The abstract is read at the publisher; the summary is JClub's.
An automated gland MRI radiomics model predicts time to pathologic progression during prostate cancer active surveillance, achieving a C-index of 0.77 and time-dependent AUC of 0.87 at 36 months.
- Why it matters: Repeated biopsies during active surveillance are burdensome, and existing MRI-based tools that avoid biopsy are constrained by challenges in automated lesion segmentation.
- What they did: In a retrospective single-institution cohort of 522 men (grade group 1 or favorable grade group 2) from 2010–2025, nnU-Net automated gland segmentation and LASSO-selected gland radiomic features trained an XGBoost accelerated failure time model.
- The result: The gland radiomics model outperformed clinicopathologic (C-index 0.68) and lesion radiomics (C-index 0.68) models (adjusted P ≤ .017), with progression-free survival differing by risk group and negative predictive value of 1.00 at a 0.25 threshold.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.