Learn2Reg 2024: New Benchmark Datasets Driving Progress on New Challenges

Lasse Hansen1Orcid, Wiebke Heyer2, Christoph Großbröhmer2, Frederic Madesta3,4, Thilo Sentker3,4, Wang Jiazheng5, Yuxi Zhang5, Hang Zhang6, Min Liu5, Junyi Wang7, Xi Zhu7, Yuhua Li8, Liwen Wang8, Daniil Morozov9,10, Nazim Haouchine9, Joel Honkamaa11, Pekka Marttinen11, Yichao Zhou12, Zuopeng Tan12, Zhuoyuan Wang13, Yi Wang13, Hongchao Zhou14, Shunbo Hu14, Yi Zhang15, Qian Tao15, Lukas Förner16, Thomas Wendler16, Bailiang Jian10,17, Christian Wachinger10,17, Jin Kim18, Dan Ruan18, Marek Wodzinski19,20, Henning Müller21,22, Tony C.W. Mok23, Xi Jia24, Jinming Duan24, Mikael Brudfors25, Seyed-Ahmad Ahmadi26, Yunzheng Zhu27, William Hsu27, Tina Kapur9, William M. Wells9, Alexandra Golby9, Aaron Carass28, Harrison Bai28, Yihao Liu29, Perrine Paul-Gilloteaux30, Joakim Lindblad31, Nataša Sladoje31, Andreas Walter32, Junyu Chen28, Reuben Dorent9,33, Alessa Hering34,35, Mattias P. Heinrich2,35
1: EchoScout GmbH, Lübeck, DE, 2: Institute of Medical Informatics, Universität zu Lübeck, Lübeck, DE, 3: Institute of Applied Medical Informatics, University Medical Center Hamburg-Eppendorf, Hamburg, DE, 4: Institute of Computational Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, DE, 5: School of Artificial Intelligence and Robotics, Hunan University, Changsha, CN, 6: Cornell University, New York, US, 7: University of Electronic Science and Technology of China, Chengdu, CN, 8: Mechanical Engineering Department, Tianjin University, Tianjin, CN, 9: Harvard Medical School, Brigham and Women’s Hospital, Boston, US, 10: Technical University of Munich, Munich, DE, 11: Aalto University, Espoo, FI, 12: Canon Medical Systems (China) Co., Ltd., Beijing, CN, 13: Smart Medical Imaging, Learning and Engineering (SMLE) Lab, School of Biomedical Engineering, Shenzhen University Medical School, Shenzhen University, Shenzhen, CN, 14: School of Information Science and Engineering, Linyi University, Linyi, CN, 15: Department of Imaging Physics, Delft University of Technology, Delft, NL, 16: Clinical Computational Medical Imaging Research, Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Augsburg, Augsburg, DE, 17: TUM Klinikum Rechts der Isar, Munich, DE, 18: University of California, Los Angeles, US, 19: AGH University of Krakow, Krakow, PL, 20: Sano Centre for Computational Medicine, Krakow, PL, 21: Institute of Informatics, University of Applied Sciences Western Switzerland, Sierre, CH, 22: Medical Faculty, University of Geneva, Geneva,CH, 23: Hong Kong University of Science and Technology, Hong Kong, HK, 24: School of Computer Science, University of Birmingham, Birmingham, UK, 25: NVIDIA Ltd., London, UK, 26: NVIDIA GmbH, Munich, DE, 27: Medical & Imaging Informatics, Department of Radiological Science, University of California, Los Angeles, US, 28: Johns Hopkins University, Baltimore, US, 29: Vanderbilt University, Nashville, US, 30: Nantes Université, CHU Nantes, CNRS, Inserm, BioCore, US16, SFR Bonamy, F44000 Nantes, FR, 31: Uppsala Universitet, Uppsala, SE, 32: Aalen University, Aalen, DE, 33: National Institute for Research in Computer Science and Control, Paris, FR, 34: Radboud University Medical Center, Nijmegen, NL, 35: These authors contributed equally to this work and share senior authorship.
Publication date: 2025/12/21
https://doi.org/10.59275/j.melba.2025-gc8c
PDF

Abstract

Medical image registration is critical for clinical applications, and fair benchmarking of different methods is essential for monitoring ongoing progress in the field. To date, the Learn2Reg 2020-2023 challenges have released several complementary datasets and established metrics for evaluations. Building on this foundation, the 2024 edition expands the challenge’s scope to cover a wider range of registration scenarios, particularly in terms of modality diversity and task complexity, by introducing three new tasks, including large-scale multi-modal registration and unsupervised inter-subject brain registration, as well as the first microscopy-focused benchmark within Learn2Reg. The new datasets also inspired new method developments, including invertibility constraints, pyramid features, keypoints alignment and instance optimisation.

Keywords

Data Challenges · (Bio-) Medical Image Registration

Bibtex @article{melba:2025:034:hansen, title = "Learn2Reg 2024: New Benchmark Datasets Driving Progress on New Challenges", author = "Hansen, Lasse and Heyer, Wiebke and Großbröhmer, Christoph and Madesta, Frederic and Sentker, Thilo and Jiazheng, Wang and Zhang, Yuxi and Zhang, Hang and Liu, Min and Wang, Junyi and Zhu, Xi and Li, Yuhua and Wang, Liwen and Morozov, Daniil and Haouchine, Nazim and Honkamaa, Joel and Marttinen, Pekka and Zhou, Yichao and Tan, Zuopeng and Wang, Zhuoyuan and Wang, Yi and Zhou, Hongchao and Hu, Shunbo and Zhang, Yi and Tao, Qian and Förner, Lukas and Wendler, Thomas and Jian, Bailiang and Wachinger, Christian and Kim, Jin and Ruan, Dan and Wodzinski, Marek and Müller, Henning and Mok, Tony C.W. and Jia, Xi and Duan, Jinming and Brudfors, Mikael and Ahmadi, Seyed-Ahmad and Zhu, Yunzheng and Hsu, William and Kapur, Tina and Wells, William M. and Golby, Alexandra and Carass, Aaron and Bai, Harrison and Liu, Yihao and Paul-Gilloteaux, Perrine and Lindblad, Joakim and Sladoje, Nataša and Walter, Andreas and Chen, Junyu and Dorent, Reuben and Hering, Alessa and Heinrich, Mattias P.", journal = "Machine Learning for Biomedical Imaging", volume = "3", issue = "December 2025 issue", year = "2025", pages = "775--791", issn = "2766-905X", doi = "https://doi.org/10.59275/j.melba.2025-gc8c", url = "https://melba-journal.org/2025:034" }
RISTY - JOUR AU - Hansen, Lasse AU - Heyer, Wiebke AU - Großbröhmer, Christoph AU - Madesta, Frederic AU - Sentker, Thilo AU - Jiazheng, Wang AU - Zhang, Yuxi AU - Zhang, Hang AU - Liu, Min AU - Wang, Junyi AU - Zhu, Xi AU - Li, Yuhua AU - Wang, Liwen AU - Morozov, Daniil AU - Haouchine, Nazim AU - Honkamaa, Joel AU - Marttinen, Pekka AU - Zhou, Yichao AU - Tan, Zuopeng AU - Wang, Zhuoyuan AU - Wang, Yi AU - Zhou, Hongchao AU - Hu, Shunbo AU - Zhang, Yi AU - Tao, Qian AU - Förner, Lukas AU - Wendler, Thomas AU - Jian, Bailiang AU - Wachinger, Christian AU - Kim, Jin AU - Ruan, Dan AU - Wodzinski, Marek AU - Müller, Henning AU - Mok, Tony C.W. AU - Jia, Xi AU - Duan, Jinming AU - Brudfors, Mikael AU - Ahmadi, Seyed-Ahmad AU - Zhu, Yunzheng AU - Hsu, William AU - Kapur, Tina AU - Wells, William M. AU - Golby, Alexandra AU - Carass, Aaron AU - Bai, Harrison AU - Liu, Yihao AU - Paul-Gilloteaux, Perrine AU - Lindblad, Joakim AU - Sladoje, Nataša AU - Walter, Andreas AU - Chen, Junyu AU - Dorent, Reuben AU - Hering, Alessa AU - Heinrich, Mattias P. PY - 2025 TI - Learn2Reg 2024: New Benchmark Datasets Driving Progress on New Challenges T2 - Machine Learning for Biomedical Imaging VL - 3 IS - December 2025 issue SP - 775 EP - 791 SN - 2766-905X DO - https://doi.org/10.59275/j.melba.2025-gc8c UR - https://melba-journal.org/2025:034 ER -
2025:034 cover