Related Experiment Video
Updated: Sep 26, 2025
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
310
Automatic Prostate Gleason Grading Using Pyramid Semantic Parsing Network in Digital Histopathology
Yali Qiu1, Yujin Hu1, Peiyao Kong1
1School of Biomedical Engineering, Health Science Center, Shenzhen University, National-Regional Key Technology Engineering Laboratory for Medical Ultrasound, Guangdong Key Laboratory for Biomedical Measurements and Ultrasound Imaging, Shenzhen, China.
Frontiers in Oncology
|April 25, 2022
Summary
An artificial intelligence system accurately detects prostate cancer and assigns Gleason grades, improving upon pathologist diagnoses. This AI tool aids in precise cancer grading, reducing treatment errors.
Area of Science:
- Uropathology
- Artificial Intelligence
- Medical Image Analysis
Background:
- Prostate biopsy histopathology and immunohistochemistry are crucial for diagnosing prostate cancer and assessing differentiation.
- Increasing demand for experienced uropathologists strains diagnostic capacity and can lead to grading inconsistencies.
- Inconsistent Gleason grading can result in suboptimal patient treatment, including overtreatment or undertreatment.
Purpose of the Study:
- To develop an artificial intelligence (AI) system for accurate prostate cancer detection and Gleason grading.
- To alleviate the pressure on pathologists and improve the consistency of cancer grading.
- To provide a clinically acceptable tool for assessing prostate cancer differentiation.
Main Methods:
- A pyramid semantic parsing network (PSPNet), inspired by semantic segmentation, was developed for automatic prostate Gleason grading.
- An auxiliary prediction output was incorporated to enhance segmentation performance during network training.
- The method demonstrated effectiveness in segmenting tissue micro-array (TMA) images.
Main Results:
- The AI system achieved first rank on the MICCAI 2019 prostate segmentation and classification benchmark using 321 biopsies.
- The method demonstrated high consistency with pathologist diagnoses and excelled in distinguishing high-risk (Gleason 4, 5) from low-risk (Gleason 3) cancer.
- The AI system achieved top performance in differentiating benign from malignant tissues based on various metrics.
Conclusions:
- The developed AI method provides effective and accurate Gleason grading results for prostate biopsies.
- The AI system shows significant potential to support uropathologists and improve patient treatment accuracy.
- The Python source code is publicly available, promoting transparency and further research.
