You are here: indexactivitiesspaacroboat

This is an old revision of the document!


Acroboat

Members

General purpose

To get a high quality of surgery, efficiency and no complications are the keypoints and the objectives researches by operators. Robotic surgery is becoming more and more widespread and with its computer interface, it is the only surgical approach allowing a very precise analysis of the procedures performed. The aim of the project is to analyse surgical practice and to understand and determine the differences between operators according to their expertise level and the factors leading to differences in the outcome of surgical procedures. To do so, we collect kinematic and video data from robot assisted hysterectomy surgeries on the daVinci Surgical System.

Description


Analysis of technical aspects.
We analyze the instrument trajectories and study any correlation with the clinical data. The analysis is performed by using motion data related metrics applied on the signals available from the Da Vinci export interface. About more than 20 different metrics are computed including duration, path length, average linear velocity, average acceleration, average jerk, bi-manual dexterity, motion smoothness, depth, idle time, relative phase, response orientation, economy of volume and working volume of the instrument’s positions. Each metric is computed, if applicable, for the right and left side and their combination. The metrics are computed on the whole procedure, as well as on the main surgical phases and steps, manually defined from the synchronized videos. From these metrics, classification approaches are then used to identify similar patterns 1) within populations of patients with similar clinical characteristics, 2) within populations of patients with similar outcome, and 3) within populations of surgeons with similar expertise levels.

Analysis of procedural aspects.
We manually define steps and phases of the operation from the synchronized video using an onthology previously established with the help of gynecologist surgeons. We then analyze the procedure using surgical workflow related metrics. Different metrics are computed including global DTW based distance, local NLTS, and pattern based distance. Following the same approach than for the first aim, from these metrics classification and machine learning approaches are then used to identify similar patterns 1) within populations of patients with similar clinical characteristics, 2) within populations of patients with similar outcome, and 3) within populations of surgeons with similar expertize levels.

Publications

Reference

Industrial Partner

inserm rennes1 ltsi