ESTRO 2024 - Abstract Book

S2332

Clinical - Urology

ESTRO 2024

Figure: Bland-Altman plot of the ADC measurement using MR-simulator and MR-Linac DWI. ADC percent difference was calculated using (ADCMR-simulator - ADCMR-Linac) / {( ADCMR-Linac + ADCMR-simulator )/2} X 100%

Significantly smaller ADC values were found in the prostate and ILs obtained from MR-Linac DWI compared to those obtained from MR-simulator DWI (prostate: 1.06±0.10 x10 -3 mm 2 /s vs 1.20±0.14 x10 -3 mm 2 /s, p<0.01; IL: 0.91±0.16 x10 -3 mm 2 /s vs 1.05±0.21 x10-3 mm 2 /s, all p<0.01). However, the SNR of b800 images was similar between the scanners for the prostate (17.09±4.79 (MR-simulator) vs 16.42±3.38 (MR-Linac), p=0.85) and the ILs (18.71±5.47 vs 19.81±4.89, p=0.18).

Table: Themean, bias, and limit of agreement of the ADC measurement using MR-simulator DWI and MR-Linac DWI. The non-parametric signed-rank test was performed with a significant level set at p-value <0.05 (denoted as *)

This study is limited by its small sample size. Additionally, the DWI acquisition could not be designed identically on the MR-Linac and MR-simulator due to differences in clinical utilization and scanner hardware configurations. The rigid image registration could not completely compensate for possible differences in image distortions and prostate deformations.

Conclusion:

This study observed a significantly larger MR-Linac ADC than MR-simulator ADC, and a small non-zero bias in the Bland-Altman analysis for both prostate glands and ILs. These findings suggest a small systematic inter-scanner