However, the study acknowledges limitations such as the lack of source video diversity and the inability to train segmentation models using both crowdsourced and expert annotations due to time constraints. The authors suggest further research to determine the limitations of this methodology when applied to increasingly complex anatomical structures and to investigate clinical outcomes with the use of this technology. The ultimate goal is to enable additional applications of AI-assisted multimodal imaging data for enhanced real-time clinical decision support for safer surgery and improved outcomes.
Key takeaways:
- The study explores the use of artificial intelligence (AI) in surgical procedures, particularly in the identification of surgical phases, critical events, and surgical anatomy.
- Most AI models in this field use supervised machine learning and require large amounts of annotated video data, which is time-consuming and costly to obtain from domain experts.
- The researchers used a gamified crowdsourcing platform to obtain annotated training data of surgical tissues, which was then used to train a soft tissue segmentation AI model.
- The study found that the AI model trained on crowdsourced data performed equally well as those trained on expert annotations, demonstrating the potential of crowdsourcing in accelerating the development and deployment of surgical AI models.