DTI: 2010–11 Funded Proposals

Daniel F. Keefe, Robert Sweet

Coupling Automatic Skill Evaluation and Visual Analytics to Improve Surgical Training

The proposed cross-disciplinary research will explore novel motion analysis and visualization tools to address far-reaching challenges in surgical training. Current approaches to surgical training suffer from several limitations, including: 1.) A lack of objective measures for evaluating surgical skill, and 2.) A lack of mechanisms for providing specific, meaningful feedback to guide training surgeons in refining their techniques, which require perfecting intricate 3D motions of the body and hands. We propose a new approach that combines local data-driven modeling of human performance in surgical tasks with interactive multi-variate data visualization tools. Using these tools, surgeons will be able to identify (and see) critical differences between hundreds of instances of the same procedure performed by various surgeons. Together these advances can address a critical need in current surgical training, leading to improved, more objective evaluations of surgical skill and providing surgeons with concrete, visual feedback that can help to accelerate progress along the extended learning curves now associated with acquiring surgical skills. The work will utilize data collected via cutting-edge robot-assisted surgeries and surgical simulators developed at the University of Minnesota SimPORTAL as well as data from a new open database of surgical performance data termed the "surgical genome project."