Artificial intelligence-assisted detection of internal openings in anal fistulas using endorectal ultrasound: a YOLOv11-based diagnostic system.

BACKGROUND

Accurate and efficient detection of internal openings (IO) of anal fistulas (AF) in endorectal ultrasound (ERUS) images is crucial for diagnosis and treatment planning. This study aimed to develop an artificial intelligence (AI) system to automatically identify these anatomical landmarks in real time, thereby improving diagnostic efficiency.

METHODS

An AI system based on the YOLOv11 architecture (YOLOv11n model) was constructed. The model was trained and evaluated using a dataset consisting of 238 ERUS cases from The Fifth People's Hospital of Chengdu. The system's performance was assessed using conventional object detection metrics, including precision, recall, and mean average precision (mAP) at various intersection-over-union (IoU) thresholds.

RESULTS

The proposed YOLOv11n model demonstrated excellent performance in detecting IO of AF. The results showed precision of 0.925, recall of 1.0, mAP of 0.99 at an IoU threshold of 0.5, and mAP of 0.602 across IoU thresholds from 0.50 to 0.95. These findings highlight the model's robustness, reliability, and accuracy in rapidly analyzing ERUS images.

CONCLUSIONS

Utilizing the YOLOv11 architecture, this AI system enables real-time, precise identification of IO of AF in ERUS images, significantly enhancing diagnostic efficiency. The system's outstanding performance offers potential for standardized training for radiologists and colorectal surgeons, effectively integrating AI advancements with clinical educational needs.