Researchers have made significant progress in the development of a deep learning (DL) model capable of recognizing standardized surgical fields and assessing surgical skills automatically. The study, recently published in JAMA Surgery, highlights the potential of artificial intelligence (AI) in revolutionizing surgical skill assessment.
Traditionally, assessments of surgical skills rely on manual, video-based reviews of surgeon performance, which can be subjective and burdensome for human reviewers. To address these challenges, the research team turned to AI technologies, leveraging their advanced video analysis capabilities and increased objectivity.
The team focused on the field of endoscopy and developed a DL model specifically designed to identify standardized surgical fields in laparoscopic sigmoid colon resection. The model was trained using intraoperative videos of laparoscopic colorectal procedures obtained from the Japan Society for Endoscopic Surgery.
The researchers carefully selected videos performed by surgeons with high Endoscopic Surgical Skill Qualification System (ESSQS) scores to construct and train the model. A total of 650 videos were used, with 60 for model development and training, and the rest for validation.
The DL model was trained to recognize the surgical field and generate an AI confidence score (AICS) that measures its similarity to standardized surgical field development parameters. The team analyzed the correlation between AICS and ESSQS scores and evaluated the model’s performance across different score groups.
The results were promising. The Spearman rank correlation coefficient between AICS and ESSQS scores was 0.81, indicating a strong relationship. Moreover, the receiver operating characteristic (ROC) curve demonstrated an area under the curve of 0.93 for the low-score group and 0.94 for the high-score group, indicating a high level of accuracy in assessing surgical skills.
Based on these findings, the researchers concluded that the DL model shows feasibility as an automatic skill assessment method and holds potential for creating an automated screening system for surgical skills. This breakthrough could contribute to improving patient outcomes and reducing the burden on human reviewers.
The integration of AI in healthcare has gained significant momentum, and its application in surgical care continues to evolve. Recent endeavors have explored AI’s role in understanding surgical care needs, enhancing perioperative care through clinical intelligence, and improving operating room scheduling.
While AI-based clinician skill benchmarking is not new, the ability to evaluate surgeon performance in near real-time is an emerging possibility. In April, researchers from the California Institute of Technology (Caltech) and the University of Southern California introduced an AI system that provides surgeons with feedback on their performance and identifies areas for improvement. By analyzing surgical videos, the tool assesses discrete motions, offering objective feedback and insights to enhance surgical skills.
As AI continues to advance, the potential for improving surgical outcomes through video-based analytics of surgeon performance becomes increasingly tangible. Automated evaluation systems have the potential to transform surgical training, promote standardization, and ultimately enhance patient safety and surgical care.
Be the first to comment