Detection of spermatogonial stem/progenitor cells in prepubertal mouse testis with deep learning

Area of Science:

• Reproductive Biology
• Computational Biology
• Histology

Background:

• Spermatogonial stem/progenitor cells (SSPCs) are vital for male fertility.
• Prepubertal testicular damage can lead to male infertility.
• Efficient detection of SSPCs is challenging but crucial for clinical diagnosis.

Purpose of the Study:

• To develop and validate a deep learning (DL) method for detecting and counting seminiferous tubules and SSPCs.
• To apply DL for analyzing newborn mouse testis sections.
• To establish a foundation for automated cell detection in infertility clinics.

Main Methods:

• Utilized YOLO object detection model based on deep learning.
• Analyzed hematoxylin and eosin (H&E) stained testicular sections.
• Employed SALL4 immune labeling as a positive control for SSPC identification.
• Created datasets from newborn mouse testis sections.

Main Results:

• Achieved high performance metrics for seminiferous tubule detection (0.98 mAP, 0.93 precision, 0.96 recall, 0.94 f1-score).
• Demonstrated strong performance for SSPC detection (0.88 mAP, 0.80 precision, 0.93 recall, 0.82 f1-score).
• The DL model showed high sensitivity in detecting these structures.

Conclusions:

• Deep learning effectively detects seminiferous tubules and SSPCs in prepubertal testes.
• The developed method minimizes human error in cell detection and counting.
• This represents a significant step towards automated diagnostic systems for male infertility.