A Deep Learning Model for Prostate Adenocarcinoma Classification in Needle Biopsy Whole-Slide Images Using Transfer Learning
Problem
Manual microscopic diagnosis of prostate adenocarcinoma across large numbers of needle biopsy specimens is time-consuming and limited by human resources, so rapid, accurate screening techniques are needed.
Why it matters
Prostate biopsy diagnosis determines optimum cancer treatment but consumes scarce pathologist time.
What the AI did
Trained deep learning models to classify needle biopsy whole-slide images into adenocarcinoma versus benign lesions, evaluating them on needle biopsy, TUR-P, and TCGA public test sets.
Results
not stated in source
Impact
Evidence supports a model intended to assist pathologists in screening prostate biopsies; no performance or clinical adoption figures stated.
Record
| Venue | Diagnostics |
|---|---|
| Published | 2022 |
| Work type | article |
| Authors | Masayuki Tsuneki, Makoto Abe, Fahdi Kanavati |
| Citations | 30 (OpenAlex, at import time) |
| References | 56 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3390/diagnostics12030768 |
| Open access | gold |
| Keywords | deep learning, prostate cancer, whole-slide images, transfer learning, pathology |
Importance rating
significant — thousands to millions affected
Prostate biopsy diagnosis is pivotal for optimum cancer treatment; manual screening of large case volumes is time-consuming and limited by pathologist resources.