CAISA - Collaborative Artificial Intelligent Surgical Assistant

Purpose and goal

Continued improvement of results and patient safety in surgery will require a transfer of know-how from humans to machines. CAISA uses video data from real heart surgeries to train AI models to recognize the surgeon´s hands and instruments and anatomical structures on the heart. The models are combined to control a robot to act as an assistant to the surgeon. The project generates new AI models and a basic structure for how autonomous surgical systems can be developed. A demonstrator must be created, tested and validated on a special test bed.

Expected effects and result

CAISA will generate manuscripts for publication in scientific journals and disseminated by other channels to the public domains. The demonstrator must clearly show the proof-of-concept. Interest in the field of autonomous surgical robotics will hopefully increase, which should lead to more initiatives, investments and support through collaborations with the major AI companies (Google, Microsoft, ChatGPT, Apple etc.) in the world. More surgical specialties will begin data collection and contribute greater breadth in development towards autonomous surgical robotics.

Planned approach and implementation

The project a collaboration between four parties with unique areas of expertise: 1. Surgery - Children´s Heart Centre, Skånes Universitetssjukhus in Lund 2. AI and Automation research - Dept of Computer Science and Dept of Control Technology, Lunds Universitet 3. Medium-sized company in Robotics - Cognibotics AB, Lund 4. Small-size company in AI based Object Estimation - Cobotic AB, Västerås We share data, AI models, programming language for robot control; receives advice from an expert panel of senior surgeons; uses the latest developments in robotics and test bed.