Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge
Aaron Avery, Aaron S. Jackson, Abhijit Amrutkar, Abhishek Mahajan, Adel Kermi, Adri\`a Casamitjana, Ahana Roy Choudhury, Aikaterini Kotrotsou, Ajeet Vivekanandan, Alain Jungo, Albert Cerigues, Albert Cl\'erigues, Alberto Albiol, Alessandro Crimi, Alexander F. I. Osman, Aliasgar Moiyadi, Amir Gholami, Anant Madabushi, Andras Jakab, Andreas Mang, Andrew Beers, Andrew Jesson, Andrew P. French, Andriy Myronenko, Angela Zhang, Anmol Popli, Anthony Costa, Antonio Albiol, Arash Nazeri, Arlindo Oliveira, Arnau Oliver, Artur A. Scussel, Ashish Phophalia, Aura Hern\'andez-Sabat\'e, Avinash Kori, Ayush Karnawat, Bao Pham, Benedikt Wiestler, Ben Glocker, Bill Barth, Bjoern Menze, B Kainz, Boqiang Liu, Brad Paschke, Brett Young-Moxon, Bruce Rosen, B. S. Manjunath, B. Uma Shankar, Buyue Qian, C. A. Verrastro, Changchang Yin, Chang Liu, Changxing Ding, Chao Dong, Charlotte Robert, Chengen Lee, Chenhong Zhou, Chen Niu, Cheyu Hsu, Chiatse J. Wang, Christoph Berger, Christoph Meinel, Christos Davatzikos, Chunliang Wang, CP Mammen, Craig Meyer, Dacheng Tao, Daniel Marcus, Daniel Rueckert, David Fuentes, David Gering, David Molina-Garcia, Dawid Schellingerhout, Deren Kong, Diana Sima, Diego D. C. Oliveira, Dinggang Shen, Dmitrii Lachinov, Dongjin Kwon, Dong Nie, Dorit Merhof, Duo Wang, Egor Krivov, Elizabeth Gerstner, \'Elodie Puybareau, Emmett Sartor, Enzo Battistella, Enzo Ferrante, Eric Deutsch, Eric N Carver, Eric Oermann, Esther Alberts, Evan Gates, Evelyn Herrmann, Evgeny Vasiliev, Eze Benson, Fabian Isensee, Fan Zhou, Fengming Lin, Feng Shi, Francisco J. Albiol, Gagan Acharya, Ganapathy Krishnamurthi, Ganesh Anand, Gary Egan, Geena Kim, Gemma Piella, George Biros, G.N. Pillai, Grzegorz Mrukwa, Guang Yang, Guidong Song, Guilherme Escudero, Guillaume Tochon, Guotai Wang, Guoxia Sun, Haipeng Shen, Hai Shu, Halandur Nagaraja Bharath, Haley Knapp, Haocheng Shen, Haojin Yang, Hao-Yu Yang, Hassan Fathallah-Shaykh, Heitor M. Santos, Heng Li, Holly Ning, Hongdou Yao, Hongliang Ren, Hongtu Zhu, Hongwei Li, Hongyang Li, Huan Fu, Huiguang He, Hyung Eun Shin, Il Song Han, Irina S\'anchez, Irina Sanchez, Issam Mahmoudi, Jakub Nalepa, James Brown, James Levitt, James M. Snyder, Jana Lipkova, Jan Kirschke, Jayashree Kalpathy-Cramer, Jean Stawiaski, Jefferson W Chen, J. Gregory Pauloski, Jianguo Zhang, Jiawei Sun, Jinhua Yu, Jin Zhu, Joaquim Salvi, John Freymann, Jonathan Fabrizio, Jonathan Ventura, Joon Lee, Jorge M. Cardoso, Jose Bernal, Joseph Chazalon, Juan Pablo Serrano-Rubio, Julian Perez-Beteta, Ju Liu, Junlin Yang, Jun Ma, Justin Kirby, Kai Ma, Kaisar Kushibar, Kamlesh Pawar, Karthik Revanuru, Kayhan Batmanghelich, Kay Sun, Ken Chang, Keyvan Farahani, Khan M. Iftekharuddin, Klaudius Scheufele, Klaus H. Maier-Hein, Koen Van Leemput, Konstantin Harmuth, Konstantinos Kamnitsas, Kuan-Lun Tseng, Kun Cheng, Kun Zhang, Kurt Keutzer, Lasitha Vidyaratne, Laszlo Lefkovits, Laszlo Szilagyi, Laura Alexandra Daza, Laura Silvana Castillo, Lei Zhang, Leon Weninger, Liang Zhao, Lichi Zhang, Liming Zhong, Linlin Shen, Lin Luo, Linmin Pei, Lisa Kohli, Li Sun, Longwei Fang, Lucas Fidon, Luis Carlos Rivera, Luis Vera, Lutao Dai, Luyan Liu, Mahbubul Alam, Mahendra Khened, Manu Agarwal, Marc-Andre Weber, Marc Combalia, Marcel Cat\`a, Marcel Prastawa, Marc Moreno Lopez, Mariano Cabezas, Maria Vakalopoulou, Marie Piraud, Markus Rempfler, Martin Bendszus, Martin Fischer, Martin Rajchl, Martin Rozycki, Matthew R. Scott, Mauricio Reyes, Maxim Pisov, Mazhar Shaikh, Meenakshi H Thakur, Mehul Soni, Michael P. Pound, Michael Rebsamen, Michal Marcinkiewicz, Michel Bilello, Miguel A.B. Monteiro, Mikhail Belyaev, Mikhail Milchenko, Mina Rezaei, Mingming Gong, Mingyuan Liu, Miriam Mehl, M Lee, Mobarakol Islam, Mohamed Tarek Khadir, Mohammadreza Soltaninejad, Mohammed Safwan, Moo Sung Park, Mostafa Salem, M Sinclair, Muneeza Azmat, Nameetha Shah, Naomi Fridman, Naveen Himthani, Nicholas K Nuechterlein, Nicholas Tustison, Nick Pawlowski, Nigel Allinson, Nik King, Nikos Paragios, Ning Wen, N. Jon Shah, Oliver Rippel, Orjan Smedby, Pablo Arbelaez, Pablo Ribalta Lorenzo, Pallavi Tiwari, Pamela Lamontagne, Pedro H. A. Amorim, Peiyuan Xu, Peter Jin, Philippe Cattin, Philipp Kickingereder, Pietro Lio, Piyush Kumar, Po-Yu Kao, Pradipta Maji, Pranjal B., Prateek Prasanna, Qiang Wu, Qian Wang, Quan Huo, Raghav Mehta, Ramiro German Rodriguez Colmeiro, Rami Vanguri, Raphael Meier, Reza Pourreza, Richard Everson, Richard McKinley, Rivka Colen, Robert Miller, Roger Chylla, Roger Sun, Roland Wiest, Rui Hua, Ruixuan Wang, Russell Takeshi Shinohara, Sabine Van Huffel, Sachin Mehta, Sachin R. Jambawalikar, Sameer Tharakan, Sandra Gonz\'alez-Vill\'a, Sanjay Talbar, Santi Puch, Sara Sedlar, Sebastien Ourselin, Sergi Valverde, Shaocheng Wu, Shashank Subramanian, Shengcong Chen, Sheng Guo, Shidu Dong, Shilei Cao, Shiyu Chang, Shuang Sun, Sicheng Zhao, Siddhartha Chandra, Siddhesh Thakur, Silvio M. Pereira, Simon Andermatt, Simon Koppers, Simon Pezold, S McDonagh, Songtao Zhang, Sotirios Bisdas, Spyridon Bakas, Stefan Bauer, Stephen McKenna, Steven Colleman, Subhashis Banerjee, Sudeep Gupta, Sung Min Ha, Sushmita Mitra, Suting Peng, Swapnil Rane, Szidonia Lefkovits, Tal Arbel, Tengfei Li, Teng-Yi Huang, T. Grosges, Th\'eo Estienne, Thomas Batchelder, Thomas Kellermeier, Thomas Mackie, Thuyen Ngo, Tianhao Zhang, Tom Vercauteren, Tony P. Pridmore, Tran Anh Tuan, Tryphon Lambrou, Tsai-Ling Yang, Tuan Tran, Ujjwal Baid, Urspeter Knecht, Vadim Turlapov, Varghese Alex, Varun Shenoy, Veronica Vilaplana, Vesna Prkovska, Victor M. Perez-Garcia, Vinicius S. Chagas, Vivek HV, V Jeya Maria Jose, W Bai, Wei Chen, Weichung Wang, Weijian Jian, Weilin Huang, Weiwei Zong, Wenqi Li, Willi Gierke, Winston Hsu, Wojciech Dudzik, Wolfgang Wick, Woo-Sup Han, Xavier Llado, Xiang Liu, Xiangmao Kong, Xiangyu Yue, Xiang Zhang, Xiaobing Zhou, Xiaobin Hu, Xiaochuan Li, Xiaogang Li, Xiaojun Hu, Xiaomei Zhao, Xiaomeng Cui, Xiaoping Yang, Xiaowen Xu, Xiaoyue Zhang, Xiaoyu Li, Xin Wen, Xue Feng, Xuejie Zhang, Xuhua Ren, Xujiong Ye, Xu Qiao, Yan Hu, Yannick R Suter, Yanwu Xu, Yaozong Gao, Yazhuo Zhang, Yefeng Zheng, Yifan Hu, Yihong Wu, Yi-Ju Chang, Ying Zhuge, Yong Fan, Yong Xia, Yuanfang Guan, Yuanyuan Wang, Yuehchou Lee, Yuexiang Li, Yuhsiang M. Tsai, Yu-Nian Ou, Yu Sun, Yu Zhao, Zach Eaton-Rosen, Zeina Shboul, Zeju Li, Zhaolin Chen, Zheng-Shen Lin, Zhentai Lu, Zhenye Li, Zhenzhen Dai, Zhigang Luo, Zhongzhao Teng
Gliomas are the most common primary brain malignancies, with different degrees of aggressiveness, variable prognosis and various heterogeneous histologic sub-regions, i.e., peritumoral edematous/invaded tissue, necrotic core, active and non-enhancing core. This intrinsic heterogeneity is also portrayed in their radio-phenotype, as their sub-regions are depicted by varying intensity profiles disseminated across multi-parametric magnetic resonance imaging (mpMRI) scans, reflecting varying biological properties. Their heterogeneous shape, extent, and location are some of the factors that make these tumors difficult to resect, and in some cases inoperable. The amount of resected tumor is a factor also considered in longitudinal scans, when evaluating the apparent tumor for potential diagnosis of progression. Furthermore, there is mounting evidence that accurate segmentation of the various tumor sub-regions can offer the basis for quantitative image analysis towards prediction of patient overall survival. This study assesses the state-of-the-art machine learning (ML) methods used for brain tumor image analysis in mpMRI scans, during the last seven instances of the International Brain Tumor Segmentation (BraTS) challenge, i.e., 2012-2018. Specifically, we focus on i) evaluating segmentations of the various glioma sub-regions in pre-operative mpMRI scans, ii) assessing potential tumor progression by virtue of longitudinal growth of tumor sub-regions, beyond use of the RECIST/RANO criteria, and iii) predicting the overall survival from pre-operative mpMRI scans of patients that underwent gross total resection. Finally, we investigate the challenge of identifying the best ML algorithms for each of these tasks, considering that apart from being diverse on each instance of the challenge, the multi-institutional mpMRI BraTS dataset has also been a continuously evolving/growing dataset.
MedFL-Stress shows that FedBN reduces the best-to-worst hospital Dice gap by 41% versus FedAvg in federated brain tumor segmentation under simulated cross-hospital MRI shifts, improving the weakest site by 3.5 points ...
HiPerfGNN uses VQ-VAE on DSC perfusion curves to form hierarchical tumor habitat graphs that predict IDH mutation (AUC 0.96 internal, 0.89 external), 1p/19q codeletion, and WHO grade.
MedRCube is a new fine-grained evaluation framework that benchmarks 33 MLLMs on medical imaging, ranks Lingshu-32B highest, and finds a significant positive link between shortcut behaviors and diagnostic performance.
TaCo contrastively embeds semantic, generative, and transformation tasks from medical imaging into a joint space to reveal which tasks cluster, blend, or remain distinct.
Multimodal glioma survival models achieve better performance through additive integration of image and RNA features rather than cross-modal synergies, as quantified by lower measured interactions in stronger architectures.
Patient identity and clinical features predict brain tumor segmentation accuracy more strongly than model choice, with localized spatial biases consistent across models and no formal fairness guarantees in any.
MK-ResRecon predicts missing slices with a multi-kernel texture-aware loss while IdentityRefineNet3D refines the combined 3D volume to enable accurate reconstruction from highly sparse 2D inputs.
InfiltrNet uses a CNN-Transformer hybrid with distance-transform labels to generate infiltration risk maps from MRI and outperforms baselines on BraTS 2020 and 2025 datasets.
UniME combines a pretrained unified ViT encoder with modality-specific CNN encoders to improve brain tumor segmentation performance when some MRI modalities are missing.
A multi-dataset cross-domain knowledge distillation approach improves unified performance on medical image segmentation, classification, and detection by transferring domain-invariant features from a joint teacher mod...
ADRUwAMS reports Dice scores of 0.9229 (whole tumor), 0.8432 (tumor core), and 0.8004 (enhancing tumor) on BraTS 2020 after training on BraTS 2019/2020 datasets.
BraTS 2021 provides a 2,040-patient mpMRI benchmark for brain tumor sub-region segmentation and MGMT methylation classification, hosted on Synapse and Kaggle with $60,000 in awards.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.