Effect of slit scan imaging techniques on image quality on radiotherapy electronic portal imaging

Transcription

1 The University of Toledo The University of Toledo Digital Repository Theses and Dissertations 2008 Effect of slit scan imaging techniques on image quality on radiotherapy electronic portal imaging Dean R. Walton Medical University of Ohio Follow this and additional works at: Recommended Citation Walton, Dean R., "Effect of slit scan imaging techniques on image quality on radiotherapy electronic portal imaging" (2008). Theses and Dissertations This Thesis is brought to you for free and open access by The University of Toledo Digital Repository. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of The University of Toledo Digital Repository. For more information, please see the repository's About page.

2 Health Science Campus FINAL APPROVAL OF THESIS Master of Science in Biomedical Sciences Effect of Slit Scan Imaging Techniques on Image Quality in Radiotherapy Electronic Portal Imaging Submitted by: Dean R. Walton In partial fulfillment of the requirements for the degree of Master of Science in Biomedical Sciences Examination Committee Signature/Date Major Advisor: Michael Dennis, Ph.D. Academic E. Ishmael Parsai, Ph.D. Advisory Committee: John Feldmeier, D. O. Senior Associate Dean College of Graduate Studies Michael S. Bisesi, Ph.D. Date of Defense: May 28, 2008

3 EFFECT OF SLIT SCAN IMAGING TECHNIQUES ON IMAGE QUALITY IN RADIOTHERAPY ELECTRONIC PORTAL IMAGING DEAN R. WALTON UNIVERSITY OF TOLEDO 2008

4 DEDICATION This thesis is dedicated to my wife, Ann, and my sons Josh and Brian. Their continual love, support, and encouragement helped me to continue when I needed it the most. ii

5 ACKNOWLEDGEMENTS I would like to express my heart felt thanks to my major advisor, Dr. Michael J. Dennis for his patience and encouragement throughout my graduate education. I also wish to acknowledge Dr. Ishmael Parsai and Dr. John Feldmeier, the other members of my research committee, for the time invested and guidance they provided. iii

6 TABLE OF CONTENTS DEDICATION... II ACKNOWLEDGEMENTS... III TABLE OF CONTENTS...IV INTRODUCTION... 1 LITERATURE... 8 MATERIALS METHODS RESULTS DISCUSSION CONCLUSION SUMMARY BIBLIOGRAPHY APPENDIX A APPENDIX B APPENDIX C iv

7 INTRODUCTION In the treatment of cancer with teletherapy radiation systems, portal images are routinely obtained to verify appropriate treatment fields. The purpose of this work is to investigate the potential benefit and application of high energy slit scan digital radiography for portal imaging. Background Improvements in the treatment of cancer by external beam radiation are the result of advances in technology that provide three-dimensional (3D) anatomical information for planning radiotherapy treatments. Treatment techniques such as 3D conformal radiotherapy and intensity-modulated radiotherapy can focus the radiation dose on the tumor while minimizing the dose to the surrounding health tissue. The benefits of these techniques can only be realized if the actual treated dose distribution is delivered accurately to the tumor. Inaccuracies between the planned treatment and the actual delivered treatment can lead to normal tissue complications or recurrence of disease (White, 1982). The use of the radiotherapy beam to create a portal image of the area being treated has long been utilized as a method to verifying correct delivery of the radiation dose. The portal image records the exact irradiated area and can identify errors in the patient set-up or of the shape of the collimated treatment beam (Marks, 1974; Rabinowitz, 1985). Historically, portal images have been acquired with film to detect setup and treatment errors (Hulshof, 1989; Rosenthal, 1992). Using film as a portal imaging technique is labor intensive that requires extra time for film positioning, 1

8 processing and review. Due to the time consuming process of using film, the ability to quickly evaluate proper treatment setup prior to treatment is a limitation. Furthermore, the ability to visually evaluate positioning errors with film is subjective and difficult to perform (Taborsky, 1981). Electronic portal imaging devices (EPID) have become a portal imaging technique used to replace film due to EPID s ability to improve treatment workflow by producing real-time digital images (Herman, 2001; Antonuk, 2002). With the ability to obtain real-time digital images it is possible to verify the correct radiation field placement prior or during field delivery efficiently. The portal image is captured digitally which allows image analysis tools to be used that improve clinical evaluation of correct treatment positioning. Megavoltage Imaging Technologies There are three different techniques used to generate useful portal images with acceptable image quality that have been commercially available. The earlier used techniques for EPIDs were a matrix ion chamber (Meertens, 1985; van Herk, 1988) and the camera-based EPIDs (Baily, 1980; Munro, 1990). These two technologies represent the first generation of portal imaging widely used. The third technology, currently being used, utilizes an active matrix flat panel imaging system (Antonuk, 1990; Zhao, 1995; Wang, 1996). The design of the camera-based system used a metal plate and a phosphorous screen which generated light that reflected off a mirror positioned at a 45º angle. The reflected light was detected by a camera and the light signal 2

9 was converted to a digital image. The camera-based EPIDs were limited by the light collection efficiency of the device. The design of the matrix ion chamber involved two planes of electrodes separated by a 0.8mm gap. The gap was filled with a fluid (2,2,4- trimethylpentane) which is ionized when exposed to radiation. The electrodes consisted of a 256x256 matrix. The active matrix flat panel imager (AMFPI) consists of a 2D digital pixel array, an overlay which converts the x-ray to a measurable signal, and an acquisition system that has control over the function of the array. The AMFPI array is designed with a thin piece of glass that is plasma etched to form thousands of individual circuit elements. Every pixel in the active matrix array is composed of a circuit which contains a thin film switch connected to some type of a capacitor. While the imager is in operation the switches are left open so that the charge generated from the radiation is integrated. The image is created by closing the switches and then the charge stored by the capacitors is recorded. After the pixel switches are sampled they are reinitialized to collect another image. Currently, the majority of AMFPI designs use thin film transistors made with hydrogenated amorphous silicon. To convert the radiation into a charge stored by each pixel, there are two ways that the imaging signal is generated. The first process is an indirect detection that uses a scintillating screen, to convert the x-rays into light, which is positioned directly over the active matrix array. The light generates a charge captured by photosensors which correspond to the capacitors in each pixel. The second process is direct detection where a 3

10 photoconductor directly senses the radiation. Presently, EPID systems that use AMFPI utilize the indirect detection method. Currently available EPIDs have a 1mm copper plate in front of the active detector layer which acts as buildup and absorbs scatter radiation. The active matrix is equivalent to 8mm of water which is less than the dose maximum for a 6MV photon beam. The effect of buildup was investigated by McCurdy et al. (2001) using Monte Carlo simulation. This investigation showed that additional buildup was needed to increase EPID sensitivity for high energy photons. The effect of the copper buildup results in the dose maximum being reached that increases the detector sensitivity. The copper buildup also filters out some of the low energy photons resulting from scatter radiation. Limitations of Portal Imaging Megavoltage x-ray portal imaging is the standard treatment verification method used even though the image quality is poor when compared to kilovoltage x-ray imaging. The quality of a megavoltage energy image is the result of a number of reasons, such as; with decreased tissue contrast due to the high-energy x-rays being more penetrating and dominated by Compton scatter interactions (Herman, 2001). Spatial resolution is limited by the larger sourcesize associated with radiotherapy along with the typical magnification used, and limitations in the detector resolution. The effect of scatter radiation reduces tissue contrast and contrast-tonoise ratio (CNR) for the structure being detected. It has long been recognized in diagnostic imaging, that scatter radiation degrades image quality (Niklason, 4

11 1981; Barnes, 1991). These methods include changes in beam energy, minimizing patient thickness, reduction of radiation field size, use of an air gap between the patient and detector, and the use of anti-scatter radiographic grids. Changes in beam energy, patient thickness and the use of radiographic grids are impractical for clinical portal imaging at megavoltage energies. The use of air gap technique is commonly used and successfully employed in portal imaging to reduce scatter. One needs to image a larger field-of-view, however the data may be acquired as a successive series of smaller fields. This is the basic principle of scatter reduction that is employed in slit scan imaging that is being evaluated in this work. General Objectives of the Study The general objective of this project is to evaluate the scatter reduction and image quality of a radiotherapy slit scan technique utilizing an electronic portal imaging device. Specific Objectives of the Study This project is divided into the following eight specific objectives. Specific Objective #1 Determine the scatter and primary components of radiation measurements for various geometries with an ionization chamber. These measurements will cover a range of field sizes for a variety of source to detector distances. A scattering material will be introduced between the source and detector to generate scatter radiation. 5

12 Specific Objective #2 Determine the scatter and primary components of radiation measurements for various geometries with an EPID. These measurements will duplicate the experimental settings used in the objective 1. Specific Objective #3 Compare the EPID and ionization chamber data to evaluate differences in response to scatter and primary components of the radiation beam. The comparison will make use of the data collected in objectives 1 and 2. Specific Objective #4 Evaluate the effect of an air gap on the scatter-to-primary ratio for various imaging geometries. A number of air gap distances will be used to investigate how scatter reacts to this change. This will be done to determine in what situations the slit scan technique will be most effective. Specific Objective #5 Evaluate the effect of a scattering material on scatter-to-primary ratio for various field sizes. The goal of this objective is to characterize the additional scatter generated from varying material thickness and establish which clinical settings will the slit scan technique improve image quality. Specific Objective #6 Compare various slit scan field sizes to evaluate the scatter reduction achieved for image improvement. This evaluation will be used to determine which field size would be optimal for this project. 6

13 Specific Objective #7 Measure the contrast to noise ratio and object detectability for various geometries using the slit scan technique. The slit scan results will be compared to open field measurements. The Las Vegas phantom will be used as the evaluation tool to determine these values. Specific Objective #8 Demonstrate the ability to generate a slit scan megavoltage image using a series of slit images. 7

14 LITERATURE Characterization of Scatter at the Image Plane Reviewed The effect of scatter radiation on image quality is been known to decrease image contrast. The contribution of scatter to the megavoltage image has been investigated not for image quality purposes but for the development of portal dosimetry (Hansen, 1996; McNutt, 1996; Boellaard, 1997; Pasma, 1998) where the dose delivered to a patient is computed from the signal detected in a portal image. To calculate a backprojection of the dose delivered the primary component must be accurately known. Therefore, the scatter must be removed from the portal image to obtain the primary component. Much of the work done to characterize scatter, detected by the image receptor, has been based on Monte Carlo simulations and convolution algorithms. Swindell et al. (1991) used a Monte Carlo simulation to investigate scatter radiation with restricted geometries. These calculations did not consider the effect of field size, patient thickness, and a polyenergetic spectra. He characterized the scatter radiation in terms of scatter-to-primary ratio (SPR) for a 2MV monoenergetic photon beam. The SPR is the ratio of the scattered radiation signal to the primary radiation signal at the image plane. Work done by Jaffray et al. (1994) modeled scatter with Monte Carlo calculations and experimental measurements for 6MV photons. Jaffray examined how scatter and primary radiation would be affected by various scattering geometries and polyenergetic spectra. The amount of scatter generated by different geometries was expressed by the scatter fraction at the 8

15 image detector. The scatter fraction (SF) is defined as the scatter radiation component divided by the total signal composed of the scatter and primary radiation. At a later date, Swindell et al. (1996) further defined the scatter radiation for a 6 MV photon beam. They developed an analytical model to calculate the SPR and compared the results to experimental measurements as well as Monte Carlo calculations. McNutt et al. (1996) used the convolution/superposition method to determine the dose delivered in the portal image plane. This method was used to predict the scatter and primary components detected by the portal image. In this work, it was proposed to remove the scatter to enhance the megavoltage portal image. Hansen et al. (1997) used a forward convolution method to remove scatter signal in images obtained by EPIDs. Scatter Radiation Reduction Techniques Reviewed The quality of images generated from megavoltage x-rays are of poor visual quality compared to kilovoltage x-rays. There are numerous factors that contribute to the degraded quality of portal images other than the familiar principle that subject contrast decreases as the energy of the x-ray beam increases. These specific factors are patient thickness, source size, and the function and position of the image receptor. A number of techniques have been developed for diagnostic imaging to reduce scatter radiation during image acquisition. The air gap technique involves increasing the distance between the patient and the image detector. By doing so the amount of scatter radiation reaching the detector is reduced (Jaffary, 1994; Swindell, 1996). The 9

16 disadvantage of the air gap technique is reduced field-of-view and possible increased focal spot blurring. This technique is currently use for megavoltage imaging as well. Another diagnostic technique for scatter reduction is the use of anti-scatter grids. The benefit of the anti-scatter grid is that it is simple and effective in eliminating scatter radiation (Sorenson, 1985; Neitzel, 1992; Shaw, 1994). The major short comings of the anti-scatter grid are that it attenuates the primary radiation as well which increases the relative noise level in the image. This technique is not practical for megavoltage imaging given the weight and size that would be required (Swindell, 1996). An alternative to anti-scatter grids is based on a diagnostic slit scan imaging method. This technique uses a narrow collimated beam which is dynamically scanned across the patient (Jaffe, 1975:Sorenson, 1980: Plenkovich, 1986: Barnes, 1994: Samei, 2004; Samei, 2005; Lui, 2007). The strength of this method is that the primary radiation is not attenuated. This method can be applied to megavoltage imaging by the use of asymmetric jaws. To date, there have been no reported investigations into the use of this technique for megavoltage energies and is the main subject of this work. 10

17 MATERIALS The following is the list of the materials utilized for this project: 1. A Varian 2100EX linear accelerator equipped with a Varian as500 electronic portal imaging device mounted on a retractable robotic arm (Varian Medical Systems, Palo Alto, CA). 2. A cm 2 IC-10 ionization chamber with a 4 mm diameter spherical volume (Wellhoffer Dosimetrie, Schwarenbruck, Germany) cm poly-methyl-methacrylate (PMMA) build up cap 4. Inovision Model electrometer 5. PMMA 41x41cm 2 scattering material at varied thicknesses of 0.15, 0.297, and 1.20cm to provide scatter material thicknesses of 10, 20 and 30cm. 6. Wooden frame for holding PMMA scattering material in the upright position. 7. Las Vegas phantom used for contrast and object detectability (Varian Medical Systems, Palo Alto, CA). 8. Portal Vision software (version 6.1) for image acquisition and analysis (Varian Medical Systems, Palo Alto, CA). 9. Image J software (version 1.39) for image analysis and processing (National Institute of Health, Bethesda, MD). 11

18 METHODS Specific Objective #1 - Ionization Chamber Radiation Measurements Experimental setup for specific objective #1 Figure 1. The geometric setup of the linear accelerator, scattering phantom material and ionization chamber for specific objective 1. The scatter and primary components of radiation for various geometries with an ion chamber were measured using the Varian 21EX linear accelerator operating at 6 MV photon beam energy. The linear accelerator geometry was set at a gantry and collimator angle of 270º and table angle of 90º (see Figure 1). This machine set up was used to reduce excess scatter from the floor and the surroundings. The distance to the bottom of the phantom was maintained at 100cm SSD. Measurements were made for a range of field sizes (1x1, 1.5x1.5, 2x2, 3x3, 4x4, 6x6, 8x8, 10x10, 15x15, 20x20, 30x30cm²). The source-to- 12

19 detector distance (SDD) was set to 110, 120, 130 and 140cm. All the measurements were made using an IC-10 ionization chamber connected to an Inovision model electrometer. The ionization chamber was fitted with a 1.5cm radial PMMA buildup cap for the 6 MV photon beam used. Use of the buildup cap minimized any signal due to electrons originating from the scatter material. The SDD was measured to the horizontal plane containing the center of the ion chamber. For each measurement the ion chamber was positioned in the center of the radiation field. The ionization chamber position was verified visually with the light field crosshairs for each combination of SDD and scatter material thickness. Three measurements were made for each field size and set up geometry. The ionization chamber was exposed to 100 MU at a dose rate of 100 MU/min for each measurement. The measured data was averaged for each set-up and normalized to a 10x10 cm 2 field. Three sets of data were acquired for each SDD setting with 0, 10, and 20 cm thicknesses of PMMA scattering material. The data set measured with no scattering material was performed to determine the inherent beam fluence of the linear accelerator to correct for changes in output with field size. Correcting for the characteristic beam fluence changes with field size is due to extra-focal radiation exposing the detector and changes in the amount of electrons backscattering to the beam monitoring chamber (Luxton, 1988). Correcting the 10 cm and 20 cm scatter material data sets results in measurements 13

20 independent of accelerator output which reflects only the effect of the air gap and scatter material (Jaffray, 1994; Swindell, 1996). The ionization chamber measurements represent a combination of the primary and scatter components of the radiation beam. In order to separate the scatter from the primary component the measured data was extrapolated to a Determination of Zero Field Size Value 1.08 Signal Nomalized to 10x10cm Field Size y = x R 2 = Field Size (cm on a side) Measured Data Linear Regression Figure 2. Change in output with field size normalized to a 10x10cm 2 field size for no scatter material present at a 110 cm SDD. The linear regression performed on field sizes 2x2, 3x3, 4x4, and 6x6cm 2. zero field size. The apparent zero field size value represents only the primary component where the scatter component reduces to zero (Johns, 1983). To calculate the zero field size value, a linear regression was done with the values of field sizes 2x2, 3x3, 4x4, and 6x6cm 2 (see Figure 2). The y-intercept of the 14

21 linear regression corresponds to the zero field size value or the primary component signal. Once the zero field size value was determined the output factors for all the field sizes were normalized to a zero field size. The scatter component for each field size can be determined by subtracting the primary component from the total signal. Since the output factors are normalized to the primary component, to determine the scatter component requires subtracting one from the given field size output factor. Specific Objective #2 - EPID Radiation Measurements Experimental setup for specific objective #2 Figure 3. The geometric setup of the linear accelerator, scattering phantom material and ionization chamber for specific objective #2. The scatter and primary components of radiation for various geometries with the Varian as500 EPID were measured using the Varian 21EX linear 15

22 accelerator operating at 6 MV photon beam energy. The linear accelerator geometry was set at a gantry and collimator angle of 270º and table angle of 90º (see Figure 3). This machine set up was used to reduce excess scatter from the floor and the surroundings. The distance to the bottom of the phantom was maintained at 100cm SSD. Images were captured for a range of field sizes (1x1, 1.5x1.5, 2x2, 3x3, 4x4, 6x6, 8x8, 10x10, 15x15, 20x20cm²). Also images were taken of field sizes 15x1, 15x1.5 and 15x2cm 2 to evaluate the field widths to be used for the slit scan technique. The SDD was measured to the phosphor layer inside the EPID. The SDD was set to 110, 120, 130 and 140cm to evaluate the air gap effect on the scatter rejection. For each image, the EPID was positioned in the center of the radiation field. The EPID position was verified visually with the light field crosshairs for each combination of air gap and phantom thickness. Four sets of data were acquired for each SDD setting with 0, 10, 20 and 30 cm thicknesses of PMMA scattering material. The data set measured with no scattering material was performed to determine the inherent beam fluence of the linear accelerator to correct for changes in output with field size as mentioned earlier. Before the image set was taken a dark field calibration image was taken to correct for additive electronic noise. Also, a flood field calibration field was taken to correct the gain for each individual pixel. Measurements were acquired using the Varian PortalVision software with each measurement represented by a 9x9 pixel region of interest in the center of the radiation field. The average, standard deviation, and noise for the region of interest were recorded for each image. The 16

23 EPID was exposed to 100 MU at a dose rate of 100 MU/min for each image. The data for each set up was normalized to a 10x10 field. The EPID images represent a combination of the primary and scatter components of the radiation beam. In order to separate the scatter from the primary component the measured data was extrapolated to a zero field size. As mentioned before, the apparent zero field size value represents only the primary component where the scatter component reduces to zero (Johns, 1983). To calculate the zero field size value, a linear regression was done with the values of field sizes 2x2, 3x3, 4x4, and 6x6cm 2 (see Figure 2). The y-intercept of the linear regression corresponds to the zero filed size value or the primary component signal. As with the ionization chamber measurements, the scatter component for each field size can be determined by subtracting the total signal from the primary value. Specific Objective #3 - Comparison of Detector Responses Using the previously measured EPID and ionization chamber data, a comparison of each detector response to scatter and primary radiation was performed. The geometric setup defined in Figures 1 & 3 for the measurements taken were used. The field size dependent output factors for each geometric setup was normalized to the calculated zero field size output factor. To perform a comparison, a ratio of each detector output factor was calculated (Greer, 2003). The following equation was used to calculate the output detector ratio (ODR) where ODR = OF EPID OF IC 17

24 with OF EPID is the EPID output factor and OF IC is the ionization chamber output factor. Specific Objective #4 - Evaluation of Air Gap Effect The effect of an air gap between a scattering material and a detector was evaluated. To evaluate the effect of an air gap, measurements taken previously were used to calculate the scatter fraction (SF). Measurements taken with no scattering material in the beam were used to correct data so the result is independent of variation in collimator scatter. To correct the scatter component of the scattering material measurement was calculated as follows S t,x = OF t,x - OF t,0 OF 0,x OF 0,0 where OF t,x represents output factor for a scattering material thickness of t and field size x, OF 0,x is the output factor for no scattering material and a field size of x. OF t,0 represents output factor for a scattering material thickness of t and the zero field size. OF 0,0 is the output factor for no scattering material and a zero field size. Once the corrected scatter (S) and primary (P) components are known, the SF can be calculated where (Motz, 1978) SF = S S + P This value represents the amount of scatter generated from the scattering material in the radiation beam, as well as, scatter originating from the collimator. Measurements were taken with air gaps of 10cm, 20cm, 30cm, and 40cm. One set of measurements were taken with an ionization chamber so the results could 18

25 be compared to previously published data (Jaffray, 1994) to validate the experimental technique. Another set of measurements were taken with an EPID to assess the scatter radiation effected on the size of an air gap. For this set of measurements the scatter-to-primary ratio (SPR) was used to evaluate the effect of the air gap on scatter. The SPR can be calculated by SPR = S P Since the output factors are normalized to the primary component of the radiation an alternative method of determining the SPR is given by SPR = OF t,x - 1 where OF t,x represents a output factor for a scattering material thickness of t and field size x. Specific Objective #5 - Determine Scattering Material Effect The effect of a scattering material between the radiation source and a detector was evaluated. The thicknesses used for the scattering material was 10cm, 20cm, 30cm and a set with no scattering material. To evaluate the effect of scattering material, measurements taken previously with the EPID in objective 2 were used to calculate the SPR for a range of field sizes. These values were used to determine how varying the thickness of the scattering material effected scatter radiation detected by the EPID. Specific Objective #6 - Slit Scan Field Size Evaluation The slit scan field sizes were 15x1cm², 15x1.5 cm², and15x2 cm². Profiles along the narrow portion of the slit scan images were created on the Varian 19

26 PortalVision software and the numerical data loaded into an Excel spreadsheet. The profile data was review to determine the best field width to be used in further measurements to be taken in Objectives 7 and 8. Profiles were compared to determine how many pixels formed a plateau or flat portion to allow an adequate region of interest size for measurements and generate a slit scan image. The SPR was calculated for these slit scan field sizes from the previously measured data for the EPID. To evaluate the decrease in scatter radiation, by using the slit scan technique, the SPR for a given slit scan field size (SPR SS ) was subtracted from the 15x15cm 2 field size SPR (SPR 15 ). SPR 15-SS = SPR 15 - SPR SS In doing so, the increase in the scatter radiation for the 15x15cm 2 field relative to the slit scan field could be analyzed to evaluate the effectiveness of the slit scan technique in improving image quality. Specific Objective #7 Slit Scan Image Quality Measurements The contrast-to-noise ratio and object detectability using the Las Vegas phantom was measured (Low, 1996) for various geometries with the Varian as500 EPID using the Varian 21EX linear accelerator operating at 6 MV photon beam energy. The linear accelerator geometry was set at a gantry and collimator angle of 270º and table angle of 90º (see Figure 4). This machine set up was used to reduce excess scatter from the floor and the surroundings. The distance to the entrance surface of the scattering material was maintained at 80cm SSD. The Las Vegas phantom was placed on top of a Styrofoam block against the scattering material side closest to the radiation source. The table was adjusted to center of the Las Vegas phantom to the light field crosshair. The SDD was 20

27 measured to the phosphor layer inside the EPID. The SDD was set to 120, 130, 140 and 150cm. For each image the EPID was positioned in the center of the radiation field. The EPID position was verified visually with the light field crosshairs for each combination of air gap and phantom thickness. Before the complete set of images was taken a dark field calibration image was taken to correct for additive electronic noise. Also a flood field calibration field was taken to correct the gain for each individual pixel. The EPID was exposed to 10 MU at a dose rate of 100 MU/min for each measurement. Four sets of data were acquired for each SDD Experimental setup for specific objective #7 Figure 4. The geometric setup of the linear accelerator, scattering phantom material and ionization chamber for specific objective #7. 21

28 setting with 0.3cm, 10cm, 20cm and 30cm thicknesses of PMMA scattering material in addition to the 1.3cm Al thickness of the Las Vegas phantom. For each geometric setup, images were taken for a field size of 15x2 cm² with the Y jaws set asymmetrically at 5 positions centered along the 5 rows of holes on the Las Vegas phantom. Images were also taken of an open field 15x15 cm² field size for comparison to the slit scan measurements. Each image was acquired on the Varian PortalVision software. The object detectability was determined by optimizing the window and level of the image and then counting the number of holes that were visible on the Las Vegas phantom (see Figure 5). Evaluation of the object detectability was performed for SDD of 120cm and 150cm. The total numbers of holes detected Las Vegas Phantom Figure 5. The regions of interest were measured in locations A and B. The total number of holes visible represented object detectability. 22

29 on the image of the Las Vegas phantom were counted with a value of 1 for a fully visible hole and 0.5 for a partially visible hole. A contrast-to-noise ratio (CNR) was determined by creating two 5x5 pixel region of interest. The two regions of interest were located on the Las Vegas phantom image as illustrated in Figure 5. Region of interest A was positioned in the largest hole in the second row and region B was located adjacent to the hole on the thick portion of the Las Vegas phantom. The average pixel value, and standard deviation for the region of interest were recorded for each measurement. The CNR was calculated by (Motz, 1978) CNR = I A - I B σ where I A and I B are the average pixel values of the regions of interest A and B. The noise (σ) is given by the following equation σ = (σ A 2 + σ B 2 ) ½ where σ A and σ B are the standard deviation values of regions of interest A and B. The CNR for the slit scan and 15x15 cm 2 open field were compared. The ratio of the slit scan CNR to the open field CNR was calculated to evaluate the difference in the two CNR measurements. Specific Objective #8 - Generate a Slit Scan Image These measurements were made using the Varian 21EX linear accelerator operating at the energy of 6 MV photons. The linear accelerator geometry was set at a gantry and collimator angle of 270º and table angle of 90º (see Figure 4 above). The distance to the entrance surface of the scattering material was maintained at 80cm SSD. The Las Vegas phantom was placed on 23

30 top of a Styrofoam block against the scattering material side closest to the radiation source. The table was adjusted to center of the Las Vegas phantom to the light field crosshair. Images were captured of a series of 15x2 cm² field sizes with the jaws set asymmetrically for 15 positions across the Las Vegas phantom. The asymmetric jaw settings relative to isocenter are listed in Table I. Images were also taken of a 15x15 cm² field size. The SDD was set to 120 and Table I. List of Asymmetric Jaw Settings and Image Processing Edges Asymmetric Jaw Settings Image Processing Edge Image # Y1 Jaw (cm) Y2 Jaw (cm) Edge 1 (cm) Edge 2 (cm) cm. The SDD was measured to the phosphor layer inside the EPID. For each image the EPID was positioned in the center of the radiation field. The EPID position was verified visually with the light field crosshairs for each 24

31 combination of air gap and phantom thickness. Before the complete set of images was taken a dark field calibration image was taken to correct for additive electronic noise. Also a flood field calibration field was taken to correct the gain for each individual pixel. Each image was acquired on the Varian PortalVision software. The EPID was exposed to 10 MU at a dose rate of 100 MU/min for each image. Two sets of data were acquired for each SDD setting with 0.3, and 30 cm thicknesses of PMMA scattering material. The 15 slit scan images were copied from the Varian PortalVision software as dicom images and loaded into the Image J software for image processing. Each image was processed to result in a 15x1 cm² field size centered on the image. The processed image edge positions relative to the isocenter are listed in Table *I. All image data outside the 1 cm width was zeroed out and the new image saved. By zeroing out the data on each side of the acquired slit image, most of the scatter radiation, transmission radiation penetrating the collimator jaws and collimator scatter were eliminated. After all 15 images had been processed, the images were added together to generate a 15x15 cm² field size image. The slit scan image was visually compared to the open 15x15 cm² field size image for image quality differences. To visually evaluate the effect of the slit scan technique a profile of pixel intensity along the column of hole furthest to the left on the Las Vegas image was created for both the slit scan and open field exposure. A comparison of the slit scan image and open field for a 10cm air gap and 30cm PMMA scattering material. The dicom image was loaded into the Image J software were a profile 25

32 can be created. Once the profile was created the numerical data was copied and pasted into an Excel spreadsheet to be graphed 26

33 RESULTS Specific Objective #1 - Ionization Chamber Radiation Measurements The aim of the first objective was to determine the scatter and primary radiation components of various measurement geometries with an ionization chamber. First, the primary component was calculated by determining the zero field size value for each set-up geometry. Clinical output factors were calculated by normalizing the readings to the 10x10 cm 2 field size. Then the zero field size was determined by performing a linear regression with the output values of field sizes 2x2, 3x3, 4x4, and 6x6cm 2 (see Figure!2). The y-intercept of the linear regression corresponds to the zero field size value or the primary component (P) relative to a 10x10cm 2 field. The regression coefficient (R 2 ) was no less than Table II list the primary component for a variety of measurement geometries. Once the primary component was determined the clinical output factors were used to determine the output factors normalized to the primary component. By normalizing to the primary component of the radiation an output factor greater than unity represents the increase in scatter radiation as the field size increases due to collimator scatter and scatter originating from a scattering material incident to the radiation beam. Figures 6-9 illustrate how the scatter component changes versus field size for each experimental set-up geometry. The results of this objective were used to evaluate specific characteristics of how scatter radiation is generated by changing the air gap, thickness of scattering material and field size. This information was also used to compare 27

34 the scatter radiation response of the ionization chamber to the EPID. The complete set of numerical and graphical data can be found in the Appendix. Table II. List of Primary Component Values Relative to 10x10cm 2 Field Using an Ionization Chamber SDD (cm) Thickness (cm) Air Gap P R

35 cm PMMA 10.0cm PMMA 20.0cm PMMA Output Factors for 110cm SDD 1.60 Output Factor Field Size (cm on a side) Figure 6. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 10 cm air gap measured with an ionization chamber cm PMMA 10.0cm PMMA 20.0cm PMMA Output Factors for 120cm SDD 1.60 Output Factor Field Size (cm on a side) Figure 7. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 20 cm air gap measured with an ionization chamber. 29

36 cm PMMA 10.0cm PMMA 20.0cm PMMA Output Factors for 130cm SDD 1.60 Output Factor Field Size (cm on a side) Figure 8. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 30 cm air gap measured with an ionization chamber. Output Factors for 140cm SDD cm PMMA 10.0cm PMMA 20.0cm PMMA 1.60 Output Factor Field Size (cm on a side) Figure 9. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 40 cm air gap measured with an ionization chamber. 30

37 Specific Objective #2 - EPID Radiation Measurements The intent of this objective was similar to objective one but the measurements were made with the EPID. Clinical output factors were calculated by normalizing the readings to the 10x10 cm 2 field size. As before, the zero field size was determined by performing a linear regression with the output values of sizes 2x2, 3x3, 4x4, and 6x6cm 2 (see Figure 2). The y-intercept of the linear regression corresponds to the primary component (P). The regression coefficient (R 2 ) was no less than for all calculations. Table III list the primary component for a variety of measurement geometries. Table III. List of Primary Component Values Using an EPID SDD (cm) Thickness (cm) Air Gap P R

38 Once the primary component was determined the clinical output factors were normalized to the primary component to yield the output factors relative to the primary beam. As done with the EPID measurements, normalizing to the primary component of the radiation results in output factors representing the increase in scatter radiation as the field size increase. Figures illustrate the scatter component change versus field size for each experimental set-up geometry. The results of this objective were used to evaluate specific characteristics of how scatter radiation is generated by changing the air gap, thickness of scattering material and field size. This information was also used to compare the scatter radiation response of the ionization chamber to the EPID. The complete set of numerical data can be found in the Appendix. 32

39 Output Factors for 110cm SDD cm PMMA 10.0cm PMMA 20.0cm PMMA 30.0cm PMMA Output Factor Field Size (cm on a side) Figure 10. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 10 cm air gap measured with the EPID. Output Factors for 120cm SDD cm PMMA 10.0cm PMMA 20.0cm PMMA 30.0cm PMMA Output Factor Field Size (cm on a side) Figure 11. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 20 cm air gap measured with the EPID. 33

40 Output Factors for 130cm SDD cm PMMA 10.0cm PMMA 20.0cm PMMA 30.0cm PMMA Output Factor Field Size (cm on a side) Figure 12. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 30 cm air gap measured with the EPID. Output Factors for 140cm SDD cm PMMA 10.0cm PMMA 20.0cm PMMA 30.0cm PMMA Output Factor Field Size (cm on a side) Figure 13. Output versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 40 cm air gap measured with the EPID. 34

41 Specific Objective #3 - Comparison of Detector Responses The goal of this objective was to compare the EPID and ionization chamber data to evaluate differences in response to scatter radiation. The output factors calculated in previous objectives for both the ionization chamber and EPID were used to evaluate the differences in sensitivity to scatter radiation of the two detectors. The ratio of the output factor of the EPID to the output factor of the ionization chamber is an indication of the relative sensitivity of the EPID to the scatter radiation. The results evaluated two factors that effect signal generated by the two detectors. The first factor was the effect of the scattering material on the detector signal. Measurements were taken with PMMA scattering material thicknesses of 10cm and 20cm as well as a measurement set with no scattering material in the radiation beam. The second factor was the effect of the air gap on the detector signal. A comparison was performed for a SDD of 110cm, 120cm, 130cm and 140cm. Figures demonstrate the differences of the EPID signal for scatter radiation compared to the ionization chamber response. The complete set of numerical data can be found in the Appendix. 35

42 Ratio of EPID/Ionization Chamber Ratio of EPID/ Ionization Chamber at 110cm SDD No PMMA 10cm PMMA 20cm PMMA Field Size (cm on a side) Figure 14. Ratio of EPID/Ionization chamber versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 10 cm air gap. Ratio of EPID/Ionization Chamber Ratio of EPID/ Ionization Chamber at 120cm SDD No PMMA 10cm PMMA 20cm PMMA Field Size (cm on a side) Figure 15. Ratio of EPID/Ionization chamber versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 20 cm air gap. 36

43 Ratio of EPID/Ionization Chamber Ratio of EPID/ Ionization Chamber at 130cm SDD No PMMA 10cm PMMA 20cm PMMA Field Size (cm on a side) Figure 16. Ratio of EPID/Ionization chamber versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 30 cm air gap. Ratio of EPID/Ionization Chamber Ratio of EPID/ Ionization Chamber at 140cm SDD No PMMA 10cm PMMA 20cm PMMA Field Size (cm on a side) Figure 17. Ratio of EPID/Ionization chamber versus field size normalized to a 0x0cm 2 field size for varied scattering material at a 40 cm air gap. 37

44 Specific Objective #4 - Evaluation of Air Gap Effect The aim of this objective was to evaluate the effect of an air gap on the SPR for various imaging geometries. A number of air gap distances were used to investigate how detected scatter is affected by this change. This was done to determine situations where the slit scan technique will be most effective. An ionization chamber was used to measure a series of data sets. The data set with no scattering material was used to correct for or include collimator scatter and other off-focus radiation in the extrapolated zero field size output factor. This results in scatter-fraction determinations that are due mainly from scatter from the patient. These Patient Scatter Fractions were determined primarily for comparison with previously published work (Jaffrey, 1994), in Figures 18 and 19. The effect of an air gap on the signal detected by the EPID was determined by comparing results of the SPR for air gaps of 10cm, 20cm, 30cm and 40cm. For each of these air gaps the PMMA scattering material was varied for thicknesses of 10cm, 20cm, 30cm and a data set without scattering material. This results in SPR differences that are due to scatter from the patient. 38

45 Scatter Fraction versus Air Gap 4x4cm Field Size 10x10cm Field Size 20x20cm Field Size 30x30cm Field Size Scatter Fraction Air Gap (cm) Figure 18. Patient scatter fraction for varied air gap with 20cm of PMMA measured with an ionization chamber. Scatter Fraction versus Air Gap Figure 19. Comparison of published data of patient scatter fraction versus air gap for a 17 cm of PMMA measured with an ionization chamber (Jaffray, 1994). 39

46 Scatter-to-Primary Ratio for 10cm PMMA Thickness cm Air Gap 20cm Air Gap 30cm Air Gap 40cm Air Gap Scatter-to-Primary Ratio Field Size (cm on a side) Figure 20. SPR including collimator and build-up scatter. The air gap was varied with a 10cm PMMA thickness at measured with an EPID. Scatter-to-Primary Ratio for 20cm PMMA Thickness cm Air Gap 20cm Air Gap 30cm Air Gap 40cm Air Gap Scatter-to-Primary Ratio Field Size (cm on a side) Figure 21. SPR including collimator and build-up scatter. The air gap was varied with a 20cm PMMA thickness at measured with an EPID. 40

47 Scatter-to-Primary Ratio for 30cm PMMA Thickness 10cm Air Gap 20cm Air Gap 30cm Air Gap 40cm Air Gap Scatter-to-Primary Ratio Field Size (cm on a side) Figure 22. SPR including collimator and build-up scatter. The air gap was varied with a 30cm PMMA thickness at measured with an EPID. Specific Objective #5 - Determine Scattering Material Effect The intent of this objective was to evaluate the effect of increasing thicknesses of scattering material between the radiation source and a detector. The thicknesses used for the scattering material was 10cm, 20cm, 30cm and a set with no scattering material. The air gap was set to 10cm, 20cm, 30cm and 40cm. To evaluate the effect of scattering material, measurements taken previously with the EPID in Objective 2 were used to calculate the SPR for a range of field sizes. These values were used to determine how varying the thickness of the scattering material effected scatter radiation detected by the EPID. 41

48 Scatter-to-Primary Ratio at 110cm SDD 3x3 Field Size 6x6 Field Size 10x10 Field Size 15x15 Field Size 20x20 Field Size Scatter-to-Primary Ratio Scatter Material Thickness (cm) Figure 23. SPR for a range of field sizes with varied scatter material thicknesses at constant 10cm air gap measured with an EPID Scatter-to-Primary Ratio at 120cm SDD 3x3 Field Size 6x6 Field Size 10x10 Field Size 15x15 Field Size 20x20 Field Size Scatter-to-Primary Ratio Scatter Material Thickness (cm) Figure 24. SPR for a range of field sizes with varied scatter material thicknesses at constant 20cm air gap measured with an EPID. 42

49 Scatter-to-Primary Ratio at 130cm SDD 3x3 Field Size 6x6 Field Size 10x10 Field Size 15x15 Field Size 20x20 Field Size Scatter-to-Primary Ratio Scatter Material Thickness (cm) Figure 25. SPR for a range of field sizes with varied scatter material thicknesses at constant 30cm air gap measured with an EPID Scatter-to-Primary Ratio at 140cm SDD 3x3 Field Size 6x6 Field Size 10x10 Field Size 15x15 Field Size 20x20 Field Size Scatter-to-Primary Ratio Scatter Material Thickness (cm) Figure 26. SPR for a range of field sizes with varied scatter material thicknesses at constant 40cm air gap measured with an EPID. 43

50 Specific Objective #6 - Slit Scan Field Size Evaluation Specific objective six was a necessary step to compare various slit scan field sizes to evaluate the scatter reduction achieved for image improvement. This evaluation was used to determine which field size would be optimal for this project. Profiles along the narrow portion of the slit scan images were created on the Varian PortalVision software and the numerical data loaded into an Excel spreadsheet. Profiles were compared to determine how many pixels formed a plateau or flat portion along the center of the field. This is important to allow an adequate region of interest size for measurements and generate a slit scan image. The number of pixels on the flat portion along the center of the field was 7, 13 and 19 pixels for fields 15x1cm², 15x1.5 cm², and15x2 cm². The SPR for a given slit scan field size was subtracted from the 15x15cm 2 field size SPR. The difference was referenced to a 15x15 cm 2 field size since it is the size to be used in the following objectives to analyze the change in image quality compared to the slit scan technique. 44

51 Field Profiles at 120cm SDD x1.0cm Field Size 15.0x1.5cm Field Size 15.0x2.0cm Field Size Intensity Distance (cm) Figure 27. Profiles of slit scan field sizes at 120cm SDD without PMMA material in the radiation beam. Scatter-to-Primary Ratio Difference Scatter-to-Primary Ratio Difference for a 10cm Air Gap x15 Field Size 1.5x15 Field Size 2x15 Field Size PMMA Thickness (cm) Figure 28. SPR difference from a 15x15cm 2 fields for various slit scan field sizes at a 10cm air gap for various PMMA scattering material thicknesses. 45

52 Scatter-to-Primary Ratio Difference Scatter-to-Primary Ratio Difference for a 20cm Air Gap x15 Field Size 1.5x15 Field Size 2x15 Field Size PMMA Thickness (cm) Figure 29. SPR difference from a 15x15cm 2 fields for various slit scan field sizes at a 20cm air gap for various PMMA scattering material thicknesses. Scatter-to-Primary Ratio Difference Scatter-to-Primary Ratio Difference for a 30cm Air Gap x15 Field Size 1.5x15 Field Size 2x15 Field Size PMMA Thickness (cm) Figure 30. SPR difference from a 15x15cm 2 fields for various slit scan field sizes at a 30cm air gap for various PMMA scattering material thicknesses. 46

53 Scatter-to-Primary Ratio Difference Scatter-to-Primary Ratio Difference for a 40cm Air Gap x15 Field Size 1.5x15 Field Size 2x15 Field Size PMMA Thickness (cm) Figure 31. SPR difference from a 15x15cm 2 fields for various slit scan field sizes at a 40cm air gap for various PMMA scattering material thicknesses. Specific Objective #7 Slit Scan Image Quality Measurements The aim of this objective was to measure the contrast-to-noise ratio (CNR) and object detectability for various geometries using the slit scan technique. The slit scan results were compared to open field measurements. A field size of 15x2cm 2 was used for the slit scan technique and a 15x15cm 2 field size for the open field. The 1.3cm thick aluminum Las Vegas phantom was used as the evaluation tool to determine these values. To determine the CNR the equations described in the methods section were used for each image. As outline in Tables IV-VII the CNR was calculated for each geometric setup. To assist in the evaluation of the CNR tabulated data a ratio of the slit scan and open CNR was calculated to investigate trends. 47

54 Table IV. List of CNR Values for Phantom plus 0.3cm PMMA Material Used Air Gap (cm) SDD (cm) Open CNR Slit CNR Ratio Table V. List of CNR Values for Phantom plus 10cm PMMA Material Used Air Gap (cm) SDD (cm) Open CNR Slit CNR Ratio Table VI. List of CNR Values for Phantom plus 20cm PMMA Material Used Air Gap (cm) SDD (cm) Open CNR Slit CNR Ratio Table VII. List of CNR Values for Phantom plus 30cm PMMA Material Used Air Gap (cm) SDD (cm) Open CNR Slit CNR Ratio The object detectability was performed at a SDD of 120cm and 150cm with PMMA scattering material thicknesses of 0.3cm, 10cm, 20cm and 30cm in addition to the 1.3cm aluminum Las Vegas phantom. The results of the object detectability analyzes are list in Tables VIII-IX. 48

55 Table VIII. List of Object Detectability Values for 120cm SDD for 15x15cm 2 Open Field and 2x15cm 2 Slit Scan Air Gap (cm) PMMA (cm) Open (holes) Slit (holes) Table IX. List of Object Detectability Values for 150cm SDD for 15x15cm 2 Open Field and 2x15cm 2 Slit Scan Air Gap (cm) PMMA (cm) Open (holes) Slit (holes) Specific Objective #8 - Generate a Slit Scan Image The goal of this objective was to demonstrate the ability to generate a megavoltage large field image using multiple slit scan images. As described in the methods section, this objective required a series of 15 individual slit scan images of a 15x2cm 2 field size to be taken. The images were copied as dicom images and opened in J Image software for image processing. The final product of the image processing was a 15x15cm 2 field size generated from 15 slit scan images. Also a 15x15cm 2 image was taken with the same image setup geometry. A comparison of these image techniques are displayed in Figures 32 and 33. To further visually evaluate the slit scan image a profile of pixel intensity along the column of holes furthest to the left on the Las Vegas image was 49

56 created for both the slit scan and open field exposure at seen in Figures 34 and 35. A total of five holes are along the profile. 50

57 Open 15x15cm 2 Las Vegas Phantom Without PMMA Material Image A Image B Figure 32. EPID image of Las Vegas phantom at a 10cm air gap without PMMA material. Image A is the open field and image B is the slit scan technique. Open 15x15cm 2 Las Vegas Phantom With 30cm of PMMA Material Image A Image B Figure 33. EPID image of Las Vegas phantom at a 10cm air gap and 30cm PMMA material. Image A is the open field and image B is the slit scan technique. 51