Performance of Synthetic CT for Partial Brain IGRT
Morris E, Price R, Kim J, Schultz L, Chetty I, and Glide-Hurst C. Performance of synthetic CT for partial brain IGRT. Med Phys 2017; 44(6):3039.
Performance of synthetic CT for partial brain IGRT. Med Phys 2017; 44(6):3039.
Purpose: Recent advancements in synthetic computed tomography (synCTs) derived from MRI data have made MR-only based treatment planning clinically feasible in the brain, although their performance for IGRT is not well understood. This work compares the geometric equivalence of digitally reconstructed radiographs (DRRs) generated via CTs and synCTs derived from MRI data for a cohort of brain cancer patients and quantifies their performance for partial brain IGRT. Methods: Ten brain cancer patients (12 lesions, 7 post-surgical) underwent MR-SIM and CT-SIM within 4 days. SynCTs were generated by combining a novel ultra-short echo time (UTE) sequence with T1, T2, and FLAIR datasets using a previously developed voxel-based weighted summation approach. DRRs for synCT and CT were compared using a patient-specific thresholding technique and assessed via overlap index (OI), Dice similarity coefficient (DSC), and Jaccard index (JI). Planar IGRT images for 22 treatment fractions were evaluated to quantify differences between CT-generated DRRs and synCT-generated DRRs in 6 unique quadrants. A previously validated in-house MATLAB program was implemented to perform 2D-2D rigid registrations using the normalized mutual information (NMI) metric, as well as calculate shifts. Overall, 1490 unique rigid registrations were assessed. Results: DRR agreement, in the AP (anterior-posterior) and Lateral views, for OI, DSC, and JI were 0.99 ± 0.01, 0.97 ± 0.02, 0.95 ± 0.04, respectively. With the exception of one quadrant, NMI results were not statistically different between the CT and synCT. Additionally, rotational registration results were negligible (<0.07±). Statistically significant differences between CT and synCT were observed in 9/18 cases (p < 0.05). Mean absolute shifts for the population were <1 mm from 0, with relative CT and synCT differences being <0.75 mm. Patientspecific surgical resection cavities caused discrepancies in the synCT, yielding shifts up to 2.8 mm. Conclusion: Though statistically significant, differences between CT and synCTwere small and not clinically significant. Thus, shift differences may be patient dependent.