S. van Kranen

Validation of deformable registration in head and neck cancer using analysis of variance.

PURPOSE Deformable image registration (DIR) is often validated based on a distance-to-agreement (DTA) criterion of automatically propagated anatomical landmarks that were manually identified. Due to human observer variability, however, the performance of the registration method is diluted. The purpose of this study was to evaluate an analysis of variance (ANOVA) based validation to account for such observer variation. METHODS Weekly cone beam CTs (CBCTs) of ten head and neck cancer patients undergoing five weeks of radiotherapy were used. An expert identified 23 anatomical features (landmarks) on the planning CT. The landmarks were automatically propagated to the CBCT using multiregion-of-interest (mROI) registration. Additionally, two human observers independently localized these landmarks on the CBCTs. Subsequently, ANOVA was used to compute the variance of each observer on the pairwise distance (PWD). RESULTS ANOVA based analysis demonstrated that a classical DTA approach underestimated the precision for the mROI due to human observer variation by about 25%. The systematic error (accuracy) of mROI ranged from 0.13 to 0.17 mm; the variability (1 SD) (precision) ranged from 1.3 to 1.5 mm demonstrating that its performance is dominated by the precision. CONCLUSIONS The PWD-ANOVA method accounts for human observer variation allowing a better estimation of the of DIR errors.

Development and Validation of an Average Anatomy Model for Adaptive Radiation Therapy of Locally Advanced Lung Cancer Patients

Purpose/Objective(s): Multiple imaging techniques are available for imaging bone marrow and guiding radiation therapy planning. This study aimed to compare FDG-PET and FLT-PET for the purpose of identifying active pelvic bone marrow (BM), to quantify variation in the location of active BM, and determine which technique is likely to be better for bone marrow-sparing radiation planning. Materials/Methods: We sampled 41 patients from three prospective clinical trials, of which 25 underwent pretreatment FDG-PET/CT only, 7 underwent pretreatment FLT-PET/CT only, and 9 underwent both. Following registration of each PET/CT with the planning CT, an active BM subvolume was defined as the subset of the pelvic BM with standardized uptake values (SUV) above a designated threshold. Three SUV thresholds were chosen for each PET image such that they defined an active BM subvolume comprising 40%, 50%, and 60% of the total pelvic BM volume. We used the Dice similarity coefficient to quantify the percent overlap of active BM volumes of equal size. Differences in the spatial distribution of active BM were assessed using a regiongrowing algorithm. Results: For patients with both FDG-PET and FLT-PET scans, the mean Dice coefficients for the 40%, 50%, and 60% subvolume thresholds were 0.683 (95% confidence interval (CI), 0.654-0.712), 0.732 (95% CI, 0.7110.753), and 0.781 (95% CI, 0.767-0.795), respectively. For the 34 patients with FDG-PET scans, comparing individual active BM subvolumes to the mean image, Dice coefficients varied from a minimum of 0.598 at the 40% threshold to a maximum of 0.889 at the 60% threshold. The corresponding Dice coefficients for 16 patients with FLT-PET scans ranged from 0.739 at the 40% threshold to 0.912 at the 60% threshold. At each threshold, active BM subvolumes identified using FLT-PET required significantly more iterations to converge on region-growing analysis compared to FDG-PET, indicating that proliferating BM is more highly clustered than metabolically active BM. Conclusion: We found significant agreement between FDG-PET and FLTPET in identifying active BM; however, FLT-PET was associated with significantly less individual variation and is likely to be superior to FDGPET for BM-sparing radiation therapy. Author Disclosure: J.C. Wyss: None. R. Carmona: None. R. Karunamuni: None. J. Pritz: None. C.K. Hoh: None. L.K. Mell: None.