# Deep Residual Learning for Instrument Segmentation in Robotic Surgery

## TL;DR

This paper introduces a deep residual learning approach that improves binary segmentation and extends it to multi-class segmentation for robotic surgical instruments, enhancing accuracy in minimally invasive surgery tasks.

## Contribution

It advances surgical instrument segmentation by applying deep residual networks and dilated convolutions, and extends the method to multi-class segmentation of different instrument parts.

## Key findings

- Improved segmentation accuracy over previous methods
- Effective multi-class segmentation of surgical tools
- Validated on MICCAI Endoscopic Vision Challenge dataset

## Abstract

Detection, tracking, and pose estimation of surgical instruments are crucial tasks for computer assistance during minimally invasive robotic surgery. In the majority of cases, the first step is the automatic segmentation of surgical tools. Prior work has focused on binary segmentation, where the objective is to label every pixel in an image as tool or background. We improve upon previous work in two major ways. First, we leverage recent techniques such as deep residual learning and dilated convolutions to advance binary-segmentation performance. Second, we extend the approach to multi-class segmentation, which lets us segment different parts of the tool, in addition to background. We demonstrate the performance of this method on the MICCAI Endoscopic Vision Challenge Robotic Instruments dataset.