurements were made for step-and-shoot IMRT fields for clinical prostate and head and neck plans, generated by Pinnacle V.7.6C (Philips Medical System, Fitchburg, WI). To compensate for the overshoot/undershoot phenomenon11 of dose delivery on Varian units due to the communication delay between the MLC controller and the dosimetry system, the centi-MU (cMU) count available from the console dose rate integrating board was utilized. All the results presented for Varian step-and-shoot IMRT have been normalized with respect to the corresponding cMU counts.

II.C.3. IQM response reproducibility (constancy of IMRT delivery)

Repeated measurements were performed for a number of clinical IMRT fields over several days to assess the reproducibility of IQM response. These mea-surements were also made simultaneously with a Mapcheck detector array (Sun Nuclear Corporation, Melbourne, FL) to verify consistent field delivery. The effect of linac output variation on the IQM chamber signal was corrected using the average Map-check signal. The Mapcheck system has been shown to achieve a long term reproducibility12 within 0.5%.

II.C.4. IQM performance for simulated errors (MLC leaf errors)

To evaluate the sensitivity of the IQM system in terms of capturing common error conditions, comparative measurements were made for a small sample of MLC defined fields. Measurements were performed on a Varian linear accelerator, equipped with a 120-leaf Millennium MLC for a set of standard fields and subsequently for the same fields with deliberate errors in the MLC leaf positions. The chamber gradient was parallel to the direction of the MLC leaf motions.

II.D. Influence of chamber on the beam

The influence of the ionization chamber on the beams was assessed in terms of beam attenuation, change in surface dose, and change in beam quality (percent depth dose and profiles)


13

III. RESULTS

III.A. Parameters for the IQM_CALC

The spatial response function σ(x,y) for the large area chamber for 14x24 cm2 jaw defined field is shown in Fig. 5. The response values were normalized at the center of the chamber. The change in the relative sensitivity was found to be approximately 0.55% mm-1 at the center of the chamber, with some variation along the gradient due to changing electrode separations. Along the symmetry axis, at +/-6.0 cm off-axis distances (corresponding to larger/smaller plate separations), the sensitivity values were 0.36% mm-1 and 1.1% mm-1, respectively. In the non-gradient direction, the sensitivity varied along the off-axis direction very slowly, as expected. The rapid fall off of the sensi-tivity at points close to the edge of the chamber is attributed to the loss of electrons into the chamber walls. The AOF values for 6 MV beams are shown in Fig. 6.

These values were utilized in the IQM_CALC program; for rectangular fields, the AOF values were determined using an equivalent square formula.13

III.B. Performance tests

III.B.1. System characteristics

The combined ion chamber and electrometer system was found to be stable and reproducible, and highly linear in dose response. The readings of the electrometer with a field size of 10x10 cm2 for ten consecutive measurements were found to be highly reproducible, with a standard deviation of 0.08%. The linearity of the chamber response was measured over a wide range of 1–2000 MU. The signal vs MU showed a highly linear correlation with a R2 value of 0.999. The dose rate dependence of the system’s response in the range of 100–600 MU/min was found to be within 0.2%. Signal saturation was avoided due to the dual integrator configuration of the electrometer.


14

FIG. 5. The spatial response function σ (x,y) of the chamber for a jaw setting of 14x24 cm2. The response values are normalized with respect to the value at the beam central axis. The in plane and cross plane dimensions refer to the directions along and perpendicular to the MLC leaf motions, respectively.

FIG. 6. The AOF as a function of square and equivalent square fields for 6 MV beams. The open circle represent data for square fields and the diamonds represent data for equivalent squares of the rectangular fields. The solid line represents a polynomial fit function of the data.

overshoot/undershoot phenomenon11 of dose delivery onVarian units due to the communication delay between theMLC controller and the dosimetry system, the centi-MU�cMU� count available from the console dose rate integratingboard was utilized. All the results presented for Varian step-and-shoot IMRT have been normalized with respect to thecorresponding cMU counts.

II.C.3. IQM response reproducibility „constancy ofIMRT delivery…

Repeated measurements were performed for a number ofclinical IMRT fields over several days to assess the reproduc-ibility of IQM response. These measurements were alsomade simultaneously with a Mapcheck detector array �SunNuclear Corporation, Melbourne, FL� to verify consistentfield delivery. The effect of linac output variation on the IQMchamber signal was corrected using the average Mapchecksignal. The Mapcheck system has been shown to achieve along term reproducibility12 within 0.5%.

II.C.4. IQM performance for simulated errors „MLCleaf errors…

To evaluate the sensitivity of the IQM system in terms ofcapturing common error conditions, comparative measure-ments were made for a small sample of MLC defined fields.Measurements were performed on a Varian linear accelerator,equipped with a 120-leaf Millennium MLC for a set of stan-dard fields and subsequently for the same fields with delib-erate errors in the MLC leaf positions. The chamber gradientwas parallel to the direction of the MLC leaf motions.


The influence of the ionization chamber on the beams wasassessed in terms of beam attenuation, change in surfacedose, and change in beam quality �percent depth dose andprofiles�.

III. RESULTS


The spatial response function ��x ,y� for the large areachamber for 14�24 cm2 jaw defined field is shown in Fig.5. The response values were normalized at the center of thechamber. The change in the relative sensitivity was found tobe approximately 0.55% mm−1 at the center of the chamber,with some variation along the gradient due to changing elec-trode separations. Along the symmetry axis, at �6.0 cm off-axis distances �corresponding to larger/smaller plate separa-tions�, the sensitivity values were 0.36% mm−1 and1.1% mm−1, respectively. In the nongradient direction, thesensitivity varied along the off-axis direction very slowly, asexpected. The rapid fall off of the sensitivity at points closeto the edge of the chamber is attributed to the loss of elec-trons into the chamber walls.

The AOF values for 6 MV beams are shown in Fig. 6.These values were utilized in the IQM_CALC program; for

rectangular fields, the AOF values were determined using anequivalent square formula.13



The combined ion chamber and electrometer system wasfound to be stable and reproducible, and highly linear in doseresponse. The readings of the electrometer with a field sizeof 10�10 cm2 for ten consecutive measurements werefound to be highly reproducible, with a standard deviation of0.08%. The linearity of the chamber response was measuredover a wide range of 1–2000 MU. The signal vs MU showeda highly linear correlation with a R2 value of 0.999. The doserate dependence of the system’s response in the range of100–600 MU/min was found to be within 0.2%. Signal satu-ration was avoided due to the dual integrator configuration ofthe electrometer.

-10-5

05

10

0.6

0.8

1.0

1.2

1.4

1.6

-10

-5

05

10

RelativeSensitivity

Cross-Plane(cm)

In-Plane(cm)

FIG. 5. The spatial response function ��x ,y� of the chamber for a jaw set-ting of 14�24 cm2. The response values are normalized with respect to thevalue at the beam central axis. The in plane and cross plane dimensions referto the directions along and perpendicular to the MLC leaf motions,respectively.

5 10 15 20

0.96

0.98

1.00

1.02

1.04

AOF

Side of square field (cm)

FIG. 6. The AOF as a function of square and equivalent square fields for 6MV beams. The open circle represent data for square fields and the dia-monds represent data for equivalent squares of the rectangular fields. Thesolid line represents a polynomial fit function of the data.



overshoot/undershoot phenomenon11 of dose delivery onVarian units due to the communication delay between theMLC controller and the dosimetry system, the centi-MU�cMU� count available from the console dose rate integratingboard was utilized. All the results presented for Varian step-and-shoot IMRT have been normalized with respect to thecorresponding cMU counts.

II.C.3. IQM response reproducibility „constancy ofIMRT delivery…

Repeated measurements were performed for a number ofclinical IMRT fields over several days to assess the reproduc-ibility of IQM response. These measurements were alsomade simultaneously with a Mapcheck detector array �SunNuclear Corporation, Melbourne, FL� to verify consistentfield delivery. The effect of linac output variation on the IQMchamber signal was corrected using the average Mapchecksignal. The Mapcheck system has been shown to achieve along term reproducibility12 within 0.5%.

II.C.4. IQM performance for simulated errors „MLCleaf errors…

To evaluate the sensitivity of the IQM system in terms ofcapturing common error conditions, comparative measure-ments were made for a small sample of MLC defined fields.Measurements were performed on a Varian linear accelerator,equipped with a 120-leaf Millennium MLC for a set of stan-dard fields and subsequently for the same fields with delib-erate errors in the MLC leaf positions. The chamber gradientwas parallel to the direction of the MLC leaf motions.


The influence of the ionization chamber on the beams wasassessed in terms of beam attenuation, change in surfacedose, and change in beam quality �percent depth dose andprofiles�.

III. RESULTS


The spatial response function ��x ,y� for the large areachamber for 14�24 cm2 jaw defined field is shown in Fig.5. The response values were normalized at the center of thechamber. The change in the relative sensitivity was found tobe approximately 0.55% mm−1 at the center of the chamber,with some variation along the gradient due to changing elec-trode separations. Along the symmetry axis, at �6.0 cm off-axis distances �corresponding to larger/smaller plate separa-tions�, the sensitivity values were 0.36% mm−1 and1.1% mm−1, respectively. In the nongradient direction, thesensitivity varied along the off-axis direction very slowly, asexpected. The rapid fall off of the sensitivity at points closeto the edge of the chamber is attributed to the loss of elec-trons into the chamber walls.

The AOF values for 6 MV beams are shown in Fig. 6.These values were utilized in the IQM_CALC program; for

rectangular fields, the AOF values were determined using anequivalent square formula.13



The combined ion chamber and electrometer system wasfound to be stable and reproducible, and highly linear in doseresponse. The readings of the electrometer with a field sizeof 10�10 cm2 for ten consecutive measurements werefound to be highly reproducible, with a standard deviation of0.08%. The linearity of the chamber response was measuredover a wide range of 1–2000 MU. The signal vs MU showeda highly linear correlation with a R2 value of 0.999. The doserate dependence of the system’s response in the range of100–600 MU/min was found to be within 0.2%. Signal satu-ration was avoided due to the dual integrator configuration ofthe electrometer.

-10-5

05

10

0.6

0.8

1.0

1.2

1.4

1.6

-10

-5

05

10

RelativeSensitivity

Cross-Plane(cm)

In-Plane(cm)

FIG. 5. The spatial response function ��x ,y� of the chamber for a jaw set-ting of 14�24 cm2. The response values are normalized with respect to thevalue at the beam central axis. The in plane and cross plane dimensions referto the directions along and perpendicular to the MLC leaf motions,respectively.

5 10 15 20

0.96

0.98

1.00

1.02

1.04

AOF

Side of square field (cm)

FIG. 6. The AOF as a function of square and equivalent square fields for 6MV beams. The open circle represent data for square fields and the dia-monds represent data for equivalent squares of the rectangular fields. Thesolid line represents a polynomial fit function of the data.



III.B.2. Validation of IMRT delivery „measurementvs calculation…

III.B.2.a. Sample prostate step-and-shoot IMRT field. Theprostate treatment field was designed with a 6 MV beam�Varian iX� using 11 beam segments, with segment MU val-ues ranging from 3 to 25 MU. As shown in Fig. 7, the aver-age agreement between the measurements and calculationsfor individual segments was found to be 0.5%, with a maxi-mum deviation of 3.5%. Larger discrepancies were typicallyobserved for the smaller area beam segments.

III.B.2.b. Sample results for a head and neck IMRT field.This treatment field was designed with a 6 MV beam �VarianiX� using eight beam segments with segment MU valuesranging from 10 to 15 MU. As shown in Fig. 8, the averageagreement between the measurements and calculations forindividual segments is 0.9%, with a maximum deviation of5%.

III.B.3. IQM response reproducibility „constancy ofIMRT delivery…

The reproducibility of two prostate IMRT fields over tendifferent measurements, delivered during a period of twoweeks, using 6 MV beams, on an Elekta linac, is shown inFig. 9. The error bars shown represent 1 standard deviationof the spread of the segment-by-segment data.

III.B.4. IQM response to simulated errors „MLC leaferrors…

The ability of the IQM system to detect probable errorswas investigated through a simulation of MLC positioningerrors on a Varian linac. Symmetrical square apertures ofvarious sizes were defined by the MLC, with the MLC back-ing jaw positioned 0.5 cm behind the leaf ends. The changein the IQM signal for a 1 mm decrease in the field width �dueto leaf bank error� for a 15�15 cm2 aperture was found tobe 0.7%, while the same error in the leaf bank in3�3 cm2 generated a 3% difference in the signal. Changingthe position of a single 0.5 cm wide leaf by 3 mm in the3�3 cm2 aperture generated a 2% difference in the signal.These measurements demonstrate that the IQM system is ca-pable of detecting errors that could arise from MLC leafbank calibration error or a failure of a single leaf. However,detection of this type of error is most sensitive for small-areasegments.

The advent of on-line IGART techniques may lead to in-tentional displacements of MLC apertures from their originalplanned position. This situation was simulated by displacingall the leaf positions by 3 mm for an IMRT prostate field,effectively shifting the entire field by 3 mm. The change insignals for different segments differed by various amountsdepending on the location along the chamber gradient; how-ever, the average change in the signal was found to be 2.5%.This result implies that the IQM system can be utilized forverifying an IGART shift typical in response to a variance inpatient setup.

III.C. Influence of AIMS on beams

The AIMS �ionization chamber� was found to reduce theintensity of a 6 MV beams for a 10�10 cm2 field by 7%.The surface dose for the same field increased from 19.5% to22.5% of the dmax dose for a source-to-surface distance of 90

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

2 4 6 8 10-4-202

1k

2k

3k

28k

30k

ElectrometerCounts

MeasuredCalculated

∆(%)

SegmentNumber(b)(a)

FIG. 7. Sample results for a step-and-shoot prostate IMRT field. �a� Thefigure on the left shows the MLC apertures for each segment and the com-posited fluence map. �b� The results of IQM calculation, measured values,and corresponding percent differences.

2 4 6 8-6-4-202

5k

10k

15k

20k

25k

30k

Electrom

eterCounts Measured

Calculated

∆(%)

SegmentNumber

02 0301

04 0605

07 08

(b)(a)

FIG. 8. Sample results for a step-and-shoot head and neck IMRT field. �a�The figure on the left shows the MLC apertures for each segment and thecomposited fluence map. �b� The results of IQM calculation, measured val-ues, and corresponding percent differences.

2 4 6 8 100

2000

4000

6000

8000

10000

ElectrometerCounts

Segment #

2 4 6 8 10 12

2000

4000

6000

8000

10000

FIG. 9. IQM response reproducibility. The reproducibility of the IQM sig-nals for two prostate IMRT fields, delivered on an Elekta Synergy linearaccelerator. The signals were reproducible to within 0.5% �1��, as shown inthe “top hat” of the bars.



III.B.2. Validation of IMRT delivery „measurementvs calculation…

III.B.2.a. Sample prostate step-and-shoot IMRT field. Theprostate treatment field was designed with a 6 MV beam�Varian iX� using 11 beam segments, with segment MU val-ues ranging from 3 to 25 MU. As shown in Fig. 7, the aver-age agreement between the measurements and calculationsfor individual segments was found to be 0.5%, with a maxi-mum deviation of 3.5%. Larger discrepancies were typicallyobserved for the smaller area beam segments.









X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

2 4 6 8 10-4-202

1k

2k

3k

28k

30k

ElectrometerCounts

MeasuredCalculated

∆(%)

SegmentNumber(b)(a)


2 4 6 8-6-4-202

5k

10k

15k

20k

25k

30k

Electrom

eterCounts Measured

Calculated

∆(%)

SegmentNumber

02 0301

04 0605

07 08

(b)(a)


2 4 6 8 100

2000

4000

6000

8000

10000

ElectrometerCounts

Segment #

2 4 6 8 10 12

2000

4000

6000

8000

10000




III.B.2. Validation of IMRT delivery (measurement vs calculation)

III.B.2.a. Sample prostate step-and-shoot IMRT field.

The prostate treatment field was designed with a 6 MV beam (Varian iX) using 11 beam segments, with segment MU values ranging from 3 to 25 MU. As shown in Fig. 7, the average agreement between the measurements and calculations for individual segments was found to be 0.5%, with a maximum deviation of 3.5%. Larger discrepancies were typically observed for the smal-ler area beam segments.

III.B.2.b. Sample results for a head and neck IMRT field.

This treatment field was designed with a 6 MV beam (Varian iX) using eight beam segments with segment MU values ranging from 10 to 15 MU. As shown in Fig. 8, the average agreement between the measurements and calcula-tions for individual segments is 0.9%, with a maximum deviation of 5%.


15

FIG. 7. Sample results for a step-and-shoot prostate IMRT field. a) The figure on the left shows the MLC apertures for each segment and the composited fluence map. b) The results of IQM calculation, measured values, and corresponding percent differences.

FIG. 8. Sample results for a step-and-shoot head and neck IMRT field. a) The figure on the left shows the MLC apertures for each segment and the composited fluence map. b) The results of IQM calculation, measured values, and corresponding percent differences.

III.B.3. IQM response reproducibility (constancy of IMRT delivery)

The reproducibility of two prostate IMRT fields over ten different measurements, delivered during a period of two weeks, using 6 MV beams, on an Elekta linac, is shown in Fig. 9. The error bars shown represent 1 standard deviation of the spread of the segment-by-segment data.

III.B.4. IQM response to simulated errors (MLC leaf errors)

The ability of the IQM system to detect probable errors was investigated through a simulation of MLC positioning errors on a Varian linac. Symmetrical square apertures of various sizes were defined by the MLC, with the MLC ba-cking jaw positioned 0.5 cm behind the leaf ends. The change in the IQM sig-nal for a 1 mm decrease in the field width (due to leaf bank error) for a 15x15 cm2 aperture was found to be 0.7%, while the same error in the leaf bank in 3x3 cm2 generated a 3% difference in the signal. Changing the position of a single 0.5 cm wide leaf by 3 mm in the 3x3 cm2 aperture generated a 2% dif-ference in the signal. These measurements demonstrate that the IQM system is capable of detecting errors that could arise from MLC leaf bank calibration error or a failure of a single leaf. However, detection of this type of error is most sensitive for small-area segments.

The advent of on-line IGART techniques may lead to intentional displace-ments of MLC apertures from their original planned position. This situation was simulated by displacing all the leaf positions by 3 mm for an IMRT prostate field, effectively shifting the entire field by 3 mm. The change in signals for different segments differed by various amounts depending on the location along the chamber gradient; however, the average change in the signal was found to be 2.5%. This result implies that the IQM system can be utilized for verifying an IGART shift typical in response to a variance in patient setup.


16

FIG. 9. IQM response reproducibility. The reproducibility of the IQM signals for two prostate IMRT fields, delivered on an Elekta Synergy linear accelerator. The signals were reproducible to within 0.5%, as shown in the “top hat” of the bars.III.B.2. Validation of IMRT delivery „measurement

vs calculation…III.B.2.a. Sample prostate step-and-shoot IMRT field. The

prostate treatment field was designed with a 6 MV beam�Varian iX� using 11 beam segments, with segment MU val-ues ranging from 3 to 25 MU. As shown in Fig. 7, the aver-age agreement between the measurements and calculationsfor individual segments was found to be 0.5%, with a maxi-mum deviation of 3.5%. Larger discrepancies were typicallyobserved for the smaller area beam segments.









X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

X1 X2

Y1

Y2

2 4 6 8 10-4-202

1k

2k

3k

28k

30k

ElectrometerCounts

MeasuredCalculated

∆(%)

SegmentNumber(b)(a)


2 4 6 8-6-4-202

5k

10k

15k

20k

25k

30k

Electrom

eterCounts Measured

Calculated

∆(%)

SegmentNumber

02 0301

04 0605

07 08

(b)(a)


2 4 6 8 100

2000

4000

6000

8000

10000

ElectrometerCounts

Segment #

2 4 6 8 10 12

2000

4000

6000

8000

10000




(1σ)


The AIMS (ionization chamber) was found to reduce the intensity of a 6 MV beams for a 10x10 cm2 field by 7%. The surface dose for the same field increased from 19.5% to 22.5% of the dmax dose for a source-to-surface distance of 90 cm. The changes in percent depth dose and profiles were found to be approximately within 1%, as shown in Fig. 10.

IV. DISCUSSION

The prototype IQM system for step-and-shoot IMRT fields yielded promising results. The agreement between the measurements and calculations for indi-vidual field segments were mostly within 1%, except for the smallest segments measured, in which the deviations increased up to 5%. One approach to dealing with small segments would be to combine these with a larger adja-cent segment.

For clinical implementation of the IQM system, a realistic tolerance criteria needs to be developed. The criteria will depend on the IQM_CALC accuracy, tolerance of MLC leaf positioning error, location of the aperture in the slope of the chamber, and also daily machine output variations. The system will have limitations in terms of capturing small leaf position error for an individual leaf, especially in the context of large aperture size. As presented earlier, the spatial sensitivity of the chamber varies along the slope, having lower values on the thinner side. The uniform sensitivity values can be obtained by utilizing the signal from an independent complimentary chamber with opposite slope. While spatial sensitivity could be increased by increasing the chamber gradient and hence the chamber thickness, there is a trade-off between sensitivity and maintaining adequate geometrical clearance during patient treatment.

Although the prototype system was tested for step-and-shoot IMRT, the me-thod could be extended to dynamic IMRT and VMAT. For dynamic IMRT, the field would be partitioned into discrete intervals. The signal in each interval would be calculated by assigning values to F(x,y), which correspond to the appropriate fractional exposure time and blocked transmission time for each element. During delivery of the dynamic IMRT, the start and end of an inter-val can be recognized through the eavesdropping mechanism. In the case of a VMAT plan, the field can be partitioned according to gantry angles.


FIG. 10. Measured data showing very small difference between the profiles (top two plots at 1.5 and 10 cm depth for 30x30 cm2 field) and percent depth dose (bottom two plots for 10x10 and 30x30 cm2 fields).

17cm. The changes in percent depth dose and profiles werefound to be approximately within 1%, as shown in Fig. 10.

IV. DISCUSSION

The prototype IQM system for step-and-shoot IMRTfields yielded promising results. The agreement between themeasurements and calculations for individual field segmentswere mostly within 1%, except for the smallest segmentsmeasured, in which the deviations increased up to 5%. Oneapproach to dealing with small segments would be to com-bine these with a larger adjacent segment.

For clinical implementation of the IQM system, a realistictolerance criteria needs to be developed. The criteria willdepend on the IQM_CALC accuracy, tolerance of MLC leafpositioning error, location of the aperture in the slope of thechamber, and also daily machine output variations. The sys-tem will have limitations in terms of capturing small leafposition error for an individual leaf, especially in the contextof large aperture size. As presented earlier, the spatial sensi-tivity of the chamber varies along the slope, having lowervalues on the thinner side. The uniform sensitivity values canbe obtained by utilizing the signal from an independent com-plimentary chamber with opposite slope. While spatial sen-sitivity could be increased by increasing the chamber gradi-ent and hence the chamber thickness, there is a trade-offbetween sensitivity and maintaining adequate geometricalclearance during patient treatment.

Although the prototype system was tested for step-and-shoot IMRT, the method could be extended to dynamicIMRT and VMAT. For dynamic IMRT, the field would bepartitioned into discrete intervals. The signal in each intervalwould be calculated by assigning values to F�x ,y�, whichcorrespond to the appropriate fractional exposure time andblocked transmission time for each element. During deliveryof the dynamic IMRT, the start and end of an interval can berecognized through the eavesdropping mechanism. In thecase of a VMAT plan, the field can be partitioned accordingto gantry angles.

A number of simplifying approximations were incorpo-rated into the predictive calculation algorithm for the IQMsystem. A more generalized approach and comprehensivelinac characterization are anticipated to improve the agree-ment between measured and predicted signals. In the imple-mentation reported here, the variation in the machine outputwas assumed to be dependent on the jaw settings only andwas parametrized by the AOF, which neglected the effects oftertiary collimators, i.e., MLC. The incorporation of a sec-ondary source model into the IQM calculation is required tocapture such effects as well as changes in fluence profiles forasymmetric fields.14–16 In such a calculation, the AOF wouldtrack residual effects such as backscatter into the monitorchamber. Other improvements may be gained from a morecareful treatment of the MLC transmission and interleaf leak-age.

The AIMS was designed with aluminum plates for opera-tional stability and durability. A trade-off of this design, how-ever, is the loss of the light field, cross hair, and in somelinacs, the surface-to-source distance indicator. While mod-ern radiation therapy practice is increasingly using imageguidance for patient setup and moving away from the use oflight fields, the chamber mount can be designed in such away that the therapist can easily remove and install thechamber in position for access to the light fields. A trade-offfrom the simplicity in output of the system is that it is pos-sible to have the same output signal for multiple combina-tions of aperture size, position, and monitor units, resultingin a false positive output. While this is possible for indi-vidual segments, it would be highly improbable for everysegment of a field to generate a false positive. Another limi-tation is that the IQM system cannot identify the cause oferrors; however, it would provide the important first step byrecognizing a discrepancy. Following the discovery of thediscrepancy, conventional methods and tools can be used toinvestigate the source of error.

Since the transmitted beam through the ion chamber doesnot change the beam characteristics significantly, other thanfor the attenuation, the IQM system can be implementedclinically into the planning process with minimal efforts.Only a modest user interaction will be required to have thesystem monitor the accuracy of beam delivery in routineclinical use.

The system could be configured to operate mostly in anautomated fashion without significant user input. The systemcould be implemented at several levels of treatment monitor-ing. In the most passive form, the IQM system would simplymonitor and record treatment delivery for later review. Amore active implementation would provide a real-time �dur-ing treatment delivery� display of the expected vs measuredsignal to the treatment unit staff, who could terminate thedelivery if predefined tolerances are exceeded. In a fully au-tomated implementation, the IQM system would provide asignal to the linac’s interlock interface to disable the beamwhen an out of tolerance condition is detected.

In addition to meeting the QA challenges of the IMRT, theIQM system can play an essential role for the on-line IGARTand validation of VMAT. In IGART, treatment fields may be

0.00.20.40.60.81.0

0.0

0.2

0.4

0.6

0.2

0.4

0.6

0.8

1.0

0.2

0.4

0.6

0.8

1.0

-30 -20 -10 0 10 20 30-202

-30 -20 -10 0 10 20 30-202

0 5 10 15 20 25 30-202

0 5 10 15 20 25 30-202

RelativeDose

1.5 cm Depth:Chamber outChamber in

10 cm Depth:Chamber outChamber in

RelativeDose

10x10 cm2 field:Chamber outChamber in

30x30 cm2 field:Chamber outChamber in

Off- Axis Distance (cm)

∆Dose(%)

Off-Axis Distance (cm)

Depth (cm)

∆Dose(%)

Depth (cm)

FIG. 10. Measured data showing very small difference between the profiles�top two plots at 1.5 and 10 cm depth for 30�30 cm2 field� and percentdepth dose �bottom two plots for 10�10 and 30�30 cm2 fields�.



A number of simplifying approximations were incorporated into the predictive calculation algorithm for the IQM system. A more generalized approach and comprehensive linac characterization are anticipated to improve the agreement between measured and predicted signals. In the implementation reported here, the variation in the machine output was assumed to be dependent on the jaw settings only and was parametrized by the AOF, which neglected the effects of tertiary collimators, i.e., MLC. The incorpo-ration of a secondary source model into the IQM calculation is required to capture such effects as well as changes in fluence profiles for asymmetric fields.14–16 In such a calculation, the AOF would track residual effects such as backscatter into the monitor chamber. Other improvements may be gained from a more careful treatment of the MLC transmission and interleaf leakage.

The AIMS was designed with aluminum plates for operational stability and durability. A trade-off of this design, however, is the loss of the light field, cross hair, and in some linacs, the surface-to-source distance indicator. While mod-ern radiation therapy practice is increasingly using image guidance for pa-tient setup and moving away from the use of light fields, the chamber mount can be designed in such a way that the therapist can easily remove and install the chamber in position for access to the light fields. A trade-off from the simplicity in output of the system is that it is possible to have the same output signal for multiple combinations of aperture size, position, and monitor units, resulting in a false positive output. While this is possible for individual seg-ments, it would be highly improbable for every segment of a field to generate a false positive. Another limitation is that the IQM system cannot identify the cause of errors; however, it would provide the important first step by recog-niz- ing a discrepancy. Following the discovery of the discrepancy, conventi-onal methods and tools can be used to investigate the source of error.

Since the transmitted beam through the ion chamber does not change the beam characteristics significantly, other than for the attenuation, the IQM system can be implemented clinically into the planning process with minimal efforts.

Only a modest user interaction will be required to have the system monitor

the accuracy of beam delivery in routine clinical use.

The system could be configured to operate mostly in an automated fashion without significant user input. The system could be implemented at several levels of treatment monitoring.

In the most passive form, the IQM system would simply monitor and record treatment delivery for later review. A more active implementation would provide a real-time (during treatment delivery) display of the expected vs measured signal to the treatment unit staff, who could terminate the delivery if predefined tolerances are exceeded. In a fully automated implementation, the IQM system would provide a signal to the linac’s interlock interface to disable the beam when an out of tolerance condition is detected.

In addition to meeting the QA challenges of the IMRT, the IQM system can play an essential role for the on-line IGART and validation of VMAT. In IGART, treatment fields may be changed on-line following imaging of the patient’s positioning; therefore, some form of on-line QA will be necessary.

The AIMS of the IQM system can be modified to have two dimensional spatial sensitivity by including two ion chambers with their gradients orthogonal to each (other along and perpendicular to the direction of MLC motion) enab-ling the validation of an arbitrary shift in the beam aperture or a completely new beam aperture selected from a library of beams. In VMAT mode the beam is delivered with variable dose rates and leaf speeds simultaneous to the gantry motions. In combination with an independent gantry angle sen-sor, such as an inclinometer attached to the gantry structure, the IQM can be utilized for the verification of segment-by-segment VMAT treatment deli-very accuracy.


18

V. CONCLUSION

We have developed a prototype independent beam monitoring system for modern radiation therapy, which provides segment-by-segment verification of the beam delivery in real time. The system consists of a large area ionizati-on chamber with a gradient in the electrode plate separation, mounted be-low the MLC, and a calculation algorithm to predict the signal from the ioni-zation chamber based on the field parameter information received directly from the TPS. The signal from the ionization chamber provides a spatially de-pendent dose-area-product signal for each beam segment. Initial test results evaluating IMRT field segments show an average agreement between the measured and predicted IQM signal to within 1%. However, further investiga-tion is required to evaluate the influence of the accepted variance in clinical delivery, including machine out-put, MLC leaf positions, jaw positions, and beam flatness on the IQM signal. These results would help establish appropri-ate tolerances for effective monitoring of treatment delivery.

ACKNOWLEDGMENTS

This work has been partially supported by the Natural Sciences and Engineer- ing Research Council of Canada (NSERC Grant No. I2IJ340869-06). The authors gratefully acknowledge the contributions of Canming Huang and Graham Wilson for computer programming support and Jurij Ivanoski for help with measurements.

Author to whom correspondence should be addressed. Electronic addresses: [email protected] / [email protected] Fax: +1 (416) 946 6566.


19

Literature

1S. Webb, “Optimizing the planning of intensity-modulated radiotherapy,” Phys. Med. Biol. 39 (12), 2229–2246 (1994).

2A. L. Boyer and C. X. Yu, “Intensity-modulated radiation therapy with dynamic Multileaf collimators,” Semin. Radiat. Oncol. 9 (1), 48–59 (1999).

3L. E. Antonuk, J. Yorkston, W. Huang, H. Sandler, J. H. Siewerdsen, and Y. el-Mohri, “Megavoltage imaging with a large-area, flat-panel, amorphous silicon imager,” Int. J. Radiat. Oncol., Biol., Phys. 36 3, 661–672 (1996).

4D. A. Jaffray and J. H. Siewerdsen, “Cone-beam computed tomography with a flat-panel imager: Initial performance characterization,” Med. Phys. 27 (6), 1311–1323 (2000).

5J. S. Tsai, D. E. Wazer, M. N. Ling, J. K. Wu, M. Fagundes, T. DiPetrillo, B. Kramer, M. Koistinen, and M. J. Engler, “Dosimetric verification of the dynamic intensity-modulated radiation therapy of 92 patients,” Int. J. Radiat. Oncol., Biol., Phys. 40 (5), 1213–1230 (1998).

6L. E. Reinstein, X. H. Wang, C. M. Burman, Z. Chen, R. Mohan, G. Kutcher, S. A. Leibel, and Z. Fuks, “A feasibility study of automated inverse treatment planning for cancer of the prostate,” Int. J. Radiat. Oncol., Biol., Phys. 40 (1), 207–214 (1998).

7C. Burman, C. S. Chui, G. Kutcher, S. Leibel, M. Zelefsky, T. LoSasso, S. Spirou, Q. Wu, J. Yang, J. Stein, R. Mohan, Z. Fuks, and C. C. Ling, “Planning, delivery, and quality assurance of intensity-modulated radiotherapy using dynamic multileaf collimator: A strategy for large-scale implementation for the treat-ment of carcinoma of the prostate,” Int. J. Radiat. Oncol., Biol., Phys. 39 (4), 863–873 (1997).

8I. J. Yeo, A. Beiki, Y. B. Cho, M. Heydarian, T. Zhang, and M. Islam, “EDR2 film dosimetry for IMRT verificati-on using low-energy photon filters,” Med. Phys. 31 (7), 1960–1963 (2004).

9B. R. Paliwal, M. Zaini, T. McNutt, E. J. Fairbanks, and R. Kitchen, “A consistency monitor for radiation therapy treatments,” Med. Phys. 23 (10), 1805–1807 (1996).

10B. Poppe, C. Thieke, D. Beyer, R. Kollhoff, A. Djouguela, A. Ruhmann, K. C. Willborn, and D. Harder, “DAVID–A translucent multi-wire transmission ionization chamber for in vivo verification of IMRT and conformal irradiation techniques,” Phys. Med. Biol. 51 (5), 1237–1248 (2006)

11G. Ezzell and S. Chungbin, “The overshoot phenomenon in step-andshoot IMRT delivery,” J. Appl. Clin. Med. Phys. 2 (3), 138–147 (2001).

12P. A. Jursinic and B. E. Nelms, “A 2-D diode array and analysis software for verification of intensity modulated radiation therapy delivery,” Med. Phys. 30 (5), 870–879 (2003).

13S. Kim, T. C. Zhu, and J. R. Palta, “An equivalent square formula for determining head scatter factors for rectangular field,” Med. Phys. 24 (11), 1770–1774 (1997).

14A. Ahnesjo, “Analytic modeling of flattening filters in photon therapy beams,” Med. Phys. 21 (8), 1227–1235 (1994).

15J. Olofsson, D. Goerg, and M. Karlsson, “A widely tested model for head scatter influence on photon beam output,” Radiother. Oncol. 67, 225–238 (2003).

16S. Kim, J. R. Palta, and T. C. Zhu, “A generalized solution for the in-air output factors for irregular fields,” Med. Phys. 25 (9), 1692–1701 (1998).

Contact us at [email protected]

Call us at +49 261 915450

More information is always available at www.i-rt.de

iRT Systems GmbHSchloßstraße 1· 56068 Koblenz· Germany

See IQM in action

The contents of this publication are copyright by iRT Systems GmbH and cannot

be altered or reproduced without the written consent of iRT Systems GmbH

Not for sale or distribution in every countryIQM® is a registered trademark of iRT SystemsIQM® is patented technology

Publications Vol. 1

Documents

Transcript of Publications Vol. 1