QUBIQ: Uncertainty Quantification for Biomedical Image Segmentation Challenge
Abstract
Uncertainty in medical image segmentation tasks, especially inter-rater variability, arising from differences in interpretations and annotations by various experts, presents a significant challenge in achieving consistent and reliable image segmentation. This variability not only reflects the inherent complexity and subjective nature of medical image interpretation but also directly impacts the development and evaluation of automated segmentation algorithms. Accurately modeling and quantifying this variability is essential for enhancing the robustness and clinical applicability of these algorithms. We report the set-up and summarize the benchmark results of the Quantification of Uncertainties in Biomedical Image Quantification Challenge (QUBIQ), which was organized in conjunction with International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2020 and 2021. The challenge focuses on the uncertainty quantification of medical image segmentation which considers the omnipresence of inter-rater variability in imaging datasets. The large collection of images with multi-rater annotations features various modalities such as MRI and CT; various organs such as the brain, prostate, kidney, and pancreas; and different image dimensions 2D-vs-3D. A total of 24 teams submitted different solutions to the problem, combining various baseline models, Bayesian neural networks, and ensemble model techniques. The obtained results indicate the importance of the ensemble models, as well as the need for further research to develop efficient 3D methods for uncertainty quantification methods in 3D segmentation tasks.
Cite
@article{arxiv.2405.18435,
title = {QUBIQ: Uncertainty Quantification for Biomedical Image Segmentation Challenge},
author = {Hongwei Bran Li and Fernando Navarro and Ivan Ezhov and Amirhossein Bayat and Dhritiman Das and Florian Kofler and Suprosanna Shit and Diana Waldmannstetter and Johannes C. Paetzold and Xiaobin Hu and Benedikt Wiestler and Lucas Zimmer and Tamaz Amiranashvili and Chinmay Prabhakar and Christoph Berger and Jonas Weidner and Michelle Alonso-Basant and Arif Rashid and Ujjwal Baid and Wesam Adel and Deniz Ali and Bhakti Baheti and Yingbin Bai and Ishaan Bhatt and Sabri Can Cetindag and Wenting Chen and Li Cheng and Prasad Dutand and Lara Dular and Mustafa A. Elattar and Ming Feng and Shengbo Gao and Henkjan Huisman and Weifeng Hu and Shubham Innani and Wei Jiat and Davood Karimi and Hugo J. Kuijf and Jin Tae Kwak and Hoang Long Le and Xiang Lia and Huiyan Lin and Tongliang Liu and Jun Ma and Kai Ma and Ting Ma and Ilkay Oksuz and Robbie Holland and Arlindo L. Oliveira and Jimut Bahan Pal and Xuan Pei and Maoying Qiao and Anindo Saha and Raghavendra Selvan and Linlin Shen and Joao Lourenco Silva and Ziga Spiclin and Sanjay Talbar and Dadong Wang and Wei Wang and Xiong Wang and Yin Wang and Ruiling Xia and Kele Xu and Yanwu Yan and Mert Yergin and Shuang Yu and Lingxi Zeng and YingLin Zhang and Jiachen Zhao and Yefeng Zheng and Martin Zukovec and Richard Do and Anton Becker and Amber Simpson and Ender Konukoglu and Andras Jakab and Spyridon Bakas and Leo Joskowicz and Bjoern Menze},
journal= {arXiv preprint arXiv:2405.18435},
year = {2024}
}
Comments
initial technical report