(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: To compare gross tumor volume (GTV) contour variability in the radiotherapy treatment planning of sarcomas and chordomas drawn using CT only, CT and MR, or CT, MR, and PET.
Methods: For each tumor, GTVs were drawn across three modality groups: CT only (group 1), CT and MR (group 2), and CT, MR, and PET (group 3). Image registration was performed using MIMVista. For each group, three contouring trials were drawn at least 24 hours apart to assess intra-modality contour variability. Contouring was randomized within each group for all subjects and performed sequentially with group 1, then 2 and finally 3. Sixteen subjects have been included so far. Intra-observer variability was assessed in 2 ways: i. by calculating the 3D Dice score between pairs of contours, to measure the similarity between masks (higher Dice score indicates greater agreement), and ii. by calculating the Hausdorff metric, which measures distances between subsets of contours in a metric space (smaller distances indicate greater agreement). The Warfield’s simultaneous truth and performance level estimation (STAPLE) method was applied to create a consensus contour from which distances are measured. Variability was then stratified by tumor size for each group.
Results: In large tumors, no significant intra-modality contour variability was found. In tumors with a major axis <12 cm, intra-modality contour variability was significantly lower in groups 2 and 3 as compared to group 1 (pd= 0.034; ph = 0.045 and pd= 0.0024; ph=0.0061, respectively; where pd and ph are the p-values obtained by using the Dice score and the Hausdorff method respectively). In 13% of subjects, distant disease was detected on PET/CT/MR that was not visible otherwise.
Conclusion: Multi-modality GTV delineation of sarcomas and chordomas provided valuable reduction in variability in the challenging case of small lesions but did not significantly improve performance in larger tumors.
Funding Support, Disclosures, and Conflict of Interest: Research Support: T32EB013180, R01CA165221, R21EB021710 and P41EB022544.
Keywords
Taxonomy
IM- Multi-Modality Imaging Systems: MRI/PET - human
Contact Email
