A deep learning system for prostate cancer diagnosis and grading in whole slide images of core needle biopsies
Problem
Gleason grading of prostate core needle biopsies is subjective and depends on pathologist expertise, and deep learning systems degrade on slides from other labs.
Why it matters
Grading accuracy directly drives prostate cancer risk stratification and treatment decisions.
What the AI did
A deep learning system segmented and graded epithelial tissue in whole slide images using a training methodology that learns domain-agnostic features.
Results
Accuracy 89.4% and κquad 0.92 on internal test set of 425 CNB WSIs; accuracy 85.3% and κquad 0.96 on external set of 1201 images.
Impact
Strong performance on an external test set suggests robust grading support across centers, though only retrospective.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2022 |
| Work type | article |
| Authors | Nitin Singhal, Shailesh Mahendra Soni, Saikiran Bonthu, Nilanjan Chattopadhyay, Pranab Samanta, Uttara Joshi, Amit Jojera, Taher Chharchhodawala et al. |
| Citations | 92 (OpenAlex, at import time) |
| References | 25 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-022-07217-0 |
| Open access | gold |
| Keywords | deep learning, Gleason grading, whole slide images, prostate cancer, domain shift |
Importance rating
significant — thousands to millions affected
Grading accuracy drives prostate cancer risk stratification and treatment decisions; robustness across centers suggests broad clinical utility, though retrospective only.