Application of AI-assisted pathological diagnosis and section quality assessment for prostate biopsy specimens
XIA Cunyan
PENG Yan
LEI Ting
LI Qing
Abstract：Objective To observe the role of artificial intelligence (AI) in prostate biopsy pathological diagnosis and simultaneously conduct AI tissue section quality assessment, summarizing and analyzing the application of AI in prostate cancer. Methods A total of 5478 prostate biopsy paraffin sections from the Department of Pathology, Changzhou First People's Hospital, from March 2025 to July 2025 were scanned digitally. An AI-assisted pathology diagnosis system was used for examination, and the results were compared with those of manual diagnosis to observe diagnostic sensitivity and specificity. At the same time, a quality control model was used for tissue section quality assessment. Results The sensitivity of AI diagnosis in prostate biopsy tissues was 98%, and the specificity was 94%. The quality control scores were as follows: 5145 sections rated as Grade A, 329 as Grade B, and 4 as Grade C. The main scoring items were incomplete preparation and wrinkles, with an occurrence rate of 94.5%; followed by contamination of the sections, with a rate of 3%, and bubbles in the slides, with a rate of 2.5%. Conclusion Prostate biopsy is the gold standard for preoperative diagnosis of prostate cancer. Currently, each patient has more than 12 biopsy samples, which consumes a large amount of manpower in preparation and diagnosis. AI-assisted pathological diagnosis of prostate biopsy tissues shows good consistency with manual examination, and has obvious advantages in shortening the diagnostic time, avoiding missed or misdiagnosis. AI also plays an efficient, convenient, and easy-to-analyze role in tissue section quality control. The application of AI in prostate cancer is increasingly becoming an indispensable auxiliary tool for the diagnostic and technical teams.
Keywords：Artificial IntelligenceAssisted Pathological DiagnosisProstate BiopsyHE SectionsQuality Control
Publication Date：2025-12-28
Online Publishing Date：2026-01-28(First online date of this platform, not the publication date of the document)
Pages：4( 1697-1700 )