Institute of Molecular Pathology and Immunology of the University of Porto (Ipatimup)

Pathologists faced time-intensive manual review of prostate needle biopsy whole-slide images, with high rates of ancillary IHC staining requests and second-opinion consultations. Ensuring diagnostic accuracy across all histologic grades and tumor sizes placed significant burden on generalist pathologists, prolonging turnaround times.

Ipatimup deployed Paige Prostate Detect and Paige Prostate Grade & Quantify to assist 4 generalist pathologists evaluating 105 consecutive prostate needle biopsy whole-slide images. A crossover study design compared unassisted versus AI-assisted reading, measuring accuracy, concordance, slide read time, IHC requests, and second-opinion requests.

AI assistance reduced slide reading times by 21.9% across benign and malignant cases, cut second-opinion requests by 39.2%, and reduced IHC requests by 24.7% for cancer cases and 17.3% for non-cancer cases. Paige Prostate correctly classified 100% of whole-slide images with corrected diagnoses and identified four additional patients whose diagnoses were upgraded from benign/suspicious to malignant.

• AI-assisted pathology can simultaneously improve both efficiency (turnaround time, IHC usage) and diagnostic accuracy, not just one at the expense of the other.
• High negative predictive value makes AI a reliable screening and second-read tool, reducing downstream resource consumption.
• Generalist pathologists—not only specialists—benefit measurably from AI support in subspecialty oncology cases.