Recently, there has been renewed interest in decreasing radiation dose to patients from diagnostic imaging procedures. So far, efforts to decrease radiation dose have focused on the amount of radiation delivered from typical techniques and fail to capture the variation in radiation dose between patients. Despite the feasibility of estimating patient-specific radiation doses and the potential for this practice to aid in quality assurance, it is not currently standard procedure for hospitals to monitor radiation dose for all patients. To address this shortcoming, we have developed an institution-wide patient-specific radiation dose monitoring program for computed tomography, digital radiography, and nuclear medicine.

The purpose of this research was to evaluate use of portal monitor in a radiological emergency, measure CT beam quality and CT table attenuation, study the variation of MOSFET sensitivity with integrated dose history and assess the accuracy of AAPM 96 formalism for estimating e ective dose from pediatric CT scans.

With the advent of new, more e cient, rotational therapy techniques such as vol- umetric modulated arc therapy (VMAT), radiation therapy treatment precision re- quires evolving quality assurance. Two dimensional (2D) detector arrays have shown angular dependence that must be compensated for by the creation of angular correc- tion factor tables. Currently available correction factor tables have several underlying assumptions that leave room for improvement: rst, these correction factors assume that the response of all ion chambers is identical for each angle; second, that the ion chamber array response from gantry angles 0 - 180 are equivalent to the response from 180 - 360; and, third, that the response is independent of the direction of rotation.

With continuing advances in medical imaging technologies, there is an increased demand to extract quantitative information from images. This has been particularly vital in the effort to increase the efficacy and accuracy of diagnoses. Quantitative information is readily available in images because the acquisition techniques intrinsically involve physical processes. Quantitative image quality metrics are critical in the evaluation of medical images for diagnostic merit, particularly when used for the characterization and comparison of different systems. When such metrics are based on measurable physical parameters, they can provide valuable information for system optimization. Image quality describes the “goodness” of an image in displaying information for a task. This thesis explored methods of measuring image quality for two scenarios: (1) to characterize 2D flat-panel detector performance and (2) to measure directional spatial resolution for 3D images from breast tomosynthesis.

Positron Emission Tomography is a coincidence-detection-based nuclear imaging modality that has increased in clinical prevalence over the last two decades. Mea- sures have recently been taken to improve the practice, speci cally the synergistic combination with CT, and implementation of iterative reconstruction. The time-of- ight (TOF) technique is another improvement theorized early in PET development, which reduces image noise by measuring the di erence in coincident photon detec- tion times. It was di cult to implement at the time of inception because of limited technologies, but better detectors and electronics have recently made TOF feasible for clinical use. Its gain in image quality has been measured by various methods, but is di cult to quantify because of tradeo s inherent in count-based imaging. This work set out to investigate the image quality gained with TOF imaging by determin- ing the e ective non-TOF scan time required to achieve equivalent image quality as TOF.

Purpose: The goal of this planning study is to determine which sectors of the gantry rotation are most and least important in the treatment of head-and-neck carcinomas with Intensity Modulated Arc Therapy, and then use this knowledge to optimize the arc arrangement by adding arcs to reinforce the sectors that are most significant.

Purpose: Cervical spine (C-spine) lesions are located more centrally than the rest of the spinal column making the c-spine an ideal target for the RapidArc therapy technique. In this study, we explore the use of RapidArc to treat c-spine patients and investigate the target coverage, MU reduction, dose to the body, and intra-fraction motion. Materials and Methods: We studied 8 c-spine SBRT patients. The patients are immobilized with Brainlab head and neck frame. Single-arc RapidArc plans are used for treatment. Before treatment, CBCT images are acquired for target localization. After treatment, post-CBCT images are acquired to assess intra-fraction motion. Actual dose to the spinal cord in CBCT was calculated with Eclipse planning system. Dose to body was measured by looking at volume covered by lower isodose lines. Results: RapidArc plans can provide comparable target coverage and critical organ dose sparing as IMRT. The typical MU reduction compared to IMRT was 24.4%. The intra-fraction motion was -0.1 ± 0.2 cm axial, 0.1 ± 0.1 cm sagittal, -0.1 ± 0.2 cm coronal, and 0.7 ± 0.4 degrees. Typical treatment delivery time (not including set-up) was reduced by approximately 71%. The actual average cord dose received in the patient was 1.5 ± 2 % greater than the planned dose to the cord. The actual dose at the highest 0.5% of the short cord next to the target was 2.9 ± 2% greater than the planned dose. Conclusion: RapidArc is well suited for c-spine treatments and provides an effective way to substantially decrease patient treatment time which aids in decreasing intra-fraction motion as well as improving the comfort level for the patient.

Accurate delivery of external beam radiation therapy relies on localization of the treatment target. Real-time electromagnetic localization systems (ELS) provide real-time tracking of the prostate during radiation therapy treatments using three implanted transponders. In this work, the Calypso® 4D Localization SystemTM (Calypso Medical, Seattle, WA), a commercial ELS, was evaluated. This study had four specific aims. First, the accuracy of the ELS for use in humans was verified through comparison with an on-board imaging kilo-voltage (kV) x-ray system used to produce both planar 2D orthogonal (planar) and volumetric 3D cone-beam computed tomography (CBCT) images. Second, the ELS was used to analyze prostate deformations and rotations occurring over the course of radiation therapy and the associated dosimetric impact was evaluated. Third, motion studies were conducted to investigate the use of the ELS during intensity-modulated arc therapy (IMAT) for prostate cancer. Fourth, appropriate planning margins were determined and new treatment plans utilizing a smaller planning margin were applied to weekly CBCTs for two patients who exhibited large transponder displacements (translational or rotational) to investigate the dosimetric effect of using smaller margins.

The MVS image sets consisted of the 25 raw projection images acquired over an arc of approximately 45° using a Siemens prototype breast tomosynthesis system. The mammograms were acquired using a commercial Siemens FFDM system. The raw data was taken from both of these systems for 27 cases and realistic simulated mass lesions were added to duplicates of the 27 images at the same local contrast. The images with lesions (27 mammography and 27 MVS) and the images without lesions (27 mammography and 27 MVS) were then post-processed to provide comparable and representative image appearance across the two modalities. All 108 image sets were shown to five full-time breast imaging radiologists in random order on a state-of-the-art stereoscopic display. The observers were asked to give a confidence rating for each image (0 for lesion definitely not present, 100 for lesion definitely present). The ratings were then compiled and processed using ROC and variance analysis.

In prostate IMRT treatment planning, the variation in patient anatomy makes it difficult to a priori estimate the maximum extent of dose reduction possible to rectum and bladder, and the current planning process may not provide the best treatment plan to the patient. Previous work shows the capacity to decrease the dose to the critical organs by applying a treatment plan from a database as a starting point. The matching treatment plan is found through an anatomical match of the seven Beam's-Eye-View projection images, compared using mutual information.
This study makes use of 100 treatment plans from patients treated at Duke. For each case, seven Beam's-Eye-View projection images were created, and the components that contribute to the overall mutual information score were manipulated in such a way as to provide a better correlation between the new, weighted mutual information and the resulting treatment plan quality. The optimization of the weighting was done by manipulating the weights along a vector of steepest descent as defined by an objective function.
The initial correlation between mutual information and treatment plan quality was R2 = 0.34, and after optimizing the weighting of the individual components that constitute mutual information, the correlation increased to R2 = 0.50. This result shows that through a manipulation of the components of mutual information based on a fit to known treatment plan qualities, a better correlation between the known weighted mutual information and the expected imposed treatment plan quality may be drawn.

Purpose: Quality control of soft-copy displays is critical to ensure the proper contrast rendition of medical images. The American Association of Physicists in Medicine’s (AAPM) Task Group 18 (TG-18) has developed a set of testing parameters for the acceptance testing and quality control of medical grade displays. This paper addresses practical challenges associated with the broad implementation of TG-18 in a clinical setting. Methods: First, a computer model was developed to determine the effects of ambient light variations on the contrast response of a DICOM GSDF calibrated display. The model was based on an LCD displays with diffuse reflection coefficients of 0.0017 sr-1, 0.0060 sr-1, 0.0080 sr-1, and 0.0200 sr-1. Second, the influence on display assessment due to interdevice variability and measurement techniques was established. Finally, the utility of a commercially available quality control program for remote monitoring of soft-copy displays was examined by confirming the accuracy and precision of the program. Results: In terms of ambient light effects, the results suggest that the maximum allowable increase in ambient lighting can be determined for primary and secondary class displays by the following equations.

Purpose: The ability to measure tumor hypoxia in vivo is exceedingly important as studies have shown hypoxia to play a vital part in malignant progression and treatment efficacy. The aim of this study is to employ optimal filter sets in conjunction with a fiber optic probe to measure the diffuse reflectance of various phantoms over the wavelength range of 400nm to 650nm and then extract the optical properties of the phantom using the neural network algorithm. Materials and Methods: Initially, the neural network and genetic algorithms are used to obtain optimal filter combinations. Seven different phantoms with various scattering and absorption levels were then created and then the diffuse reflectance of each phantom was measured using a filter/fiber optic probe instrument in combination with a spectrometer. The neural network algorithm was then again employed to extract the absorption and reduced scattering coefficients from each phantom with each of the eight optimized filter sets using both a low and high scatter reference phantom. Results: Filter sets 1, 2, and 5 extracted both optical properties with the least amount of error overall when using both reference phantoms. The errors were higher when a low scatter phantom was used as reference for a high scatter phantom and vice versa. Discussion and Conclusions: This study demonstrated the feasibility of using a neural network based model to extract optical properties of phantoms with different absorption and scattering levels from collected diffuse reflectance data. In the future, this study will be expanded to examine data with varying saturation levels and also compare the results with the current Monte Carlo model which models reflectance over the entire spectrum.

The goal of this thesis project was to investigate the detectability of pulmonary nodules with chest tomosynthesis as compared to conventional chest radiography and dual energy chest imaging in human observer studies. Based on observer readings, statistical analysis (JAFROC) was performed to determine the accuracy of diagnosis in chest tomosynthesis in comparison to conventional and dual energy radiography.

This study took imaging studies from 34 subjects who each had a conventional chest exam, dual energy exam, tomosynthesis exam, and CT exam. The CT exam for each patient functioned as the reference method. From this reference method, the location of nodules was determined on images from the other three modalities. Two imaging physicists served as observers in a jackknife free response receiver operating characteristic (JAFROC) study. Each observer marked the location of objects thought to be nodules on the images and gave a confidence rating of that object being a nodule on a scale from 1 to 100. The detectability performance was determined using the lesion localization fraction, non-lesion localization fraction, and JAFROC figure of merit of Chakraborty. Four thoracic radiologists also served as observers in a preliminary cohort of cases in a larger observer study. The reading time of the radiologists was noted for conventional, dual energy, and tomosynthesis images.

Chest tomosynthesis substantially and significantly improved the detectability of lung nodules and the accuracy of observer performance, with non-expert observers. The reading time with chest radiologists was longer with tomosynthesis than with combined dual energy and conventional radiography. Radiologist observers are continuing to read a cohort of 72 human subjects to complete an assessment of the performance of trained clinical observers.

Purpose: To assess the quantitative potential of breast tomosynthesis by estimating the percent density of voxelized anthropomorphic breast phantoms. Method and Materials: A Siemens breast tomosynthesis system and voxelized anthropomorphic breast phantoms were modeled using Monte Carlo simulation. The phantoms represented 35%- and 60%-dense breasts of 5cm thickness. The images generated by the simulation were reconstructed using Siemens filtered back-projection software. The non-uniform background due to scatter, heel effect, and limited angular sampling was corrected with a voxel-by-voxel scaling factor based on a simulated, uniform 100% fatty breast phantom. To estimate the density of each slice, the total number of fatty and glandular voxels was calculated using two thresholding algorithms, with and without the use of prior knowledge about the anatomy. Finally, the estimated density of the reconstructed slice was compared to the known percent density of the corresponding slice from the voxelized phantom. Results: When thresholding with prior information, overall density estimation errors for the central eleven slices were -5.0% and -2.6% for the 35% and 60% dense phantoms, respectively. With the full automated, errors for central eleven were -8.0% and 6.3%, respectively. Voxel to voxel matching of the phantom vs. reconstructed slice demonstrated 75.7% and 75.2% respectively of voxels were correctly classified. Conclusion: The errors in percent density estimation were <8% for both the phantoms thus implying that quantification of breast density using tomosynthesis is possible. However, limitations of the acquisition and reconstruction process continue to pose challenges in density estimation leading to potential voxel to voxel errors that warrant further investigation.

Purpose: An Investigation and characterization of the dosimetric properties of a potential standardized formulation of a 3D radiochromic plastic dosimeter. Also, the feasibility of using the PRESAGE/Optical-CT system for 3D dosimetry around a brachytherapy source was investigated. Method and Materials: Radiation induced optical density measurements of small volumes of the radiochromic dosimeter were used to quantify the dosimetric characteristics of interest via Thermo Scientific Genesys 20 Spectrophotometer. Dose-rate dependence, dose sensitivity, stability of response, energy sensitivity, and temperature dependence were the key dosimetric quantities of interest. Brachy dose distributions were obtained by irradiation of cylindrical PRESAGE volumes 6cm in diameter by 8cm height with a GammaMed 12i Ir-192 HDR unit (Varian Medical Systems). Reconstructed optical density values were normalized to a norm point chosen based on the irradiation prescription. Results: Optical density change was found to be linear with dose after irradiation. After the first irradiation optical density cleared in two weeks stored at room temperature. Three subsequent irradiations, allowing each to fully clear, of the same small volumes showed an equal increase in sensitivity to dose from the first irradiation. Line profile comparisons between PRESAGE and Eclpise show very good agreement, even in low dose regions. Conclusion: The “reusable” formulation is reusable when stored at room temperature (~22°C). Also, the PRESAGE/Optical-CT shows potential in evaluation of 3D dose distributions around brachytherapy sources.

The purpose of this work is to improve the NPS, and thus DQE, of CR images by correcting for pixel-to-pixel gain variations specific to each plate. Ten high-exposure open field images were taken with an RQA5 spectrum, with a sixth generation CR plate suspended in air without a cassette. Image values were converted to exposure, the plates registered using fiducial dots on the plate, the ten images averaged, and then high-pass filtered to remove low frequency contributions from field inhomogeneity. A gain-map was then produced by converting all pixel values in the average into fractions with mean of one. The resultant gain-map of the plate was used to normalize subsequent single images to correct for pixel-to-pixel gain fluctuation. The normalized NPS (NNPS) for all images was calculated both with and without the gain-map correction. The NNPS with correction showed improvement over the non-corrected case over the range of frequencies from 0.15 – 2.5 mm-1. At high exposure (40 mR), NNPS was 50-90% better with gain-map correction than without. A small further improvement in NNPS was seen from careful registering of the gain-map with subsequent images using small fiducial dots, because of slight misregistration during scanning. CR devices have not traditionally employed gain-map corrections common with DR detectors because of the multiplicity of plates used with each reader. This study demonstrates that a simple gain-map can be used to correct for the fixed-pattern noise and thus improve the DQE of CR imaging. Such a method could easily be implemented by manufacturers because each plate has a unique bar code and the gain-map could be stored for retrieval after plate reading. These experiments indicated that an improvement in NPS (and hence, DQE) is possible, depending on exposure level, over all frequencies with this technique.

Purpose: Changes in patient volume, due to tumor shrinkage, dehydration, dysphagia and atrophy, could present issues in the accuracy of dosimetry throughout the course of treatment. The aim of this work is to study the dosimetric impacts of the volumetric changes during IMRT and to investigate the feasibilities of electronic portal imaging device (EPID) in predicting the impacts. Materials and Methods: An anthropomorphic head and neck phantom was used to represent two scenarios: symmetric and asymmetric volume loss. The phantom was simulated and planned according to the head and neck protocols used in our clinic. Dose volume histograms (DVH) were generated for each set up scenario and were used to calculate the integral dose expected at the coincident volume of the phantom. During treatment delivery, the EPID captured exit fluence of each beam at each level of bolus thickness. These images were quantitatively analyzed using gamma analysis with criteria of 3% and 3mm dose difference and distance-to-agreement respectively. Results: Comparing maximum to minimum volume in the symmetric situation with DVH generated in Eclipse show substantial fluctuations in dose. When comparing five layers of bolus material to zero layers of bolus material, the changes were most significant. The asymmetric volume change predicted dose fluctuations that were less significant than the symmetric phantom. As for gamma analysis, a quantitative evaluation of the integrated dose fluence, captured by the EPID, showed extreme variability in the images with five layers of bolus when compared to images with no bolus. Less significant variation was shown in layers of closer thicknesses, as expected. Conclusions: The phantom study indicates that volume loss could contribute to clinically considerable changes in the dose delivered to target and organs at risk. The proposed technique using EPID could provide valuable information about the variation of dose due to volumetric changes and might be potentially useful.

Introduction: To develop and characterize a novel dosimetric platform, consisting of PRESAGE/optical-CT and quadrant phantom, for verification of the accuracy of RapidArcTM head & neck radiation therapy. Materials & Methods: A 16cm x 12cm cylindrical quadrant phantom was constructed from four pieces of polyurethane. Holes were drilled into the axial and sagittal planes where metal barbs were fastened to serve as fiducial markers. Gafchromic EBT2 films were placed within the quadrants and the entire structure was placed within a waterfilled Styrofoam container for CT-planning and irradiation. As a proof-of-principle, a two field plan, consisting of 6cm x6 cm anterior-posterior and right-lateral beams was generated in ECLIPSE and the phantom was irradiated three times to characterize the dosimetric accuracy and reproducibility. In addition, the quadrant phantom and PRESAGE were irradiated with a RapidArc head & neck verification plan and the dosimetry was characterized. Dose profiles were created and 2D gamma analysis was performed between the ECLIPSE, EBT2 and PRESAGE datasets using the Normalized Dose Distribution (NDD) algorithm. Results: The proof-of-principle analysis shows excellent reproducibility between three separate irradiations, with 2.2% pixel-by-pixel percent standard deviation between all three films. Likewise, the absolute dosimetric accuracy between the EBT2 films and v ECLIPSE was promising, with 98.8% and 96.3% passing rates for (3%, 3mm) gamma criteria in the Sagittal and Axial planes, respectively. Analysis of the RapidArc plans indicates less agreement between EBT2 and ECLIPSE in both planes. Even poorer dosimetric agreement has been observed between PRESAGE and ECLIPSE, as well as PRESAGE and EBT2. Conclusion: The quadrant phantom shows a high level of reproducibility and accuracy for a simple two-field plan and reasonably good accuracy for RapidArc plans. The PRESAGE data is not as promising and, therefore, more plans will need to be evaluated to ensure whether or not this platform is well suited for RapidArc.

Mammography is known to be one of the most difficult radiographic exams to interpret. Mammography has important limitations including superposition of normal tissue that can obscure a mass, chance alignment of normal tissue to mimic a true lesion, and the inability to derive volumetric information. It has been shown that stereomammography can overcome these deficiencies by showing that layers of normal tissue lay at different depths. If standard stereomammography (i.e. a single stereoscopic pair consisting of 2 projection images) can significantly improve lesion detection, how will multi-view stereoscopy (MVS), where many projection images are used, compare to mammography? The aim of this study was to assess the relative performance of MVS compared to mammography for breast mass detection.

108 image sets were shown to five full-time breast imaging radiologists in random order on a state-of-the-art stereoscopic display. The observers were asked to give a confidence rating for each image (0 for lesion definitely not present, 100 for lesion definitely present). The ratings were then compiled and processed using ROC and variance analysis.

The mean AUC for the five observers was 0.614±0.055 for mammography and 0.778±0.052 for multi-view stereoscopy. The difference of 0.164±0.065 was statistically significant with a p-value of 0.0148. The differences in the AUCs and the p-value suggest that multi-view stereoscopy has a statistically significant advantage over mammography in the detection of simulated breast masses. This highlights the dominance of anatomical noise compared to quantum noise for breast mass detection. It also shows that significant lesion detection can be achieved with MVS without any of the artifacts associated with tomosynthesis.

There is growing need for small animal dosimetry in the bioresearch community. The choice of dosimetry calibration technique as well as the dosimeter employed to establish dose to a medium and dose rates for irradiation is essential, and should be considered for exposure in different irradiator systems.

Both X-ray irradiators were characterized and parameter values including kVp, HVLs, and in-air dose rates were established for future reference. Based on the current dosimetry techniques and devices used, employing TG-61 as compared to the MOSFETbased phantom dosimetry technique results in considerable higher dose rates, which means that the mice are really being underexposed. Further investigation of the ion chamber detectors used for TG-61 dose measurements and to calibrate the MOSFET dosimeters is necessary to ensure accuracy in resulting dose rates. The choice of calibration and correction factors employed in the TG-61 dose calculation needs to be evaluated to ensure accuracy in resulting dose rates. We conclude that use of AAPM TG- 61 for small animal dosimetry may have limitations in accurately predicting absorbed doses in small animal dosimetry, and that the MOSFET-based phantom method, in addition to better simulating real-life mice geometry and providing immediate readout after exposure, proves to be more practical for use. Mice exposure in the cesium-137 irradiator based on the new geometry that gives higher dose rates and is in the more uniform region provides shorter exposure times for research investigators as well as a more uniform dose distribution.

Radiation therapy’s ability to plan, calculate, and deliver radiation doses has accelerated well beyond its ability to fully measure and compare those doses. Therefore, there is a pressing need for robust, reliable, 3D dosimetry to verify the accuracy of the clinical therapeutic delivery of external beam radiation. PRESAGE/optical-CT/CERR is a system that is powerful and flexible enough to keep up with the complex dosimetric demands present in today’s radiation oncology clinics. This work has shown the feasibility of a novel 3D dosimetry system as applied to the verification of small field, thoracic, gated irradiations. It represents an improvement over previous 3D dosimetry techniques and shows promise as the groundwork for future collaborative investigations involving even more clinically challenging scenarios.

This study shows a new approach to commissioning radiosurgery fields using a reliable and accurate 3D dosimeter. This study makes use of the concept that voxel volume in 3D data sets can govern the accuracy of the dosimetry in a way similar to that of the active volume of a point detector. PRESAGE has the potential to provide energy, dose rate, and directionally independent measurements of many beam characteristics. The high-resolution optical CT methods employed in this work served to extend the existing applications of PRESAGE to a new area: small field dosimetry. Being both a dosimeter and a phantom, the placement errors that occur with a water tank and point detector system are almost absent. In addition, the problem of choosing a dosimeter with an appropriate sensitive volume is avoided by the inherent high resolution of most optical CT methods. In the future, PRESAGE dosimetry will benefit from larger CCD-based scanners. At this point, the OCTOPUS is the only realistic option for 3D digitization of large PRESAGE volume. This laser-based system, though quite dependable, has many limitations. This work shows the utility present in a CCD-based system; a system limited predominantly by its field of view. It is reasonable to assume that using larger telecentric lenses would accommodate larger fields of view and larger dosimeters. The use of such lenses will be the topic of future investigation. These developments in PRESAGE dosimetry will allow clinicians and researchers to digitize large volume of high-resolution 3D dosimetric data in a short period of time.

This project was treated as a feasibility study, not a recipe for extracting a respiratory trace. The question remains as to the best way to represent a set of planes as a single measurement. A summing and center of mass technique were both considered here, but neither was determined to be successful in all situations. There is still a great deal of room for improvement in the signal processing. The sinogram techniques are not devoid of merit, but the Mean Ring method has the advantage of better statistical confidence due to much higher counts in each bin.

This work has significant advantages over other data driven methods because it does not require any human intervention or image reconstruction; thus in theory it could be implemented in parallel with a clinical acquisition without adding time or cost to a protocol.

No research project description available.

The ability of radiation therapy and chemotherapy to kill tumor cells is limited in part by biological mechanisms that inhibit or repair DNA damage.   Hyperthermia (HT) has been shown to reduce the effect of many of these mechanisms.  Hyperthermia refers to the heating of tissue, usually in the range of 40°C – 45°C, to obtain known therapeutic benefits. When used with radiation therapy (RT), hyperthermia is known to increase tumor control. Currently, one of the most significant limits to HT + RT and/or CT efficacy has been the lack of a noninvasive thermometry system that would allow the accurate measurement of thermal maps that would in turn allow accurate modeling of thermal dose and effects. The work that I have done for my thesis will examine the utility of a new non-invasive thermometry method in understanding temperature sensitive liposome (TSL) delivered dose.

The dual x-ray system constructed in radiation oncology models the kV imaging systems commonly found on clinical linear accelerators. What makes this system unique is that it uses two separate imaging chains, each consisting of an x-ray tube, detector, and stage, to acquire both fluoroscopic images and tomographic data. Each detector is positioned 50 cm from the rotational center of the stage and 150 cm from the focal spot of its respective tube. Each tube and detector is aligned such that the central ray passes through the axis of rotation of the stage and the central rays for each tube are at a right angle. This system is capable of taking tomographic data as well as real-time fluoroscopic images.

The goal of this project is twofold. The first goal of this project is to design and implement a timing and communications scheme that would allow the dual x-ray system to function as one device that can be easily operated by a single user. The second goal is to begin an investigation into the effects of cross scatter on image quality.

Introduction: It has been shown that cone-beam computed tomography (CBCT) can potentially be used in dose calculation for treatment planning and verification.  This study investigates the Hounsfield Unit (HU) and dosimetric properties of various CBCT techniques using phantoms compared to conventional CT to determine which techniques are proper for dose calculation. Materials and Methods: HU of material disks in Catphan and HU profiles for homogeneous and nonhomogeneous phantoms for six different CBCT techniques (Low-Dose Head, Standard-Dose Head, High Quality Head, Pelvis Spotlight, Pelvis, and Low-Dose Thorax) were compared to CT images. Plans with a single photon beam based on CBCT techniques and conventional CT of phantoms were compared using dose values at the isocenter, isodose distributions, dose volume histograms, and gamma analysis. Results: HU values of all the CBCT techniques are very close to those of CT in Catphan and in homogeneous phantoms except for the Low-Dose Thorax (LDTH) technique, which overestimated the HU value by 85.  However, the HU profile for the LDTH technique in the lung area matched well with the CT compared to the other techniques.  The dose distributions based on all CBCT techniques in the homogeneous phantoms matched well with the CT except for LDTH, which underestimated the dose.  For the lung area, the LDTH-based dose distribution matched well with CT and the other CBCT techniques did not.  For the spine area, the dose distribution of all techniques matched relatively well with CT, with all techniques overestimating the dose by less than 2% except for LDTH technique, which underestimates the dose by less than 2% compared to CT. Conclusion: This study compared the HU values and dose calculations based on six different CBCT techniques with those of conventional CT to determine which CBCT techniques are acceptable for treatment planning and verification.  The dose calculation accuracy depends on the combination of the CBCT acquisition technique and treatment site.  A CBCT technique should be properly selected for the patient treatment site if CBCT-based dose calculation is to be performed.

In medical physics it is well understood that conformity in dose distribution is a desirable goal in radiation delivery. This assertion is a result of the unavoidable proximity of malignancies to normal, healthy tissue. The goals of external beam radiation therapy are described as the delivery of a “precisely measured dose of irradiation to a defined tumor volume with minimal damage to surrounding healthy tissue” (Chao, Perez, & Brady, 2002). This study quantifies the potential for normal tissue sparing in external beam radiation therapy boost plans for cervical cancer by employing patient-specific margins to IMRT and RapidArc. Various target volume expansion margins are evaluated, including best-case and standard-case scenarios, which encompass internal organ motion over the intrafractional treatment period. In establishing the non-invasive application of an advanced image-guided radiation therapy procedure to cervical boost fractions, patient-specific margins lead to measurable dose reduction to normal tissues.

In nuclear medicine, radiation is administered to the patient orally, inhalationaly or intravenously via a radiolabeled compound, referred to as a radiopharmaceutical or radiotracer.  The chemical then distributes itself throughout the body based on its chemical properties.  As the radionuclide attached to the radiotracer decays, it emits radiation.  This radiation passes through the body to a ring of detectors around the patient.  Not all the radiation emitted passes through the body.  A certain percentage is scattered, never reaching the detector.  Unlike transmission imaging where this attenuation is the basis of the image formation, in emission imaging, scattered radiation is a source of noise.  The end result of emission radiology is a set of projections of the radioactivity concentration along specific lines through the body.  These projections are then organized into a two-dimensional map of the radiotracer’s distribution in the patient.  With the advent of tomographic methods, these two dimensional images can be transformed into 3D data sets which allow the user to slice the data in various ways to view desired cross sections of the body.

Whole Body Counting is a technique that is used to make in vivo measurements of radioactivity in test subjects (humans and animals). Typically, the radiation is detected in the form of gamma rays that are emitted from either naturally occurring substances or contaminants present within the body. Sources of radiation contamination which in turn lead to internal exposure include food, drinking water, occupational hazards and environmental contamination from a nuclear disaster. Due to both the increase in the number of power plants in operation (the EPA estimates there are over 100 active nuclear reactors in the United States) and public concern of nuclear terrorism there is a continuous need to have protocols in place to deal with radiation contamination. These protocols should provide a method of identifying and quantifying the radiation contamination.

The results show that with fairly simply calibrations the whole body counter can produce reliable and accurate in vivo activity measurements for a wide range of energies provided that the activity levels are between 1 and 250 µCi. Below 1 µCi the statistical variations are too great to provide reliable results, unless longer counting times are used. Above 250 µCi the amount of attenuation provided by tissue is not enough to overcome the dead time effects.

It has been shown that post-surgical radiation treatment can greatly improve local tumor control in breast cancer patients.  Machtay et al. have also studied the correlation between radiation dose levels and survival, and found an estimated 18% decrease in the risk of death with every 10 Gy increase in biological equivalent dose.  It is therefore of great importance that tumor structures receive high levels of radiation for adequate tumor cell kill.  However, it is common to have radiation sensitive normal structures (termed critical structures) which must receive minimal radiation doses due to an increased risk of complications or normal tissue death post-treatment .  Avoiding dose to critical structures becomes an even greater challenge when the irradiation area is in a region accompanied by motion, which can be attributed to skeletal, cardiac, gastrointestinal, or respiratory systems.

A correlation was found to exist between the breath-hold level and intrafractional chestwall motion.  Threshold correlations with intra/interfraction motion were also found.  Results from this study can be used to guide setting the threshold for amplitude-gated deep-inspiration-breath-hold radiation therapy.

A most important criterion for the success of radiation therapy treatment is to reduce the doses to the critical structures while obtaining a good tumor control. With advanced techniques such as Intensity Modulated Radiation therapy, it is feasible to meet the above requirements with minimal limitations. IMRT makes use of multi-leaf collimators (MLC) that modulate beamlet fluences to conform the doses to the target structure while avoiding excess doses to the critical structures. In the design of the IMRT treatment plan on Eclipse planning system, the Dose Volume Histograms of the target and the critical structures can be manually modified using the optimization process. Achieving this delivery in practice requires a sophisticated linear accelerator (LINAC) delivery system. The dose patterns planned for the treatment become complex. The LINAC might not always be able to deliver such complex dose distributions exactly as planned due to its physical constraints. This leads to a deviation of the delivered dose pattern from the planned dose pattern.  The efficiency of the LINAC in delivering planned dose patterns precisely plays a key role in the success of IMRT.

No research project description available.

Investigating the potential of pharmacokinetic parameters extracted from Dynamic Contrast Enhanced-Magnetic Resonance Imaging (DCE-MRI) to estimate treatment response in locally advanced head and neck cancer patients
Advisor: Oana Craciunescu, PhD

Purpose: The purpose of this work was to assess the potential of DCE-MRI to assess changes in tumor physiology in locally advanced head and neck cancer (LAHNC) patients receiving Targeted Therapies and cisplatin based concurrent Chemoradiation. In order to generate DCE-MRI extracted parameters that accurately reflect changes in tumor physiology, certain key factors must be considered. These include knowledge of the native longitudinal relaxation time (T10) whose accurate determination depends on choice of flip angle (FA) combinations used. Additionally appropriate methods of contouring regions of interest (ROI) must be selected. Finally, it is important to establish that the variations measured within a tumor are less than that observed between tumors in different patients as well as determine the inter-user variability in cases where ROIs are independently contoured by different users.

Advisor: Adam Wax, PhD

Microbicide gel vaginal distribution is key to gel effectiveness.  The ability to measure gel coating thickness accurately is paramount in the evaluation of candidate microbicidal products.  This investigation presents a multiplexed, low coherence interferometry (LCI) instrument to be used as a new, label-free, high-resolution method for intravaginal measurement of microbicidal get distribution.  It also explores the feasibility of combining the multiplexed LCI system with a fluorescence-based confocal microscopy system. Vaginal imaging techniques using fluorescent labels have been previously employed to map thickness of deployed gels in women1-3.  However, these fluorescent techniques are limited by the contrast agents that are used to generate a signal.  LCI exploits the low coherence of broadband light to achieve depth-resolved thickness measurements with micron resolution.  LCI is also a time independent method of acquiring depth information, which makes it very desirable for in vivo studies. A benchtop LCI system was previously built, based on a Michelson interferometer geometry, which was shown to be an effective method of measuring gel thickness, with precision and accuracy exceeding fluorometry techniques4.  This study presents a new LCI system that uses six multiplexed channels to achieve broad area scanning without the need for a mechanical scanner.  The system’s linearity and accuracy were first verified with a custom calibration socket, which contained various wells of known depths.  Measurements on test gels were then performed using a custom tissue phantom, which was created to have optical properties similar to epithelial tissue. In a second branch of the study, a previously retrofit laser scanning confocal microscope was adapted in order to image a fluorescent-labeled placebo gel in vitro. A placebo gel was labeled with the fluorescent agent indocyanine green (ICG), and simple images were captured to show the possibility of using confocal microscopy to image microbicide gel distribution.  Photobleaching and low quantum yield were determined to be the two major limiting factors in imaging microbicide gel distribution in vivo with ICG as a contrast agent. 

Advisor: James Provenzale, MD

The purpose of this study is to evaluate the potential utility of dynamic CT perfusion imaging as an alternative method for characterization of microcirculation in brain tumors, in lieu of the more conventionally utilized dynamic contrast-enhanced magnetic resonance perfusion imaging. The specific question to be answered in this study is: Do the cerebral perfusion parameters obtained from dynamic CT perfusion imaging correlate well with those obtained from DCE MR perfusion imaging? We will compare hemodynamic measurements in brain tumors using CT and MR perfusion imaging by measuring the relative cerebral blood volume (rCBV) and mean transit time (MTT) obtained from each of these imaging modalities in CT and MR examinations performed within 24 hours of one another.

Advisor: Mark Oldham

Purpose: Rind radiosurgery is a highly modulated technique that aims to treat the proliferative disease of ring-enhancing intracranial lesions, and to reduce the dose to normal brain tissue when compared with conventional radiosurgery. Reduction in normal tissue dose reduces CNS toxicity with two consequential benefits – latitude for dose escalation of proliferative tissues and the potential for an extension of stereotactic radiosurgery protocols to include lesions exceeding 4.0 cm in diameter. This study aims to quantify the potential dose reductions to normal tissue for intensity modulated rind radiosurgery (IMRR) treatments.
Methods and Materials: A set of IMRT treatment plans were created in the Eclipse planning system to quantify dose sparing for a range of lesions (diameter = 3, 5, 7, 9 cm) and rind thicknesses (0.5, 1.0, 1.5, 2.0 cm). Plans were created for both the conventional (solid tumor) and rind (ring-enhancing) scenarios. A variety of beam arrangements and IMRT constraint schemes of varying levels of complexity were evaluated. All doses were computed using the pencil beam convolution algorithm with a dose grid size of 1.25 mm. The dose reduction to normal brain was evaluated by comparing the integral dose to a shell of normal tissue encompassing the majority of normal tissues as well as a 1.0 cm shell encapsulating the high dose region surrounding the PTV.
Results: The reduction in normal tissue integral dose ranged from 0 to 5.5% with the greatest savings correlating with smaller lesion sizes. Maximum dose sparing of 5.5% was achieved using a 12 non-coplanar beam arrangement, and a constraint scheme which held no limits on dose deposited in the center of the tumor. A dramatic increase in the number of monitor units of up to 123% was observed in the delivery of rind distributions.
Conclusions: These results indicate that the integral dose to normal brain tissue can be reduced with IMRR treatments. The dose sparings were less however than previously reported with the radiosurgery cone technique, and this is likey due to leakage associated with the extreme modulation required for the delivery of rind treatment plans.

Information-Theoretic CAD in Mammography: Effect of Region Size on Mass Detection and Multi-Size Analysis for Performance Improvement
Advisor:  Georgia Tourassi, PhD

Featureless, knowledge-based CAD systems are an attractive alternative to feature-based CAD, because they require no to minimal image preprocessing. Such systems compare images directly using the raw image pixel values rather than relying on low-level image features. Specifically, information-theoretic measures such as mutual information (MI) have been shown to be an effective, featureless, similarity measure for image comparisons. MI captures the statistical relationship between the gray level values of corresponding image pixels. In a CAD system developed at our laboratory, the above concept has been applied for location-specific detection of mammographic masses. The system is designed to operate on a fixed size region of interest (ROI) extracted around a suspicious mammographic location. Since mass sizes vary substantially, there is a potential drawback. When two ROIs are compared, it is unclear how much the parenchymal background contributes in the calculated MI. This uncertainty could deteriorate CAD performance in the extreme cases, namely when a small mass is present in the ROI or when a large mass extends beyond the fixed size ROI. This thesis evaluates the effect of ROI size on the overall CAD performance and proposes various solutions to the identified limitations.

Advisor: Timothy G. Turkington, PhD

Slow rotation CT scanners, such as the one found on the Infinia Hawkeye SPECT/CT system (GE Healthcare), can have images affected by motion artifacts due to data inconsistencies when reconstructed using filtered backprojection. This project seeks to improve CT image quality by masking portions of the CT data that contain motion and then reconstructing the image using iterative algorithms which have more flexibility than filtered backprojection. Streak artifacts appear in FBP images from slow rotation CT systems, caused by data inconsistencies due to breathing motion. Reasonable images can be produced with many projection angles missing when there is no motion with iterative reconstruction. With respiratory motion, there are too many projections affected by motion to remove all of them. Applying POMs to the portions of the sinogram with motion along with iterative reconstruction techniques improves the image quality of simple phantoms. Further study is needed to determine if this technique may be directly applied to patient data to decrease motion artifacts, allowing for the accurate reconstruction of CT images.

Advisor: Ehsan Samei, PhD

Ambient lighting in diagnostic radiology reading rooms is typically kept at a minimal level to maintain perceived image contrast.1 Under low ambient lighting conditions, however, a radiologist’s pupils will contract and dilate as the visual focus shifts between the high luminance display and the low luminance surrounding background. This pupillary action may lead to radiologist discomfort and performance degradation.2, 3 The introduction of Liquid Crystal Displays (LCDs) to reading rooms, however, has allowed a re-evaluation of optimum illuminance conditions. With high luminance ratios and low diffuse reflection coefficients, medical-grade LCDs may be calibrated to compensate for moderate levels of increased ambient lighting without any appreciable loss of perceived image contrast.4 These intrinsic properties may make it possible to maintain detection performance while increasing reading room ambient lighting so that the difference between the luminance level of the diffusely reflected light from the background surroundings (Ls) and the luminance level to which the eye adapts while reading an image (Ladp) is minimized. As a result of the decreased discrepancy between these luminance levels, pupillary action may decline, potentially leading to improved radiologist comfort and maintained or enhanced diagnostic performance.
Four studies were performed to determine the effect of increased ambient lighting on detection performance. The first study utilized the biologic contrast response of the human visual system to determine a range of clinically representative Ladp values of typical medical images. The second study examined the effect of increased ambient lighting on the detection of subtle objects embedded within a uniform background, while the third and fourth studies examined radiologist detection performance of subtle cancerous lesions in mammograms and chest radiographs under increased ambient lighting.

Advisor: Timothy G. Turkington,

The purpose of the thesis project was to develop a method to investigate the degree of misalignment of the lung/liver interface in the non-attenuation corrected PET images and in the CT image and to compare the resulting value to the visual appearance of respiration artifacts for studies with the CT acquired with an end tidal volume breath hold. The effect of factors such as lung disease and patient weight and age were also investigated. Images were acquired with a Discovery ST scanner with a LightSpeed 16 slice CT scanner. The CT was performed with an end tidal volume expiration breath hold with a rotation time of 0.5 seconds and a rotation speed of 27.5 mm/rotation. Profiles were drawn on coronal PET and CT images and the axial position of the lung/liver interface in the two images was compared. Also, the attenuation corrected PET images were placed into one of three groups based on the presence or severity of apparent artifacts: artifact present, intermediate artifact present and artifact absent. The investigated PET/CT protocol where the CT image is taken at end tidal volume expiration leads to an average non-attenuation corrected PET lung/liver interface position of 5.98 mm above the CT for the right lung and 5.72 mm above for the left lung. This distance increases for patients who have lung cancer to 6.9 mm above the CT for the right lung and 13.1 mm above for the left lung. Respiration artifacts appeared in 60% of the examined studies. The appearance of more severe respiration artifacts is correlated to higher differences in lung/liver interface location between the non-attenuation corrected PET image and the CT image. The average distance between the interface in the PET and CT image was 20.7 mm whereas the values for intermediate and absent artifacts were 6.9 and 8.0 mm respectively. In the majority of studies when the difference between lung/liver interface locations in the non-attenuation corrected PET and the CT was greater than three slices, a respiratory artifact was present. This situation occurred in 28% of the general studies examined and in 43% of the lung cancer studies examined.

Advisor: Mark Oldham

Purpose: PRESAGE is a novel radiochromic plastic material that—in combination with optical-CT imaging—has been shown to have outstanding potential for high-resolution three-dimensional (3D) dosimetry. This work explores the role and presents the first results of the application of PRESAGE/optical-CT dosimetry to commissioning a new linear accelerator for IMRT treatments. Commissioning was achieved using a modified Radiologic Physics Center (RPC) IMRT credentialing phantom to enable true 3D dosimetry.
Methods and Materials: The RPC phantom was CT scanned with the standard insert and with the modified PRESAGE insert. Using the Eclipse planning system, a 9-field IMRT treatment plan was created on the standard insert CT-scan to deliver 6.6Gy to the primary PTV and meet all other constraints of the credentialing test. A second plan was then created by recalculating the original plan on the CT scan incorporating the PRESAGE insert. The prescribed dose to the PRESAGE insert was reduced to 4Gy to avoid scanning limitations of the optical-CT system. Consistent relative fluence between the two plans was verified by MapCHECK and DynaLog file analysis. After irradiation, evaluation of the PRESAGE/optical-CT system was performed by comparing PRESAGE measurements with the independent film and TLD measurements made at the RPC.

Advisor: Mark Oldham

Purpose: To quantify the intra-fraction motion of the cervix using a novel Tool for Organ Motion AnalysiS (TOMAS). Using TOMAS cervical motion was assessed quantitatively for the exploration of custom PTV margin expansion feasibility and benefit.
Methods and Materials: Up to 501 single shot fast spin echo MRI images were acquired through the central sagittal slice of the cervix for 9 healthy volunteers and 9 cancer patients (4 patients with pelvic/abdominal malignancies and 5 patients undergoing external beam radiation therapy for cervical cancer). Images were acquired at 6-second intervals for approximately 20 minutes. For the patients with cancer, MRI scans were performed at 3 time points; pre-treatment, mid-treatment and posttreatment. Healthy volunteers were scanned once. An in depth analysis was performed in a range of techniques including anatomical point placement, contour comparison and contour profile analysis. Cervical contours were delineated on all images using in-house software, which also enabled quantification of motion and deformation. Using these contours, the extent of margin expansion to achieve 100% coverage was measured.
Results: During each intra-fraction period a general trend was observed in that the cervix moved superiorly and posteriorly with time for both the volunteers and cancer patients (up to 5 mm for volunteers and 7 mm for patients). This motion was observed to correlate with bladder filling. For patient scans acquired at different time
points, both the anterior and posterior motions increased as a function of treatment progression. To construct a customized margin expansion for the volunteers, it was found that the initial contour would need to be expanded on average by 3.8mm, 6.9mm, 6.2mm and 9.6mm in the anterior, posterior, superior and inferior directions
respectively. For the cervical cancer patients, these dimensions would be 2.4-3.7mm, 3.3- 3.9mm, 2.8-3.5mm and 1.4-3.4mm respectively and dependant on stage of treatment considered.
Conclusions: This work suggests that as treatment of the cervix progresses, significant changes in intra-fraction motion can occur. Further, each patient’s motion is unique and requires individual assessment. Motion analysis may provide guidance as to when an IMRT boost would be most successful and lead to appropriate PTV
expansion.

The project involves the ongoing collaboration with Dr. Sophie Paquerault and her research group at the FDA.  The goal of our project is to perform an inter-institution comparison of two CADe algorithms, given a mutually agreed set of testing images taken from the Digital Database of Screening Mammograms (DDSM).  We are near completion of the project, and we will soon be conducting training and testing of the algorithm in order to share the results with the FDA group.

Advisor:  Joseph Izatt, PhD

We introduce a combination of optical coherence tomography (OCT) and optical projection tomography (OPT) for computed tomography in living animals, by applying the detection advantages of Fourier-domain OCT in a transmissive geometry to extend projection imaging in highly scattering tissues to the ½-cm scale. The dramatic signal-to-noise ratio (SNR) advantage of Fourier-domain OCT over time-domain detection has enabled dramatically faster acquisition rates and densely sampled high speed 3D data acquisition in living samples. Unfortunately, higher SNR cannot be effectively traded off for deeper tissue penetration in the usual OCT geometry because multiple scattering, rather than attenuation loss, typically dominates signal detection deep in tissues and thus limits usable image depth. In a transmissive geometry, there is no ambiguity between first-arriving and multiply scattered photons, thus theoretically unlimited imaging depth is possible given enough acquisition time (i.e., SNR). In this work, we introduce Fourier domain acquisition in transillumination OCT for the first time and show how its parallel depth-scan acquisition accelerates identification of unscattered first-arriving photons for reconstruction of the interior of highly scattering objects via computed tomography.

No research project description available.

Advisor: Zhiheng Wang, Ph.D.

This research at Duke University involved the implementation of video feedback for respiratory gating treatments in radiation therapy.  This was used for patients whose tumors were located in the abdominal or thoracic cavities of the body, where a large magnitude of internal organ motion is evident.  A head mounted display, or video goggles, was worn by the patient during treatment, which displayed the same RPM (Varian) software interface used by the therapists at the linac console.  By viewing their real-time plot of breathing position versus time, self-analysis of respiratory regulation was possible, with less reliance on verbal coaching from the therapist.  This technique of video feedback was compared with using audio instructions by calculating the accuracy of breath-hold and the efficiency of treatment for each method.  Improved consistency in breath-hold and efficiency of treatment for each field of delivery was observed for patients participating in the video feedback technique.