Fig. 1: Study design.
From: Tumor evolution metrics predict recurrence beyond 10 years in locally advanced prostate cancer
a, Within the IMRT clinical trial (NCT00946543), 6–12 ultrasound-guided diagnostic needle biopsies were taken per individual for routine diagnosis and were embedded in paraffin. b, Decision tree for the imaging cohort (n = 250 individuals, n = 1,923 biopsies) and sequencing cohort (n = 111 individuals, n = 578 biopsies). The DELINEATE trial cohort was not included. c, Kaplan–Meier curve for time to recurrence in the imaging cohort (n = 250 individuals). d, Experimental workflow for FFPE biopsies and matched germline (buffy coat/normal FFPE tissue). Figure created with BioRender.com. e, Computational histopathology analysis with deep learning both for Gleason segmentation and single-cell classification on H&E sections. Sample input and output is shown for FI-115-S8; SCCNN, spatially constrained convolutional neural network. f, Example of individual FI-132, where computational Gleason segmentation and CNA genomic data were compared.
