Instrument Segmentation

* Indicates equal contribution
Semantic segmentation of surgical scenes is a fundamental task that must be solved in the pursuit of autonomous robot-assisted surgeries and image-guided interventions. The objective of this task is to create an automatic method for identifying and spatially limiting the relevant objects in a surgical scene. This task is the first step towards solving computer-aided surgery problems such as overlay and 3D reconstruction of organs, instrument tracking, navigation, and programming of robot arms for precision and strength critical tasks. Our work mostly focuses on footage from robotic-assisted surgery scenes, however the methods developed for natural images are directly applicable to laparoscopic surgery scenes, thus widening the scope and range of applications for our work.

Publications and awards

ISINet: An Instance-Based Approach for Surgical Instrument Segmentation

C. González*, L. Bravo-Sánchez* and P. Arbeláez

23rd International Medical Image Computing and Computer Assisted Intervention (MICCAI), 2020


L. Bravo-Sánchez*, C. González*, and P. Arbeláez

Arxiv preprint

ROBUST-MIS Challenge 2019

L. Bravo-Sánchez*, C. González*, and P. Arbeláez
Our team Uniandes won 5 awards at the Robust Endoscopic Instrument Segmentation Challenge 2019 (ROBUST-MIS) as a part of MICCAI 2019. Our method won first place in the Multiple Instance Detection task, and in the Multiple Instance Segmentation task we won one first place and two second places.
  • Address:
    Cra. 1 E No. 19A - 40. 111711, Bogotá, Colombia - Mario Laserna Building - School of Engineering - Universidad de Los Andes
  • Phone:
    [571] 332 4327, 332 4328, 332 4329