Hering Alessa, Hansen Lasse, Mok Tony C W, Chung Albert C S, Siebert Hanna, Hager Stephanie, Lange Annkristin, Kuckertz Sven, Heldmann Stefan, Shao Wei, Vesal Sulaiman, Rusu Mirabela, Sonn Geoffrey, Estienne Theo, Vakalopoulou Maria, Han Luyi, Huang Yunzhi, Yap Pew-Thian, Brudfors Mikael, Balbastre Yael, Joutard Samuel, Modat Marc, Lifshitz Gal, Raviv Dan, Lv Jinxin, Li Qiang, Jaouen Vincent, Visvikis Dimitris, Fourcade Constance, Rubeaux Mathieu, Pan Wentao, Xu Zhe, Jian Bailiang, De Benetti Francesca, Wodzinski Marek, Gunnarsson Niklas, Sjolund Jens, Grzech Daniel, Qiu Huaqi, Li Zeju, Thorley Alexander, Duan Jinming, Grosbrohmer Christoph, Hoopes Andrew, Reinertsen Ingerid, Xiao Yiming, Landman Bennett, Huo Yuankai, Murphy Keelin, Lessmann Nikolas, van Ginneken Bram, Dalca Adrian V, Heinrich Mattias P
IEEE Trans Med Imaging. 2023 Mar;42(3):697-712. doi: 10.1109/TMI.2022.3213983. Epub 2023 Mar 2.
Image registration is a fundamental medical image analysis task, and a wide variety of approaches have been proposed. However, only a few studies have comprehensively compared medical image registration approaches on a wide range of clinically relevant tasks. This limits the development of registration methods, the adoption of research advances into practice, and a fair benchmark across competing approaches. The Learn2Reg challenge addresses these limitations by providing a multi-task medical image registration data set for comprehensive characterisation of deformable registration algorithms. A continuous evaluation will be possible at https://learn2reg.grand-challenge.org. Learn2Reg covers a wide range of anatomies (brain, abdomen, and thorax), modalities (ultrasound, CT, MR), availability of annotations, as well as intra- and inter-patient registration evaluation. We established an easily accessible framework for training and validation of 3D registration methods, which enabled the compilation of results of over 65 individual method submissions from more than 20 unique teams. We used a complementary set of metrics, including robustness, accuracy, plausibility, and runtime, enabling unique insight into the current state-of-the-art of medical image registration. This paper describes datasets, tasks, evaluation methods and results of the challenge, as well as results of further analysis of transferability to new datasets, the importance of label supervision, and resulting bias. While no single approach worked best across all tasks, many methodological aspects could be identified that push the performance of medical image registration to new state-of-the-art performance. Furthermore, we demystified the common belief that conventional registration methods have to be much slower than deep-learning-based methods.
图像配准是一项基础的医学图像分析任务，人们已经提出了各种各样的方法。然而，只有少数研究在广泛的临床相关任务中对医学图像配准方法进行了全面比较。这限制了配准方法的发展、研究成果在实践中的应用以及对各种竞争方法的公平基准测试。Learn2Reg挑战赛通过提供一个多任务医学图像配准数据集来全面表征可变形配准算法，从而解决了这些限制。在https://learn2reg.grand-challenge.org上可以进行持续评估。Learn2Reg涵盖了广泛的解剖结构（大脑、腹部和胸部）、模态（超声、CT、MR）、注释的可用性以及患者内和患者间的配准评估。我们建立了一个易于访问的框架来训练和验证3D配准方法，这使得能够汇总来自20多个不同团队的65个以上单独方法提交的结果。我们使用了一套互补的指标，包括鲁棒性、准确性、合理性和运行时间，从而能够对医学图像配准的当前技术水平有独特的见解。本文描述了挑战赛的数据集、任务、评估方法和结果，以及对新数据集可转移性的进一步分析结果、标签监督的重要性和由此产生的偏差。虽然没有一种方法在所有任务中都表现最佳，但可以确定许多方法学方面的因素将医学图像配准的性能提升到了新的技术水平。此外，我们揭开了传统配准方法一定比基于深度学习的方法慢得多这一普遍观念的神秘面纱。
