Fig. 6: Representative cases across diverse clinical scenarios. | Nature Communications

Fig. 6: Representative cases across diverse clinical scenarios.

From: Automated MRI system for clinically significant prostate cancer detection development validation and real-world implementation

Fig. 6

Case A (A–E): 51-year-old man, PSA 13.3 ng/mL, left peripheral-zone lesion, PI-RADS 3. T2W and high-b DWI show a focal finding; ProAI delineates prostate and lesion and produces a low-probability heatmap (0.114). Biopsy: chronic inflammation; lesion resolved after targeted therapy. Case B (F–J): 66-year-old man, PSA 6.5 ng/mL, subtle peripheral-zone lesion initially PI-RADS 3 and overlooked on routine reading. ProAI identifies and segments the lesion; heatmap probability 0.922. Targeted biopsy and radical prostatectomy (RP) confirm csPCa (ISUP Grade Group 2). Case C (K–O): 70-year-old man, PSA 8.3 ng/mL, transition-zone lesion PI-RADS 2, a challenging differentiation from benign hyperplasia. ProAI localises the lesion; heatmap probability 0.964. RP confirms csPCa (ISUP Grade Group 2). Case D (P–T): 64-year-old man, PSA 3.9 ng/mL, right peripheral-zone lesion PI-RADS 4. Despite low spatial resolution on T2W and motion artefact on DWI, ProAI maintains robust segmentation; heatmap probability 0.856. RP confirms high-grade csPCa (ISUP Grade Group 5). PSA prostate-specific antigen, PI-RADS Prostate Imaging Reporting and Data System, T2W, T2-weighted, DWI diffusion-weighted imaging, PZ/TZ peripheral/transition zone, RP radical prostatectomy, csPCa clinically significant prostate cancer, ISUP International Society of Urological Pathology.

Back to article page