Glioma consensus contouring recommendations from a MR-Linac International Consortium Research Group and evaluation of a CT-MRI and MRI-only workflow
Chia‐Lin Tseng, James Stewart, Gillian Whitfield, Joost J.C. Verhoeff, Joseph Bovi, Hany Soliman, Caroline Chung, Sten Myrehaug, Mikki Campbell, Eshetu G. Atenafu, Chris Heyn, Sunit Das, James Perry, Mark Ruschin, Arjun Sahgal
Abstract
INTRODUCTION: This study proposes contouring recommendations for radiation treatment planning target volumes and organs-at-risk (OARs) for both low grade and high grade gliomas. METHODS: Ten cases consisting of 5 glioblastomas and 5 grade II or III gliomas, including their respective gross tumor volume (GTV), clinical target volume (CTV), and OARs were each contoured by 6 experienced neuro-radiation oncologists from 5 international institutions. Each case was first contoured using only MRI sequences (MRI-only), and then re-contoured with the addition of a fused planning CT (CT-MRI). The level of agreement among all contours was assessed using simultaneous truth and performance level estimation (STAPLE) with the kappa statistic and Dice similarity coefficient. RESULTS: A high level of agreement was observed between the GTV and CTV contours in the MRI-only workflow with a mean kappa of 0.88 and 0.89, respectively, with no statistically significant differences compared to the CT-MRI workflow (p = 0.88 and p = 0.82 for GTV and CTV, respectively). Agreement in cochlea contours improved from a mean kappa of 0.39 to 0.41, to 0.69 to 0.71 with the addition of CT information (p < 0.0001 for both cochleae). Substantial to near perfect level of agreement was observed in all other contoured OARs with a mean kappa range of 0.60 to 0.90 in both MRI-only and CT-MRI workflows. CONCLUSIONS: Consensus contouring recommendations for low grade and high grade gliomas were established using the results from the consensus STAPLE contours, which will serve as a basis for further study and clinical trials by the MR-Linac Consortium.