Reproducibility of rectal tumor volume delineation using diffusion-weighted MRI: Agreement on volumes between observers

PURPOSE: To retrospectively evaluate the inter-observer agreement between a radiologist and a radiation oncologist and volume differences, in T2 and diffusion-weighted (DWI) MRI of gross tumor volume (GTV) delineation, in rectal cancer patients. MATERIALS AND METHODS: Two observers, a radiologist and a radiation oncologist, delineated GTVs of 50 patients on T2-weighted MRI (T2GTV) and echo planar DWI (DWIGTV). Observers agreement was assessed using DICE index, Bland-Altman analysis and intra-class correlation coefficient (ICC). Student's t-test was used for GTV comparison. RESULTS: Median T2GTV and DWIGTV were 17.09±14.12 cm3 (1.92-62.03) and 12.79±12.31 cm3 (1.23-62.25) for radiologist, and 16.82±13.66 cm3 (1.78-65.9) and 13.72±12.77 cm3 (1.29-69.75) for radiation oncologist. T2GTV were significantly larger compared to DWIGTV (P<0.001 and P<0.001, for both observers). Mean DICE index for T2GTV and DWIGTV were 0.80±0.07 and 0.77±0.06. The mean difference between the two observers were 0.26cm3 (95% CI: -5.36 to 5.88) and -1.13cm3 (95% CI: -5.70 to 3.44) for T2 and DWI volumes. The ICC for T2 volumes was 0.989 (95% CI: 0.981-0.994) (P<0.001) and 0.992 (95% CI: 0.986-0.996) (P<0.001) for DWI volumes. CONCLUSION: DWI resulted in smaller volumes delineation compared to T2-weighted MRI. Substantial and almost perfect agreements were reported for DWIGTV and T2GTV between radiologist and radiation oncologist. Due to the fact that DWI could be considered a simple technique for volume delineation for radiation oncologist, DWI could be used to improve quality in radiation planning for an accurate boost volume delineation when a dose escalation is investigated.