Index of content:
Volume 36, Issue 6, June 2009
- Joint Imaging/Therapy Scientific Session: Room 303A
- Correction Strategies
36(2009); http://dx.doi.org/10.1118/1.3182683View Description Hide Description
Purpose: Equal phase and amplitude sorting methods have been commonly used for 4D CT construction. However, effect of these sorting techniques on 4D dose construction has not been explored. In this study, we investigate an optimal sorting technique for 4D dose construction. Method and Materials: An optimization model was formed using organ motion pdf and 4D dose convolution. The objective function for optimization was defined as the maximum difference between the expected 4D dose in organ of interest and the 4D dose calculated using a 4D CT sorted by a candidate sampling method. Sorting samples, as optimization variables, were selected on the respiratory motion pdf assessed during the CT scanning. Breathing curves obtained from patients' 4DCT scanning, as well as 3D dose from treatment planning, were used in the study. The equal phase and amplitude sorting methods were compared to the one optimized for each patient and the number of sorting samples varying from 2 to 20. Results: The difference in 4D dose construction decreased rapidly as the number of sorting samples increased to 6. The equal phase sorting demonstrated the largest residual error in 4D dose construction, requiring minimal 10 phases in 4D CT to maintain the dose residual less than 1% of the expected dose. The equal amplitude sorting, on the other hand, had smaller residual in general compared to the equal phase sorting when the number of sorting samples was larger than 4. Finally, the optimized sorting could achieve acceptable residuals with using 4 ∼ 5 sorting samples. Conclusion: 4D dose construction can be improved by optimizing the sorting samples. Further increase in sorting phase number above 5 may not be necessary when using the optimal sampling point.
Conflict of Interest: Support in part by Elekta Research Grant.
TH‐D‐303A‐02: Correction Strategy to Overcome Non‐Random Target Motions for Hypofractionated Spine Body Radiotherapy36(2009); http://dx.doi.org/10.1118/1.3182684View Description Hide Description
Purpose: Non‐random target motions have been reported for hypofractionated spine stereotactic body radiotherapy(SBRT), largely due to prolonged treatment delivery as compared to conventional radiotherapy. In this study, we aim to develop an adaptive correction strategy to overcome such non‐random target motions. Method and Materials: Intra‐fraction target motions of more than 200 treatment sessions were analyzed. These target motions were detected using an in‐room dual kV x‐ray imagingsystem. Non‐random target motions was characterized in six degree of freedom (DOF) that included translations (x‐, y‐, z‐ shifts) and rotations (roll, yaw, pitch). Based on the observed incidence and motion characteristics, a correction strategy based on periodic interventions (e.g., via realigning the patient or the beams) was developed in order to correct the effects of such motions to an acceptable level. The population averaged time intervals for implementing the strategy were calculated for different treatment sites that included cervical, thoracic, and lumbar‐sacral lesions. Results: Non‐random target motions were found to be present for every case studied regardless of target locations. Cervical spine targets were found to possess the highest incidences of non‐random target motions as compared to other sites (p<0.0001). The correction strategy employing periodic intervention was found to be effective in compensating the observed target motions. The average time intervals required to maintain the target motions to within 1 mm in translation or 1 degree in rotation were 5.5 min, 5.9 min, and 7.1 min for cervical, thoracic, and lumbar‐sacral lesions, respectively. Conclusion: Interventions of approximately every 5–8 minutes or less are warranted in overcoming non‐random target motions in protracted spine SBRT treatments.
TH‐D‐303A‐03: Validation of An Analytical 1D Filtering of the Dose Distribution for the Calculation of the Expected PET Distributions in Proton Therapy.36(2009); http://dx.doi.org/10.1118/1.3182685View Description Hide Description
Purpose: A non‐invasive method for verification of treatment delivery to ensure the high quality of proton therapy is offered by Positron Emission Tomography(PET), which takes advantage of the β+‐activation produced via nuclear reactions between the protons and the nuclei of the tissue during irradiation,. Since dose distributions and measured PETimages are correlated but not identical, a procedure to provide clinical feedback on the correct dose delivery and irradiation field position is necessary. This study aims to validate the measured activity patterns by means of activity distributions calculated using a novel and fast one‐dimensional filtering model recently proposed. This way the actual dose delivery can be validated without reverting to Monte Carlo simulated PET distributions. Method: We derived the analytical expressions of the filters by converting the dose into the specific isotope profiles along the penetration depth, for all the main reaction channels which yield positron emitters in biological tissue. For the application to inhomogeneous targets a dedicated MATLAB®‐based code has been developed. Results: The new filters were first applied to monoenergetic depth dose distributions at different beam energies and the results were validated against FLUKA‐MC β+isotope distributions. All resulting distributions were found in agreement with the MC distributions, confirming filter independence from the proton beam energy. The filter functions were then applied to more realistic spread out Bragg peaks, simulated in simple inhomogeneous targets consisting of PMMA, lung and bone equivalent inserts. Conclusion: Results have shown a fairly good agreement in terms of both 50% distal fall‐off position (<1mm) and absolute value between simulated depth activity profiles and filter predictions (few percent), demonstrating how a proper filtering of the MC depth dose distribution can be used to predict in a simple and fast manner all the possible β+ activations of an arbitrary target.
36(2009); http://dx.doi.org/10.1118/1.3182686View Description Hide Description
Purpose: Respiratory motion during radiotherapy can be managed by either respiratory gating (e.g. RPM by Varian) or some form of breath hold (e.g. ABC by Elekta). In respiratory gating the duty cycle is limited when a narrow window is used to reduce residual motion. For longer treatment fields, IMRT or IMAT, the breath‐hold method may take many breath holds to complete a field and each breath hold induces extra stress on the patient. We are developing a hybrid method of respiratory motion management with a relatively high duty cycle and at the same time the patient can potentially tolerate an un‐limited number of breathing cycles repeatedly with little motion during treatment delivery. Method and Materials: The Gated Breathing Synchronizer (GBS) coaches the patient to breathe with a period of breath hold in every breathing cycle. It monitors the patient with sensors such as mass‐flow sensor and respiratory belts. It can close a valve to induce breath hold which can be overridden by the patient for safety. It can also send a beam‐hold signal to synchronize the linac with the breathing pattern of the patient. We have implemented a mode of GBS to enter breath hold at end‐exhalation (EE mode). The algorithm increases the breath hold duration starting from 3 seconds per cycle until the respiration rate (RR) is reduced to a target value. It then maintains the RR at that value. Results: We have tested the EE mode of GBS on a group of healthy volunteers. Detail results on the duty cycle, RR, and the duration that the subjects can maintain the target RR will be presented. Conclusion: We have implemented the EE mode of GBS and it is well tolerated by healthy volunteers. Supported in part by NIH P01‐CA59827.
TH‐D‐303A‐05: Role of Image Guided Patient Repositioning and Online Planning in Localized Prostate Cancer IMRT36(2009); http://dx.doi.org/10.1118/1.3182687View Description Hide Description
Purpose: Compare image‐guidedIMRT for localized prostate cancer involving on‐line patient repositioning and online planning (replanning). Method and Materials: Ten early‐stage prostate cancer patients receive approximately 10 CT scans each, totaling 108 studysets. Each CT is segmented manually to identify the prostate, bladder and rectum. Using a Philips Medical, Pinnacle 8.1x RTPS,image‐guided repositioning starts with an IMRT plan to irradiate a PTV resulting from a 3‐mm margin around the prostate, on the first CT scan of each patient, which is then registered on the 2‐N, serial scans. For replanning, an IMRT plan is made on each of the serial (2‐N) CT scans using 0 and 3 mm margins. The dose distributions from scans 2‐N are then deformed to the initial CT using a mesh‐based B‐Spline deformation method, for each method. The deformed doses are added on the initial CT scan of each patient for DVH and isodose analyses. Results: Fractional volumes of rectum receiving 90 and 95% of the prescription dose (V90 and V95) range from 2–3% for 3 mm margins with repositioning and replanning, and 1–1.5% for 0 mm margins replanning. The difference in doses to rectum and bladder in repositioning and replanning with 3 mm margins are statistically insignificant. The V95 to prostate is 96.0, 97.4 and 93.8 % for repositioning with 3 mm margins, replanning with 3 mm margins and replanning with 0 mm margins, respectively. Conclusions:Image guided IMRT using 3 mm PTV margins with patient repositioning and replanning are largely comparable in target coverage and critical organ sparing, while replanning with 0 mm margins shows a statistically significant but small reduction in the doses to rectum and bladder and target coverage. Thus, a limited need exists to replan localized prostate IMRT, under image guidance.
TH‐D‐303A‐06: Automatic Image and Contour Warping Based On 3D Salient Points for Assessing the Need for Replanning in IGRT36(2009); http://dx.doi.org/10.1118/1.3182688View Description Hide Description
Purpose: To help assess the need for RT replanning by automatically warping the CTimage and patient contours from planning onto the current fraction Cone‐Beam CTimage.Method and Materials: A non‐rigid auto‐registration scheme has been developed which uses anchor interest points in images. It involves four steps: a) Extract a patient‐specific compressed model in terms of multiscale distinctive salient points from the planning CTimage, using a 3D SIFT detector adapted for both bony and soft tissue features; b) Retrieve these points in the current CBCTimage, via multiscale template‐matching maximizing local correlation; c) Derive a thin‐plate‐spline non‐rigid transformation from point pairs; d) Warp the CTimage and/or ROIs therein onto the CBCTimage. The auto‐warped CT gray‐value densities are then useful as surrogate density attenuation parameters to update the dose map according to the treatment beams planned; along with the auto‐warped delineations, this leads to up‐to‐date DVHs, helping decisions. Four patients showing significant changes through 35‐fraction head‐and‐neck treatments were selected retrospectively, with their planning ROIs and several recontoured critical/node structures in the mid‐treatment CBCT.Results: For each patient, over 1000 truly salient points were extracted and retrieved within 2 minutes; the corresponding registration map was computed and applied within five minutes. Unlike rigid alignment, the warped image and contours clearly adapt to the deformed anatomy highlighted by the mid‐treatment CBCT, typically neck shrinking, node shifting, and spine flexions. This reveals e.g. that a gross node coverage planned at mean dose 2.1Gy/fraction can decrease to less than 1.8Gy/fraction at fraction 35. Conclusion:Image and ROI warping based on salient points is feasible, and recommendable for updated dosimetry checks. In replanning events, the delineations warped to a newly acquired CT may provide a starting point to support time‐efficient re‐contouring. Conflict of Interest: Research sponsored in part by Philips Healthcare corporation.
TH‐D‐303A‐07: Dosimetric Impact of Rotational Setup Error in Stereotactic Body Frame Radiation Therapy (SBRT)36(2009); http://dx.doi.org/10.1118/1.3182690View Description Hide Description
Purpose:SBRT uses specialized fixation devices to achieve high dose gradient with minimal margin, however setup errors cannot be totally eliminated. The dosimetric impact of rotational setup error is investigated in the context if translational shifts can compensate for it. Methods and Materials: To simulate patient rotational setup errors, CTimages of a phantom and actual clinical SBRT cases (liver: n=3, lung: n=3) were rotated around the body center using an in‐house image processing toolkit. The dosimetric impact of uncorrected roll was quantified by comparing the recalculated dose distribution to the original plan. Manual translational registration was then performed to match the target volumes on the rotated images to the original datasets to simulate translational correction. The original plan was then recalculated using the corrected CT datasets and the resultant dose distributions compared to the original ones. Results: The simulated rotation of phantom results in reduced dose‐volume coverage of CTV and PTV. While minimal for rotations <3 degrees, CTV coverage decreased sharply for larger rotation. For all clinical cases, a 3‐degree rotation resulted in less than 5% reduction in the CTV volume covered by 90% prescribed dose (CTV90). A larger rotation led to significant dose reduction in tumor targets as well as dose changes to critical organs (OAR). Manual translational registration resulted in good recovery of both CTV90 and PTV80. However, increased dose to OAR was observed in some cases where the selection of beam angle was close to the OAR. Conclusion: It is concluded that uncorrected rotational setup error larger than 3 degree could cause significant dose changes to tumor targets and OAR. Image guided translational correction can compensate for rotational setup errors. Despite the improvement of CTV matching, caution needs to be paid regarding dose increase to OAR when large rotation angle is corrected.
36(2009); http://dx.doi.org/10.1118/1.3182691View Description Hide Description
Purpose: To present a method to evaluate dose uncertainties introduced by using deformable image registration to map dose between patient poses. Methods and Materials: By definition, dose is the ratio between energy deposited E and mass M, i.e. D = E / M . The dose uncertainty is defined as D D = (¶ D / ¶ E ) D E + (¶ D / ¶ M ) D M . When dose is mapped from patient pose 1 to pose 2, this expression can be evaluated as 2 2 1 1 1 1 2 1 D D = ( E ‐ E ) / M + E ( M ‐ M ) / M where the E's and M's are evaluated over finite volumes. Practical evaluation of the dose uncertainty can be accomplished by defining multiple arbitrary finite volumes via contouring on pose 1, then deformably mapping each volume to pose 2. In each volume, the integral dose E and mass M is evaluated to allow determination of D D . The method is applied to a test patient case to demonstrate its implementation. Conclusion: A practical method has been developed to evaluate dose uncertainty in volumes deformably mapped between differing patient poses. For the test case, the average dose uncertainty in the mapped volumes is 2.3%, with a maximum dose uncertainty of 10%. Results will be dependent on the quality of the image registration, the dose mapping method, and the dose gradients. (Work supported by NIH P01CA116602).
36(2009); http://dx.doi.org/10.1118/1.3182692View Description Hide Description
Purpose: Establish the setup uncertainty (SU) for intracranial pediatric radiotherapy patients based on daily pre‐treatment CBCT and quantify the residual uncertainty (RU) based on post‐treatment CBCT. Also, quantify the inter‐observer uncertainty (OU) and mechanical uncertainty (MechU) of the CBCT system. Methods and Materials: 94 intracranial pediatric patients have completed therapy under an IRB approved localization protocol, and stratified based on treatment position (Supine/Prone) and the use of general anesthesia (GA). Patients received a localization CBCT prior to each fraction and at the end of every other fraction. The CBCT used was an investigational MV imaging beam line; the output was adjusted such that the patient isocenter dose was 1cGy per CBCT. The offset based on the registration of the pre‐fraction CBCT to the Sim‐CT, comprising the SU, were recorded in an electronic database and the patient moved into the correct position. The offset in the post‐fraction CBCT to the Sim‐CT comprised the RU. Nine individuals independently registered the first five CBCTs of the same five randomly chosen patients. The comparison established the inter‐observer uncertainty (OU). The mechanical uncertainty (MechU) was determined by acquiring a CBCT of a localization phantom monthly and recording the discrepancy between the known positions of landmarks within the phantom and the positions based on the CBCT.Results: The mean age of the 94 patients was 12.7±7.3 years. 69 were treated supine and 25 prone; 41 with GA and 53 without. The various uncertainties in mm were as follows: SU_Supine=3.5, SU_Prone=3.8, SU_GA=3.7, SU_noGA=3.6; RU_Supine=1.8, RU_Prone=2.4, RU_GA=1.5, RU_noGA=2.3. The OU was 0.9 and the MechU was 0.5 mm. Conclusion: The 1cGy CBCTs produced with the investigational imaging beam line can be used to confidently localize the patient before treatment and thereby reduce the setup uncertainty.
Conflict of Interest: Supported in part by Siemens Medical USA.
- Innovations and Frontiers in Medical Physics
WE‐D‐303A‐01: Feasibility Study of Microbeam Radiation Therapy Using a Carbon Nanotube Field Emission Based Electron Microbeam Irradiator36(2009); http://dx.doi.org/10.1118/1.3182528View Description Hide Description
Purpose: Microbeam radiation therapy(MRT) is an innovative radiation therapy treatment method. It has unique capability of eliminating tumor while sparing normal tissue. Currently all the MRT studies are carried out using synchrotron based radiation facility, which in general has limited accessibility. We have recently developed a carbon nanotube(CNT)field emission based electron microbeam irradiation system. It has the potential of providing extremely high radiationdose rate (>100 Gy/sec) with high spatial and temporal beam resolution. Method and Materials: The electron microbeam irradiator is based on carbon nanotubeelectron field emission technology. The system consists of a triode type CNTfield emissioncathode sealed in vacuum. The cathode current is controlled by an adjustable gate electrode, while a high voltage power supply is used to create an acceleration potential between the cathode and anode. A laser drilled aperture is used as the electron beam exit window and also provides required electron beam collimation. Results: The CNT microbeam irradiator is capable of delivering electron radiation at extremely high dose rate (>100 Gy/sec). The beam profile was recorded using Gafchromic film (HD‐810) and showed beam FWHM (full‐width‐at‐half‐maximum) about 62 μm. The measured PVDR (peak‐to‐valley dose ratio) is about 15:1, which is very reasonable for MRT application and can be further improved with better beam collimation. SW480 colorectal cancer cells were also irradiated using the same device for demonstration purpose. Conclusion: We have demonstrated the feasibility of generating high dose rate electron microbeam radiation using a carbon nanotubefield emission based electron microbeam irradiator. The compact size and low cost of the device will make the device more accessible for cancer research community. The preliminary study using the as‐developed electron microbeam system showed its promising future for potential MRT related research, especially at cellular level.
WE‐D‐303A‐02: Deoxyglucose Labeled Gold Nanoparticles as X‐Ray Computed Tomography Contrast Agents for Cancer Imaging36(2009); http://dx.doi.org/10.1118/1.3182529View Description Hide Description
Purpose: To study the feasibility of using 2‐Deoxy‐D‐Glucose (2‐DG) labeled gold nanoparticle (AuNP‐DG) as a metabolic functional CTcontrast agent through in vitro experiments. Method and Materials: The goldnanoparticles (AuNP) were fabricated using a citrate acid reduction method. The size of the AuNP was determined from Transmission Electron Microscopy images to be 4 nm in diameter. The conjugation of the 2‐DG with the AuNP core was accomplished using mercapto group in the 2‐carbon position by condensation reaction of 2‐amino‐deoxyglucose with mercaptosuccinic acid. The human alveolar epithelial cancer cell line, A‐549, was chosen for the in vitro cellular uptake assay. Two groups of cell samples (∼1 × 105 cells per sample) were incubated with the AuNP‐DG and the unlabeled AuNP, respectively, for 30 minutes (37°C, 7% CO2). Following the incubation, the cells were washed with sterile PBS six times to remove the excess goldnanoparticles. The cells were then spun to cell pellets using a centrifuge. The cell pellets were imaged using a microCT scanner immediately after the centrifuging (75 kVp, 135 μA, 1184 × 1120 matrix size, 360 views, averaging 5 frames per view). The reconstructedCTimages were analyzed using a commercial software package. Results: Significant contrast enhancement in the cell samples incubated with the AuNP‐DG with respect to the cell samples incubated with the unlabeled AuNP was observed in multiple CT slices. Quantitative analysis of the image data showed that (45.6 ± 14.2)% of the cells that were incubated with the AuNP‐DG exhibit enhanced contrast compared to the cells in the control group (incubated with AuNP). Conclusion: Results from these experiments strongly suggest enhanced uptake of the AuNP‐DG over the unlabeled AuNP by the highly glycolytic cancer cells in vitro and indicate that AuNP‐DG could served as a metabolic functional CTcontrast agent with tumor‐specific targeting capability.
WE‐D‐303A‐03: Magnetic Resonance Temperature Imaging Guided Laser‐Induced Thermal Therapy with Multi‐Walled Carbon Nanotubes36(2009); http://dx.doi.org/10.1118/1.3182530View Description Hide Description
Purpose: Feasibility has been studied for combining Multi‐Walled Carbon Nanotubes(MWNTs), as a near‐infrared (NIR) laser absorber and heat generator, with ProtonResonance Frequency (PRF) based Magnetic Resonance Temperature Imaging (MRTI), to improve the safety and efficacy of laser‐induced thermal therapy. Method and Materials: MWNTs‐laser‐induced therapy was evaluated using 3 tissue equivalent gel phantoms: alginate‐only (sodium alginate 3g/L), MWNTs‐instilled (0.5mg/ml) and MWNTs‐implanted (sub‐surface, simulating a subcutaneous tumor containing MWNTs).In vivo experiments used 4 RENCA kidneytumor bearing mice in their right flank, thermally treated using an external laser beam after direct MWNTs injection (100ug), and monitored by MRTI throughout the treatment. MRTI‐guided laser‐induced thermal therapy for phantom and in vivo experiments was performed using MR‐compatible laser systems (fiber‐optic and external laser beam) and a 7T MRI small animal scanner (Bruker Biospin). The 3‐D MRTI in vivo protocol has a standard deviation of < 1°C, temporal resolution of 4.2s and a high spatial resolution of 0.25mm. Results: With minimum‐invasive fiber‐optic laser heating (ø 0.6mm, 1 min @ 0.1W), phantom results show that the MWNTs—instilled phantom heated preferentially (from 20°C to 47°C; Δt=+27°C), compared to the alginated‐only phantom (from 20°C to 25°C; Δt=+5°C). With non‐invasive laser heating (ø 10mm, 0.5 min @ 1.8W), the implanted region of the MWNTs‐implanted phantom showed significantly elevated temperatures compared to the nearby alginate‐only medium (Δt=+15°C). Implanted RENCA kidney flank tumors in mice injected with MWNTs were heated to 77°C (Δt=+51°C) after a single 30s 3W non‐invasive laser irradiation compared to 44°C in the laser‐only tumors (no MWNTs Δt=+18°C). In two weeks post‐treatment study, all control tumors (no MWNT no laser) and laser‐only tumors keep growing while MWNTs‐tumors begin to shrink. Conclusion: This study indicates the significant improvement in small animal laser‐induced thermal therapy and may be applicable for superficial tumors in humans.
WE‐D‐303A‐04: Mircometer‐Sized Iron Oxide Particles (MPIO) Enhanced MRI with Granulocyte‐Colony Stimulating Factor (GCSF) Modulation in Murine Myocardial Infarction Model36(2009); http://dx.doi.org/10.1118/1.3182531View Description Hide Description
Purpose: To monitor the MPIO and enhanced green fluorescenceprotein (eGFP) labeled mesenchymal stem cells (MSCs) infiltration into the myocardial infarction (MI) site using ‐weighted MRI; To monitor the MRI contrast around the MI site post‐GCSF modulation. Methods: C57Bl/6 male mice (6–8 weeks old) were irradiated with an 8‐Gy dose. The labeled MSCs (3–7×105) were transplanted into the tibial medullary space 2 days post‐irradiation. The mice were divided into: 1) a sham‐operated group (Sham, n=7); 2) a MI group without GCSF injection (MI‐GCSF, n=7); and 3) a MI group with GCSF treatment (MI+GCSF, n=3). At 14 days post‐labeled MSCs transplantation, the two MI groups underwent surgery via permanent ligation of the left anterior descending coronary artery while the Sham group underwent open‐chest operation without perturbing the heart. The MI+GCSF group received subcutaneous GCSF injection 1 day post‐MI to enhance MSC mobilization. ‐weighted short‐axis cardiac MRI was performed at baseline, 3, 7 and 14 days (D14) post‐surgery. Results: The MRI signal at the MI site was temporally attenuated for both MI groups, with more attenuated for MI+GCSF group (SNR 18.17±6.06 vs 11.37±1.01 at D14, p<0.05), but not for Sham group (30.63±5.69). The MI+GCSF group showed a trend of cardiac function improvement relative to MI‐GCSF group (left ventricular ejection function 45.55±7.52% vs 40.80±16.69% at D14), but it is insignificant possibly due to the small sample number. Dual‐labeled cells were fluorescently detected around the infarction site. Conclusions: Migration of MPIO‐labeled MSCs from bone marrow into the injured heart can be temporally monitored by MRI and additional signal attenuation caused by GCSF treatment can be differentiated. Results of this study suggest a potential approach in cell therapy to noninvasively monitor migration of labeled cells as well as the mobilization modulation produced by pharmaceuticals in the MI related events.
WE‐D‐303A‐05: How to Measure Fluoroscopic “dose Efficiency”: The Spatio‐Temporal Detective Quantum Efficiency36(2009); http://dx.doi.org/10.1118/1.3182532View Description Hide Description
Purpose: To measure the “dose efficiency” of fluoroscopic systems on a quantitative and absolute scale by developing a spatio‐temporal detective quantum efficiency (DQE) metric of performance. Method and Materials: We developed the first comprehensive spatio‐temporal approach to fluoroscopic system performance through the development of a spatio‐temporal DQE. It is defined in terms of the presampling spatio‐temporal modulation transfer function(MTF), describing resolution, and spatio‐temporal noise power spectrum (NPS), describing noise. We developed what we call the “semi‐transparent moving slanted‐edge method” to measure the temporal component of the MTF, which uses a small‐signal approach to temporal resolution. This method overcomes temporal non‐linearity problems associated with differences between lag build‐up and decay responses of CsI‐based systems. A three‐dimensional (3D) approach to fluoroscopic noise, in which 3D spatio‐temporal ROIs are selected in space and time, was used to calculate the spatio‐temporal NPS. Spatio‐temporal DQE values of a bench‐top x‐ray image intensifier fluoroscopic system were calculated, illustrating use of the metric and gaining insight into fluoroscopic temporal dose efficiency. Results: Non‐ideal temporal performance was noted in the test system at temporal frequencies > 5 Hz, and there was a ∼50% drop in DQE values by 15 Hz, analogous to decreases in DQE values commonly seen at large spatial frequencies. This is the first time that fluoroscopic temporal performance has been quantized using a DQE metric, and the non‐ideal system temporal performance may be due to temporal noise aliasing and the statistical nature of lag. Conclusion: The spatio‐temporal DQE describes fluoroscopic system performance and dose efficiency in both space and time, providing temporal information that can not be obtained using lag‐corrected or other spatial approaches. The ∼50% decrease in performance by 15 Hz may suggest the presence of temporal quantum sinks.
36(2009); http://dx.doi.org/10.1118/1.3182533View Description Hide Description
Purpose: Accurate dosimetry is critical for effective management of persons exposed to radiation; in many cases this can only be achieved by biodosimetry, measuring the change in a biological parameter correlated to dose received. This study evaluates the suitability of FLT PETimaging as a biodosimeter by investigating the correlation between absorbed dose and FLT uptake in bone marrow in patients undergoing fractionated radiotherapy.Methods and Materials: Patients received a series of two PET/CT scans usingFLT (a cellularproliferation marker), one prior to the start of radiotherapy and another later in the treatment course (2–3 weeks). The ratio of the bone marrow SUVs of the later scan to the earlier scan was calculated, and the result was used as the surviving fraction in the linear quadratic formalism. A constant α/β ratio of 10 Gy was assumed. Alpha values were calculated for individual bones on a voxel‐by‐voxel basis, using dose information from the treatment plan. Inter‐patient and intra‐patient comparisons of the results were made, and the relationship between α and received dose was investigated. Results: High FLT uptake in bone marrow was observed for all patients in pre‐treatment scans, and a negative correlation was observed between dose received and change in bone marrow FLT uptake. Interestingly alpha was observed to decrease with dose. The decrease in α occurs exponentially, with the extrapolation value at zero of 0.15Gy−1 and the exponential dose decay constant of 0.5 Gy−1. Intra‐patient variability suggests that the calculation is sensitive to the image registration techniques employed. Conclusions: FLT‐PET imaging shows great potential for use as a biodosimeter; these results suggest that a correlation exists between change in FLT uptake and absorbed dose. Further study is needed to assess the sensitivity of these calculations both to setup errors and to different image processing techniques.
WE‐D‐303A‐07: A 3D System for Spatial Localization of Scanned Proton Beams: Characterization and Experimental Validation36(2009); http://dx.doi.org/10.1118/1.3182534View Description Hide Description
Purpose: Intensity‐modulated proton therapy (IMPT) using scanned proton beamlets relies on the delivery of a large numbers of beamlets to shape the dose distribution in a highly conformal manner. We have developed a 3D system based on liquid scintillator to evaluate the spatial localization, intensity and depth of penetration (energy) of the proton beamlets in near real‐time. Method and Materials: The liquid scintillation (LS) detectorsystem consists of a volume of Liquid Scintillator (20×20×23 cc) in a light tight enclosure connected to a CCDcamera. This camera has a field of view of 25.7 cm by 19.3 cm and a pixel size of 0.4 mm. While the LS is irradiated, the camera continuously acquire images of the light distribution produced inside the LS. Irradiations were conducted with proton pencil beams produced with a spot‐scanning nozzle. Pencil beams with nominal range in water between 9.5 cm and 17.6 cm were irradiated in a square area 10 cm wide. Image frames where acquired at 50 ms per frame. Results: The signal to noise ratio of a typical Bragg peak was about 170. Proton range measured from the light distribution produced in the LS was accurate within 0.3 mm on average. The largest deviation seen between the nominal and measured range was 0.6 mm. Lateral position of the measured pencil beam was accurate within 0.4 mm on average. The largest deviation seen between the nominal and measured lateral position was 0.8 mm. However, the accuracy of the measurement of lateral beam position could be improved by correcting light scattering artifacts. Conclusion: Our LS detectorsystem has been shown to be capable of fast, sub‐millimeter spatial localization of proton spots delivered in a 3D volume. This system could be used for quality assurance of IMPT.
Supported partly by the NCI (1R01CA120198‐01A2) and (2P01CA021239‐29A1)
36(2009); http://dx.doi.org/10.1118/1.3182535View Description Hide Description
Purpose: The most commonly used NTCP and TCP models are acknowledged to be over simplified. The typical approach to determining if a model ‘fits’ the data is to look at residuals of predicted rates, but this is not straightforward in multi‐dimensional modeling problems because data are sparse. It is therefore crucial to develop methods that investigate the goodness of fit of the models over the dataspace. The goal of this work is to investigate a supervised machine‐learning tool for producing contour plots of outcome probability, as a reference against simpler models.Method and Materials: For demonstrative purposes, we used an institutional dataset consisting of 281 patients treated for non‐small cell lungcancer. Two endpoints were investigated: radiation‐induced pneumonitis (n=219), and local control (n=56). Clinical and dosimetric variables were extracted using CERR and logistic regression model building was performed in the DREES software system. Kernel‐based support vector machines with radial basis functions (SVM‐RBF) were used to produce corresponding contour maps of risk levels. Results: The SVM projection plots identified four possible regions based on the risk group and the confidence level: (i) a region of low risk patients with high confidence prediction level; (ii) a region of low risk patients with lower confidence prediction level; (iii) a region of high risk patients with lower confidence prediction level, and (iv) a region of high risk patients with lower confidence prediction level. The SVM‐RBF produces nonlinear contours that differ significantly from the logistic regression contours, with some agreement where the data is most dense. Conclusion: SVM models provide a powerful, automated, tool to help understand NTCP and TCP model fits. We plan to integrate SVM methods within the open‐source DREES modeling system.
Partially supported by K25 CA 128809 and R01 CA85181.
36(2009); http://dx.doi.org/10.1118/1.3182536View Description Hide Description
Purpose: To develop a compact and powerful FPGA‐based dose calculation system to meet the performance requirement of emerging radiation therapy technologies. Method and Materials: One challenge for advanced radiation therapy technologies such as intensity‐modulated arc therapy (IMAT) is the speed of dose calculation since a large number of beams are used to approximate continuous rotational delivery, resulting in significantly increased dose calculation workload. To accelerate this process, we built a multi‐FPGA (field‐programmable gate array) accelerator with 8 independent computing channels, which can be plug into a PCI slot of any computer to serve as a coprocessor to the CPU and turn it into a powerful dose calculation workstation. The algorithm we implemented is the popular collapsed‐cone convolution/superposition (CCCS) algorithm. After receiving the total energy released per unit mass (TERMA), density and necessary geometrical information from the PCI bus, the accelerator will work on its own to do transport‐line generation, raytracing, kernel tilting, and dose deposition. Results: With the support of memory‐rich multi‐FPGA platform, our floating‐point‐based design working at 90Mhz provides a speedup of 20X over the commercial multi‐threaded software on a state‐of‐the‐art quad‐core system. The FPGA architecture was optimized in design to achieve the best performance specifically for the CCCS algorithm with the minimum resources as compared with a general purpose FPGA board. Our modular approach can also be easily expanded to achieve even greater speedup for cases where the computation time is still an issue by inserting additional such FPGA boards into the computer.Conclusion: We developed a complete dose calculation system powered by a multi‐FPGA plug‐in board, which is capable of delivering very fast dose calculation based on the collapsed‐cone convolution/superposition algorithm. Experimental studies demonstrated the potentials of our new approach to dynamic radiation delivery technologies such as IMAT that demand tremendous computation power for dose calculation.
- Modeling of Intrafraction Motion
TU‐C‐303A‐01: Evaluating the Accuracy of Four‐Dimensional Photon Dose Calculations with Phantom Measurements36(2009); http://dx.doi.org/10.1118/1.3182338View Description Hide Description
Purpose: Respiratory motion can cause deviations between the intended and delivered dose distributions. Recent work has focused on developing four‐dimensional (4D) dose calculation algorithms, which explicitly account for respiration in the dose calculation process. Before these dose calculations methods can be used clinically, it is necessary to verify their accuracy. The purpose of this study was to evaluate the accuracy of 4D dose calculations with phantom measurements. Methods: Measurements were made using two anthropomorphic phantoms: a rigid moving phantom and a deformable phantom. Two motion patterns were designed to drive both phantoms: a sinusoidal motion pattern and an irregular motion pattern that was extracted from a patient breathing profile. Three plans were generated on each phantom: a single‐beam, a multiple‐beam, and an IMRT plan. Doses were calculated using the 4D dose calculation capabilities of a commercial radiation treatment planning system. Each plan was used to irradiate the phantoms, and doses were measured using TLD and radiochromic film. The measured doses were compared to the 4D‐calculated doses using a measured‐to‐calculated TLD ratio and a gamma analysis. For the TLD and film, relevant passing criteria (5% for TLD and 5%/3mm for gamma) were applied to determine if the 4D dose calculations were accurate to within clinically acceptable standards. Results: All of the TLD measurements met the passing criteria. 42 out of the 48 evaluated films passed the gamma criteria. The films that did not pass the gamma criteria were from the irregular moving rigid phantom. Conclusions: In controlled conditions, 4D dose calculations are accurate to within clinically acceptable standards. In clinical terms, this means that if patient breathing is reproducible, 4D dose calculations will produce accurate dose distributions. Conversely, irregular breathing can produce inaccurately calculated 4D dose distributions. Conflict of Interest: This work is partially supported through an SRA with Philips Healthcare.
TU‐C‐303A‐02: Intrafraction Prostate Motion Monitoring with Cine‐MV and Minimal As‐Needed Onboard KV Imaging36(2009); http://dx.doi.org/10.1118/1.3182339View Description Hide Description
Purpose: To examine the feasibility and performance of using treatment MV beam imaging with prior knowledge to estimate 3D prostate intrafraction motion and to instantaneously reposition based on information from minimal usage of on‐board kV imaging during IMRT.Methods and Materials: In contrast with current motion monitoring techniques which seek to accurately and continuously localize a moving target, we attempted for the first step only to detect potential motion beyond a pre‐defined threshold using MV images and in the second step through combined MV‐kV imaging (by turning on the kV imager) to confirm the over‐threshold event as well as obtaining accurate position information which could be used for instantaneously repositioning. EPIDimages were used to measure 2D prostate displacement in the imaging plane. By taking into account the strong correlation between prostate SI and AP motion, small displacement in the LR direction, and data from a previous IMRT session (different gantry angle), we estimated in‐line prostate motion as well which in turn allowed us to detect potential 3D over‐threshold motion. To minimize the influence from training data, no statistical motionprobability distribution information was used. Simulation has been done using 536 patient‐measured trajectories from 17 patients. Experiments were performed on a Varian Trilogy linac using a motion phantom programmed for selected typical trajectories, and the results were compared with simulations. Results: Prostate displacement beyond a set threshold (3mm) was detected for over 99% of the time at the cost of negligible kV dose (< 3 images/fraction on average). The position information required for repositioning was found to have sub‐millimeter accuracy using combined MV‐kV data. Conclusion: Significant reduction of adverse effects of intrafraction prostate motion is achievable. The technique can be readily implemented in clinics and incurs minimal imaging dose to the patient as compared with other stereoscopic imaging techniques.