Related Experiment Video
Updated: Nov 6, 2025
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
487
A method for utilizing automated machine learning for histopathological classification of testis based on Johnsen
Yurika Ito1, Mami Unagami1, Fumito Yamabe1
1Department of Urology, Toho University School of Medicine, 6-11-1, Omori-Nishi, Ota-ku, Tokyo, 143-8541, Japan.
Scientific Reports
|May 10, 2021
Summary
Artificial intelligence (AI) can now automatically determine Johnsen scores from testicular tissue images, offering a potential replacement for manual pathologist evaluation. This AI tool shows high accuracy, especially with detailed image analysis, aiding in clinical practice.
Area of Science:
- Pathology
- Artificial Intelligence
- Reproductive Medicine
Background:
- The Johnsen score is crucial for evaluating testicular sperm production.
- Traditional manual scoring by pathologists can be time-consuming and subjective.
- Automating this process could improve efficiency and consistency in male infertility assessments.
Purpose of the Study:
- To investigate the efficacy of an artificial intelligence (AI) tool for automated Johnsen score determination.
- To assess if AI can serve as a viable alternative to traditional manual scoring by pathologists.
- To support and enhance the diagnostic capabilities of pathologists in evaluating testicular histology.
Main Methods:
- Utilized Google Cloud AutoML Vision for AI model development and assessment.
- Collected and stained testicular tissue samples from 264 patients.
- Created datasets of histopathology images at 400x magnification and 5.0 × 5.0 cm cutouts for AI training and validation.
- Defined four distinct Johnsen score categories (1-3, 4-5, 6-7, 8-10) for classification.
Main Results:
- The AI algorithm achieved an average precision of 82.6%, precision of 80.31%, and recall of 60.96% for the 400x magnification dataset.
- For the 5.0 × 5.0 cm cutout dataset, the AI demonstrated superior performance with average precision of 99.5%, precision of 96.29%, and recall of 96.23%.
- This study represents the first report of an AI-based algorithm for predicting Johnsen scores.
Conclusions:
- AI-powered automated Johnsen scoring shows significant promise as a tool to assist pathologists.
- The AI tool, particularly with detailed image analysis, achieves high accuracy in predicting Johnsen scores.
- This technology has the potential to streamline the evaluation of male fertility and improve diagnostic consistency.
