Volumetric and voxel-wise based analysis of the intraprostatic lesions on multiparametric MRI.
Carver E, Wen N, Liu C, Chetty I, Lee J, Elshaikh M, and Movsas B. Volumetric and voxel-wise based analysis of the intraprostatic lesions on multiparametric MRI. Med Phys 2018; 45(6):e699-e700.
Purpose: Pilot studies have shown that administering a boost dose to the intraprostatic lesions for prostate cancer treatment may increase the probability of tumor control. Multiple MR Imaging (mpMRI) techniques have shown promise for the detection and localization of intraprostatic lesions. In this study, we performed both volume and voxel-wise based analysis to evaluate the tumor foci delineated on three mpMRI sequences. Methods: 90 patients with mpMRI imaging data and biopsy proven prostate cancer from the SPIEAAPM-NCI PROSTATEx Challenge were included in the study. Annotated tumor foci with known Gleason scores were contoured by a radiation oncologist on T2 weighted, Apparent Diffusion Coefficient (ADC), and volume transfer coefficient (ktrans) images. Both ADC and ktrans image series were registered and resampled to the T2 weighted image series. Dice and Jaccard coefficients were calculated between the contours of each pair of the three modalities and a Spearman rank correlation was calculated on a voxel-based level within a combined volume of three contours. Results: Tumor foci were located in the anterior fibromuscular stroma, central zone, peripheral zone, and transitional zone in 35.2%, 5.6%, 32.4%, and 25.4% of patients, respectively. Gleason grades for scores from 1 to 5 (with grades 4, 5 grouped together) were 29.6%, 40.8%, 15.5%, and 14.1%, respectively. Ktrans values were indistinguishable between cancerous regions and the remaining prostatic regions for 19 patients. The Dice coefficient and Jaccard values were 0.74 ± 0.13, 0.60 ± 0.15 for T2-ADC and 0.61 ± 0.16, 0.46 ± 0.16 for T2-Ktrans. Spearman correlations were 0.20 ± 0.20 for T2-ADC and 0.13 ± 0.25 for T2-ktrans. Conclusion: The volume of intraprostatic lesions indicated by T2 and ADC had better agreement than T2 and ktrans. The ktrans map provided valuable information on vascular abnormality for target delineation, but was limited to Gleason score 4, 5 lesions only.