A Kumarasiri

Evaluation of a magnetic resonance guided linear accelerator for stereotactic radiosurgery treatment

INTRODUCTION The purpose of this study was to investigate the systematic localization accuracy, treatment planning capability, and delivery accuracy of an integrated magnetic resonance imaging guided Linear Accelerator (MR-Linac) platform for stereotactic radiosurgery. MATERIALS AND METHODS The phantom for the end-to-end test comprises three different compartments: a rectangular MR/CT target phantom, a Winston-Lutz cube, and a rectangular MR/CT isocenter phantom. Hidden target tests were performed at gantry angles of 0, 90, 180, and 270 degrees to quantify the systematic accuracy. Five patient plans with a total of eleven lesions were used to evaluate the dosimetric accuracy. Single-isocenter IMRT treatment plans using 10-15 coplanar beams were generated to treat the multiple metastases. RESULTS The end-to-end localization accuracy of the system was 1.0 ± 0.1 mm. The conformity index, homogeneity index and gradient index of the plans were 1.26 ± 0.22, 1.22 ± 0.10, and 5.38 ± 1.44, respectively. The average absolute point dose difference between measured and calculated dose was 1.64 ± 1.90%, and the mean percentage of points passing the 3%/1 mm gamma criteria was 96.87%. CONCLUSIONS Our experience demonstrates that excellent plan quality and delivery accuracy was achievable on the MR-Linac for treating multiple brain metastases with a single isocenter.

SU‐C‐17A‐03: Evaluation of Deformable Image Registration Methods Between MRI and CT for Prostate Cancer Radiotherapy

PURPOSE We evaluated the performance of two commercially available and one open source B-Spline deformable image registration (DIR) algorithms between T2-weighted MRI and treatment planning CT using the DICE indices. METHODS CT simulation (CT-SIM) and MR simulation (MR-SIM) for four prostate cancer patients were conducted on the same day using the same setup and immobilization devices. CT images (120 kVp, 500 mAs, voxel size = 1.1×1.1×3.0 mm3) were acquired using an open-bore CT scanner. T2-weighted Turbo Spine Echo (T2W-TSE) images (TE/TR/α = 80/4560 ms/90°, voxel size = 0.7×0.7×2.5 mm3) were scanned on a 1.0T high field open MR-SIM. Prostates, seminal vesicles, rectum and bladders were delineated on both T2W-TSE and CT images by the attending physician. T2W-TSE images were registered to CT images using three DIR algorithms, SmartAdapt (Varian), Velocity AI (Velocity) and Elastix (Klein et al 2010) and contours were propagated. DIR results were evaluated quantitatively or qualitatively by image comparison and calculating organ DICE indices. RESULTS Significant differences in the contours of prostate and seminal vesicles were observed between MR and CT. On average, volume changes of the propagated contours were 5%, 2%, 160% and 8% for the prostate, seminal vesicles, bladder and rectum respectively. Corresponding mean DICE indices were 0.7, 0.5, 0.8, and 0.7. The intraclass correlation coefficient (ICC) was 0.9 among three algorithms for the Dice indices. CONCLUSION Three DIR algorithms for CT/MR registration yielded similar results for organ propagation. Due to the different soft tissue contrasts between MRI and CT, organ delineation of prostate and SVs varied significantly, thus efforts to develop other DIR evaluation metrics are warranted. CONFLICT OF INTEREST Submitting institution has research agreements with Varian Medical System and Philips Healthcare.

SU‐E‐J‐206: Delivered Dose to Organs From CBCT‐Based IGRT of the Prostate

Purpose: To estimate the actual delivered dose to organs from CBCT‐based image‐guided radiation therapy for localized prostate cancer treatments. Methods: Seven localized prostate cancer treatments were retrospectively selected, each with ∼40 CBCT images. All patients were treated with 9 IMRT beams. The PTV margin was 6 mm in the posterior direction, and 10 mm otherwise around the prostate. For each patient, the original plan was used to calculate treatment dose on each CBCT image at treatment position. The calculated daily dose was then warped back to the original planning CT via an intensity‐based B‐spline deformable image registration (DIR) algorithm. The DIR algorithm employed mutual information with a multi‐resolution and multi‐stage scheme. From the transferred total cumulative dose, dosimetric parameters were recorded for comparison with the original plan. The quality of prostate registration was quantified using three implanted fiducial markers. Only those cases with < 2mm error were included in the dose analysis. Results: The overall marker displacement was 1.8±1.8 mm after registrations. 75 of 281 registrations (27%) exceeded the 2 mm error threshold. The primary causes were the excessive noise level on CBCT images especially for large patients, presence of rectum gas, and lower abdominal motion artifacts. Excluding these cases, a total of 206 CBCT images were included in dosimetric analysis. Changes in dose coverage to the prostate was minimal (<1%). Dose deviations for rectum (Dmax, V50, V65, V75) were (0.5±1.0%, 2.6±2.5%, 3.5±3.7%, 2.3±4.2%) and (−1.1−0.9%, −2.5±5.3%, −3.0±5.3%, −3.8±4.9%) for bladder, respectively. Conclusion: With the conventional PTV margins, deviations in dose coverage of the prostate in the face of daily anatomic changes were minor However, relatively larger deviations were observed for rectum and bladder (1SD=∼4.3%). Reduced margins may lower dose to critical organs while maintaining target coverage. However, margin reduction must be viewed cautiously and with consideration of clinical outcomes.

Changes in pharyngeal constrictor volumes during head and neck radiation therapy: Implications for dose delivery

Objective: The objective of this study was to evaluate the anatomical changes and associated dosimetric consequences to pharyngeal constrictor muscles (PCMs) that occur during head and neck (H and N) radiotherapy (RT). Materials and Methods: A cohort of 13 oropharyngeal cancer patients with daily cone beam computed tomography (CBCT) was retrospectively studied. On every 5th CBCT image, PCM was manually delineated by a radiation oncologist. The anterior-posterior PCM thickness was measured at the midline level of C3 vertebral body. Delivered dose to PCM was estimated by calculating dose on daily images and performing dose accumulation on corresponding planning CT images using a parameter-optimized B-spline-based deformable image registration algorithm. The mean and maximum delivered dose (Dmean, Dmax) to PCM were determined and compared with the corresponding planned quantities. Results: The average (±standard deviation) volume increase (ΔV) and thickness increase (Δt) over the course of 35 total fractions were 54 ± 33% (11.9 ± 7.6 cc) and 63 ± 39% (2.9 ± 1.9 mm), respectively. The resultant cumulative mean dose increase from planned dose to PCM (ΔDmean) was 1.4 ± 1.3% (0.9 ± 0.8 Gy), while the maximum dose increase (ΔDmax) was 0.0 ± 1.6% (0.0 ± 1.1 Gy). Patients who underwent adaptive replanning (n = 6) showed a smaller mean dose increase than those without (n = 7); 0.5 ± 0.2% (0.3 ± 0.1 Gy) versus 2.2 ± 1.4% (1.4 ± 0.9 Gy). There were statistically significant (P = 0.001) strong correlations between ΔDmean and Δt (Pearson coefficient r = 0.78), as well as between ΔDmean and ΔV (r = 0.52). Conclusion: The patients underwent considerable anatomical changes to PCM during H and N RT. However, the resultant increase in dose to PCM was minor to moderate. PCM thickness measured at C3 level is a good predictor for the mean dose increase to PCM.

SU-F-J-68: Deformable Dose Accumulation for Voxel-Based Dose Tracking of PTV Cold Spots for Adaptive Radiotherapy of the Head and Neck

PURPOSE To utilize deformable dose accumulation (DDA) to determine how cold spots within the PTV change over the course of fractionated head and neck (H&N) radiotherapy. METHODS Voxel-based dose was tracked using a DDA platform. The DDA process consisted of B-spline-based deformable image registration (DIR) and dose accumulation between planning CTs and daily cone-beam CTs for 10 H&N cancer patients. Cold spots within the PTV (regions receiving less than the prescription, 70 Gy) were contoured on the cumulative dose distribution. These cold spots were mapped to each fraction, starting from the first fraction to determine how they changed. Spatial correlation between cold spot regions over each fraction, relative to the last fraction, was computed using the Jaccard index Jk (Mk,N), where N is the cold spot within the PTV at the end of the treatment, and Mk the same region for fraction k. RESULTS Figure 1 shows good spatial correlation between cold spots, and highlights expansion of the cold spot region over the course of treatment, as a result of setup uncertainties, and anatomical changes. Figure 2 shows a plot of Jk versus fraction number k averaged over 10 patients. This confirms the good spatial correlation between cold spots over the course of treatment. On average, Jk reaches ∼90% at fraction 22, suggesting that possible intervention (e.g. reoptimization) may mitigate the cold spot region. The cold spot, D99, averaged over 10 patients corresponded to a dose of ∼65 Gy, relative to the prescription dose of 70 Gy. CONCLUSION DDA-based tracking provides spatial dose information, which can be used to monitor dose in different regions of the treatment plan, thereby enabling appropriate mid-treatment interventions. This work is supported in part by Varian Medical Systems, Palo Alto, CA.

SU-F-R-41: Regularized PCA Can Model Treatment-Related Changes in Head and Neck Patients Using Daily CBCTs

PURPOSE To use daily cone beam CTs (CBCTs) to develop regularized principal component analysis (PCA) models of anatomical changes in head and neck (H&N) patients, to guide replanning decisions in adaptive radiation therapy (ART). METHODS Known deformations were applied to planning CT (pCT) images of 10 H&N patients to model several different systematic anatomical changes. A Pinnacle plugin was used to interpolate systematic changes over 35 fractions, generating a set of 35 synthetic CTs for each patient. Deformation vector fields (DVFs) were acquired between the pCT and synthetic CTs and random fraction-to-fraction changes were superimposed on the DVFs. Standard non-regularized and regularized patient-specific PCA models were built using the DVFs. The ability of PCA to extract the known deformations was quantified. PCA models were also generated from clinical CBCTs, for which the deformations and DVFs were not known. It was hypothesized that resulting eigenvectors/eigenfunctions with largest eigenvalues represent the major anatomical deformations during the course of treatment. RESULTS As demonstrated with quantitative results in the supporting document regularized PCA is more successful than standard PCA at capturing systematic changes early in the treatment. Regularized PCA is able to detect smaller systematic changes against the background of random fraction-to-fraction changes. To be successful at guiding ART, regularized PCA should be coupled with models of when anatomical changes occur: early, late or throughout the treatment course. CONCLUSION The leading eigenvector/eigenfunction from the both PCA approaches can tentatively be identified as a major systematic change during radiotherapy course when systematic changes are large enough with respect to random fraction-to-fraction changes. In all cases the regularized PCA approach appears to be more reliable at capturing systematic changes, enabling dosimetric consequences to be projected once trends are established early in the treatment course. This work is supported in part by a grant from Varian Medical Systems, Palo Alto, CA.

An automatic dosimetric and geometric tracking system for head and neck adaptive radiotherapy

The ability to track changes in tumors and organs at risk over a course of radiotherapy is essential for successful adaptive radiation therapy (ART). To that end, we have developed a software suite that helps with tracking geometric and dosimetric changes of organs and targets within the clinical setting. The core engine, denoted as ART engine, is fully autonomous and consists of several modules, including data import/export, dose calculation, deformable image registration (DIR), interactive review graphical user interfaces, and messaging. The system has been successfully used for various research projects and is currently being thoroughly tested as part of our institutional software validation and quality assurance process. We have crossvalidated our DIR algorithm against four commercially available DIR algorithms using images from H&N radiotherapy. The estimated accuracy of our system was equivalent or better.

SU‐C‐BRF‐03: PCA Modeling of Anatomical Changes During Head and Neck Radiation Therapy

PURPOSE To develop principal component analysis (PCA) models from daily cone beam CTs (CBCTs) of head and neck (H&N) patients that could be used prospectively in adaptive radiation therapy (ART). METHODS For 7 H&N patients, Pinnacle Treatment Planning System (Philips Healthcare) was used to retrospectively deformably register daily CBCTs to the planning CT. The number N of CBCTs per treatment course ranged from 14 to 22. For each patient a PCA model was built from the deformation vector fields (DVFs), after first subtracting the mean DVF, producing N eigen-DVFs (EDVFs). It was hypothesized that EDVFs with large eigenvalues represent the major anatomical deformations during the course of treatment, and that it is feasible to relate each EDVF to a clinically meaningful systematic or random change in anatomy, such as weight loss, neck flexion, etc. RESULTS DVFs contained on the order of 3×87×87×58=1.3 million scalar values (3 times the number of voxels in the registered volume). The top 3 eigenvalues accounted for ∼90% of variance. Anatomical changes corresponding to an EDVF were evaluated by generating a synthetic DVF, and applying that DVF to the CT to produce a synthetic CBCT. For all patients, the EDVF for the largest eigenvalue was interpreted to model weight loss. The EDVF for other eigenvalues appeared to represented quasi-random fraction-to-fraction changes. CONCLUSION The leading EDVFs from single-patient PCA models have tentatively been identified with weight loss changes during treatment. Other EDVFs are tentatively identified as quasi-random inter-fraction changes. Clean separation of systematic and random components may require further work. This work is expected to facilitate development of population-based PCA models that can be used to prospectively identify significant anatomical changes, such as weight loss, early in treatment, triggering replanning where beneficial.

SU-E-J-100: Automatic CT-To-CT Contour Segmentation Using Deformable Image Registration Software for Head and Neck (H&N) Cancer Adaptive Radiotherapy

Purpose: The significant amount of time required for target delineation continues to be one of the major challenges associated with adaptive radiotherapy. The aim of this study is to evaluate the accuracy of efficient deformable image registration algorithms (available in the VelocityAI and Varian SmartAdapt commercial systems) for automatic segmentation of physician contours from planning CT to mid‐treatment CT images for H&N adaptive radiotherapy. Methods: Ten head and neck cancer patients were considered for this study, each with a planning CT (CT1) and a second CT (CT2) taken approximately 3 weeks into treatment. Treatment volumes and organs were manually delineated by a physician on both sets of CT scans. B‐spline‐based VelocityAI and Demons‐based SmartAdapt DIR algorithms were used to automatically deform CT1 and the relevant contour sets onto corresponding CT2 images. For each DIR, the volume of interest was set to encompass the whole contour set. The agreement of the automatically propagated contours with manually drawn contours of CT2 was visually evaluated by a physician, and the volume overlap was quantified using DICE coefficients. Results: The overall mean (1SD) DICE indices were 0.72(0.11) for VelocityAI and 0.68(0.17) for SmartAdapt. Both software attained a high degree of correlation for well differentiated and relatively large organs, with DICE indices often exceeding 0.8. Organs with small volumes and/or those with poorly defined boundaries showed less correlation (DICE: 0.5), likely due to volume averaging effects. Target volume contours generally aligned well (DICE: 0.7–0.8 for PTVs and 0.8–0.9 for GTVs). Conclusion: Use of automatic DIR‐based contour segmentation in H&N adaptive RT is likely to mitigate the need for manual delineation and thereby improve efficiency. More work is needed to evaluate the accuracy of these tools for routine clinical use, particularly for organs with small volumes, or those with poorly defined boundaries. This work is supported in part by Varian Medical Systems, Palo Alto, CA