At KLIV we are currently working on capturing the knowledge of performing a surgery as graphs. These knowledge graphs helps in capturing the intra-operative details in a systematic way. The captured knowledge has a wide range on applications starting from surgery outcome prediction to live surgery assists and performing robotic surgeries.
We are currently focused on capturing the surgical knowledge of Laparoscopic Cholecystectomy in 7 surgical phases. The surgical phases and corresponding knowledge graphs will be listed below as and when completed.
Surgical Phases of Laparoscopic Cholecystectomy:
Port Insertion Phase
Overview of the graph (for best view download and open in browser)
Calot's Triangle Dissection Phase (Ongoing)
Overview of the graph (for best view download and open in browser)
We at KLIV are focused on automatically mapping the surgical videos and the corresponding knowledge graphs with the help of Deep Learning. Currently for mapping videos Laparoscopic Cholecystectomy with the corresponding knowledge graphs we have resorted to detection of surgical action triplets from the videos using spatio-temporal neural networks. The preliminary work on this is available here, the detailed outcomes will be released shortly.
With huge amounts of surgical data being generated, the storage, annotation, validation of the such data is becoming difficult. So we at KLIV are building a Surgical Annotation Platform that will be serving as one point for uploading, annotation and validation surgical data. It would indeed server as a hub for accessing the data and publishing the outcomes for AI related tasks. It would also be an integral part and the face of an active learning framework for surgical tasks. The portal link and demo videos will be made available soon.