Related Experiment Video
Updated: Aug 15, 2025
Detection and Isolation of Cancer in Prostate Biopsies Using Stimulated Raman Histology and Artificial Intelligence
Published on: June 10, 2025
Stimulated Raman Scattering Microscopy Enables Gleason Scoring of Prostate Core Needle Biopsy by a Convolutional
Jianpeng Ao1, Xiaoguang Shao2, Zhijie Liu1
1State Key Laboratory of Surface Physics and Department of Physics, Human Phenome Institute, Academy for Engineering and Technology, Key Laboratory of Micro and Nano Photonic Structures (Ministry of Education), Yiwu Research Institute of Fudan University, Fudan University, Shanghai, P.R. China.
Stimulated Raman scattering (SRS) microscopy with deep learning offers rapid, label-free prostate cancer histology. This approach aids in precise focal therapy by providing timely tumor grading from fresh biopsies.
Area of Science:
- Biomedical optics
- Computational pathology
- Cancer diagnostics
Background:
- Focal therapy (FT) for prostate cancer requires precise tumor identification and grading.
- Current histopathology methods for prostate biopsies lack the speed and accuracy needed for same-time clinical decisions.
- Minimizing treatment toxicity necessitates preserving healthy tissue surrounding cancerous lesions.
Purpose of the Study:
- To evaluate stimulated Raman scattering (SRS) microscopy combined with a convolutional neural network (CNN) for rapid, label-free Gleason grading of prostate cancer biopsies.
- To assess the diagnostic accuracy and consistency of the SRS-CNN platform compared to traditional histopathology.
- To determine the potential of this integrated platform to support timely focal therapy decisions.
Main Methods:
- Utilized SRS microscopy for label-free, near real-time imaging of fresh prostate core needle biopsies.
- Developed and trained a CNN model using SRS images from 61 patients to classify Gleason patterns.
- Validated the CNN performance on an independent external test dataset of 22 cases and compared Gleason scoring with pathologist consensus in 21 cases.
Main Results:
- SRS microscopy effectively visualized heterogeneous histologic features in fresh prostate tissues.
- The CNN achieved 85.7% accuracy in classifying Gleason patterns on the training set and 84.4% on the external test set.
- The SRS-CNN system demonstrated 71% diagnostic consistency with expert pathologists for Gleason scoring.
Conclusions:
- Deep learning-assisted SRS microscopy provides a rapid and accurate method for evaluating prostate cancer tumor grade directly from fresh biopsies.
- This platform has the potential to streamline the diagnostic workflow, enabling timely histopathology crucial for focal therapy.
- The label-free, rapid nature of SRS-CNN analysis minimizes tissue processing, offering a significant advantage for clinical application.
