# TotalSegmentator: robust segmentation of 104 anatomical structures in CT images

**Authors:** Jakob Wasserthal, Manfred Meyer, Hanns-Christian Breit, Joshy Cyriac, Shan Yang, Martin Segeroth  
**arXiv:** [2208.05868](https://arxiv.org/abs/2208.05868)  
**Date:** 2022-08-11  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2208.05868v1/

## Key points

- Authors Jakob Wasserthal, Manfred Meyer, Hanns-Christian Breit, Joshy Cyriac, Shan Yang, and Martin Segeroth focus on improving automatic segmentation of multiple anatomical structures in whole-body CT images.
- They introduce a new dataset and segmentation toolkit to address common issues faced by existing algorithms.
- An improved workflow for creating ground truth segmentations is presented.
- TotalSegmentator is introduced as a versatile tool that can easily segment multiple anatomical structures.
- The work addresses key challenges in automatic segmentation and provides a valuable resource for researchers and practitioners working with CT imaging data.
- The code for TotalSegmentator is accessible on GitHub at https://github.com/wasserth/TotalSegmentator.
- The dataset can be found at https://doi.org/10.5281/zenodo.6802613.

## Summary

In their work titled "TotalSegmentator: robust segmentation of 104 anatomical structures in CT images," authors Jakob Wasserthal, Manfred Meyer, Hanns-Christian Breit, Joshy Cyriac, Shan Yang, and Martin Segeroth focus on improving the automatic segmentation of multiple anatomical structures in whole-body CT images. They introduce a new dataset and segmentation toolkit that addresses common issues faced by existing algorithms and present an improved workflow for creating ground truth segmentations. By training a segmentation algorithm on this new dataset, they introduce TotalSegmentator—a versatile tool that can easily segment multiple anatomical structures. This work not only addresses key challenges in automatic segmentation but also provides a valuable resource for researchers and practitioners working with CT imaging data. The code for TotalSegmentator is accessible on GitHub (https://github.com/wasserth/TotalSegmentator), while the dataset can be found at https://doi.org/10.5281/zenodo.6802613.