ESTRO 2024 - Abstract Book

S2331

Clinical - Urology

ESTRO 2024

467

Digital Poster

Inter-scanner ADC reproducibility on prostate cancer patients: 1.5T MR-Linac vs 1.5T MR-simulator

Oi Lei Wong 1 , Jing Yuan 1 , Darren M.C. Poon 2 , Sin Ting Chiu 3 , Bin Yang 4 , Cindy Xue 1 , George Chiu 3 , Kin Yin Cheung 4

1 Hong Kong Sanatorium & Hospital, Research Department, Hong Kong, Hong Kong. 2 Hong Kong Sanatorium & Hospital, Comprehensive Oncology Center, Hong Kong, Hong Kong. 3 Hong Kong Sanatorium & Hospital, Department of Radiotherapy, Hong Kong, Hong Kong. 4 Hong Kong Sanatorium & Hospital, Medical Physics Department, Hong Kong, Hong Kong

Purpose/Objective:

The potential of diffusion-weighted imaging (DWI) in MR-guided radiotherapy (MRgRT) for prostate cancer (PC) in treatment response monitoring and lesion localization is well recognized. However, the use of DWI in PCa MRgRT lacks standardization. One of the challenges is the reproducibility of the apparent diffusion coefficient (ADC), especially since pre-radiotherapy treatment planning MR scans are usually obtained using MR-simulators or diagnostic MR scanners and inter-fractional scans might be obtained using MR-Linac. This study evaluates the inter scanner ADC reproducibility on a 1.5T MR-simulator and 1.5T MR-Linac in PC patients undergoing MRgRT.

Material/Methods:

Nineteen localized-PC patients with twenty MRI-visible intra-prostatic lesions (ILs) (69±8 years) were included. DWI scans were acquired on a 1.5T MR-simulator and a 1.5T MR-Linac in the same radiation treatment position and setup, prior to any irradiation, within an interval of 5-15 days. Both scans were acquired with the same TR/TE (3000/77ms) and slice thickness (4mm). The voxel size was slightly larger for MR-Linac (3x3 mm 2 ) than MR-simulator (2.56x2.56 mm 2) . The optimized b-values were 0, 800, 1400 s/mm 2 for MR-simulator and 150, 500, 800 s/mm2 for MR-Linac. ADC maps were calculated using the scanner console for all DWI scans. The prostate and ILs were manually delineated on MR-simulator images by a radiation oncologist with >10 years of experience. For the MR-Linac DW images, the prostate and IL segmentations were obtained by translating the manual segmentation to the MR-Linac b800 images by an MR physicist with >10 years of experience. The signal-to noise ratio (SNR) of the b800 images and mean ADC were calculated and compared among the scanners using the signed-rank test with a significance level of 0.05 when appropriate. The ADC reproducibility was evaluated using Bland-Altmant analysis.

Results:

Based on the Bland-Altman plot, all the data fell within the limit of agreement for the prostate ADC, while 2 data points fell outside the limit of agreement for the ILs ADC. Further investigation revealed that these data points corresponded to 2 of the 4 poorly visualized ILs on the MR-Linac b800 image, which may be attributed to compromised gradient performance on the MR-Linac. Similar non-zero bias was observed for the prostate (12.2%) and the ILs (14.5%), indicating a potential systematic error between the two scanners.