Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge
Computer Vision and Pattern Recognition
2019-04-24 v3 Artificial Intelligence
Machine Learning
Machine Learning
Abstract
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.
Keywords
Cite
@article{arxiv.1811.02629,
title = {Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge},
author = {Spyridon Bakas and Mauricio Reyes and Andras Jakab and Stefan Bauer and Markus Rempfler and Alessandro Crimi and Russell Takeshi Shinohara and Christoph Berger and Sung Min Ha and Martin Rozycki and Marcel Prastawa and Esther Alberts and Jana Lipkova and John Freymann and Justin Kirby and Michel Bilello and Hassan Fathallah-Shaykh and Roland Wiest and Jan Kirschke and Benedikt Wiestler and Rivka Colen and Aikaterini Kotrotsou and Pamela Lamontagne and Daniel Marcus and Mikhail Milchenko and Arash Nazeri and Marc-Andre Weber and Abhishek Mahajan and Ujjwal Baid and Elizabeth Gerstner and Dongjin Kwon and Gagan Acharya and Manu Agarwal and Mahbubul Alam and Alberto Albiol and Antonio Albiol and Francisco J. Albiol and Varghese Alex and Nigel Allinson and Pedro H. A. Amorim and Abhijit Amrutkar and Ganesh Anand and Simon Andermatt and Tal Arbel and Pablo Arbelaez and Aaron Avery and Muneeza Azmat and Pranjal B. and W Bai and Subhashis Banerjee and Bill Barth and Thomas Batchelder and Kayhan Batmanghelich and Enzo Battistella and Andrew Beers and Mikhail Belyaev and Martin Bendszus and Eze Benson and Jose Bernal and Halandur Nagaraja Bharath and George Biros and Sotirios Bisdas and James Brown and Mariano Cabezas and Shilei Cao and Jorge M. Cardoso and Eric N Carver and Adrià Casamitjana and Laura Silvana Castillo and Marcel Catà and Philippe Cattin and Albert Cerigues and Vinicius S. Chagas and Siddhartha Chandra and Yi-Ju Chang and Shiyu Chang and Ken Chang and Joseph Chazalon and Shengcong Chen and Wei Chen and Jefferson W Chen and Zhaolin Chen and Kun Cheng and Ahana Roy Choudhury and Roger Chylla and Albert Clérigues and Steven Colleman and Ramiro German Rodriguez Colmeiro and Marc Combalia and Anthony Costa and Xiaomeng Cui and Zhenzhen Dai and Lutao Dai and Laura Alexandra Daza and Eric Deutsch and Changxing Ding and Chao Dong and Shidu Dong and Wojciech Dudzik and Zach Eaton-Rosen and Gary Egan and Guilherme Escudero and Théo Estienne and Richard Everson and Jonathan Fabrizio and Yong Fan and Longwei Fang and Xue Feng and Enzo Ferrante and Lucas Fidon and Martin Fischer and Andrew P. French and Naomi Fridman and Huan Fu and David Fuentes and Yaozong Gao and Evan Gates and David Gering and Amir Gholami and Willi Gierke and Ben Glocker and Mingming Gong and Sandra González-Villá and T. Grosges and Yuanfang Guan and Sheng Guo and Sudeep Gupta and Woo-Sup Han and Il Song Han and Konstantin Harmuth and Huiguang He and Aura Hernández-Sabaté and Evelyn Herrmann and Naveen Himthani and Winston Hsu and Cheyu Hsu and Xiaojun Hu and Xiaobin Hu and Yan Hu and Yifan Hu and Rui Hua and Teng-Yi Huang and Weilin Huang and Sabine Van Huffel and Quan Huo and Vivek HV and Khan M. Iftekharuddin and Fabian Isensee and Mobarakol Islam and Aaron S. Jackson and Sachin R. Jambawalikar and Andrew Jesson and Weijian Jian and Peter Jin and V Jeya Maria Jose and Alain Jungo and B Kainz and Konstantinos Kamnitsas and Po-Yu Kao and Ayush Karnawat and Thomas Kellermeier and Adel Kermi and Kurt Keutzer and Mohamed Tarek Khadir and Mahendra Khened and Philipp Kickingereder and Geena Kim and Nik King and Haley Knapp and Urspeter Knecht and Lisa Kohli and Deren Kong and Xiangmao Kong and Simon Koppers and Avinash Kori and Ganapathy Krishnamurthi and Egor Krivov and Piyush Kumar and Kaisar Kushibar and Dmitrii Lachinov and Tryphon Lambrou and Joon Lee and Chengen Lee and Yuehchou Lee and M Lee and Szidonia Lefkovits and Laszlo Lefkovits and James Levitt and Tengfei Li and Hongwei Li and Wenqi Li and Hongyang Li and Xiaochuan Li and Yuexiang Li and Heng Li and Zhenye Li and Xiaoyu Li and Zeju Li and XiaoGang Li and Wenqi Li and Zheng-Shen Lin and Fengming Lin and Pietro Lio and Chang Liu and Boqiang Liu and Xiang Liu and Mingyuan Liu and Ju Liu and Luyan Liu and Xavier Llado and Marc Moreno Lopez and Pablo Ribalta Lorenzo and Zhentai Lu and Lin Luo and Zhigang Luo and Jun Ma and Kai Ma and Thomas Mackie and Anant Madabushi and Issam Mahmoudi and Klaus H. Maier-Hein and Pradipta Maji and CP Mammen and Andreas Mang and B. S. Manjunath and Michal Marcinkiewicz and S McDonagh and Stephen McKenna and Richard McKinley and Miriam Mehl and Sachin Mehta and Raghav Mehta and Raphael Meier and Christoph Meinel and Dorit Merhof and Craig Meyer and Robert Miller and Sushmita Mitra and Aliasgar Moiyadi and David Molina-Garcia and Miguel A. B. Monteiro and Grzegorz Mrukwa and Andriy Myronenko and Jakub Nalepa and Thuyen Ngo and Dong Nie and Holly Ning and Chen Niu and Nicholas K Nuechterlein and Eric Oermann and Arlindo Oliveira and Diego D. C. Oliveira and Arnau Oliver and Alexander F. I. Osman and Yu-Nian Ou and Sebastien Ourselin and Nikos Paragios and Moo Sung Park and Brad Paschke and J. Gregory Pauloski and Kamlesh Pawar and Nick Pawlowski and Linmin Pei and Suting Peng and Silvio M. Pereira and Julian Perez-Beteta and Victor M. Perez-Garcia and Simon Pezold and Bao Pham and Ashish Phophalia and Gemma Piella and G. N. Pillai and Marie Piraud and Maxim Pisov and Anmol Popli and Michael P. Pound and Reza Pourreza and Prateek Prasanna and Vesna Prkovska and Tony P. Pridmore and Santi Puch and Élodie Puybareau and Buyue Qian and Xu Qiao and Martin Rajchl and Swapnil Rane and Michael Rebsamen and Hongliang Ren and Xuhua Ren and Karthik Revanuru and Mina Rezaei and Oliver Rippel and Luis Carlos Rivera and Charlotte Robert and Bruce Rosen and Daniel Rueckert and Mohammed Safwan and Mostafa Salem and Joaquim Salvi and Irina Sanchez and Irina Sánchez and Heitor M. Santos and Emmett Sartor and Dawid Schellingerhout and Klaudius Scheufele and Matthew R. Scott and Artur A. Scussel and Sara Sedlar and Juan Pablo Serrano-Rubio and N. Jon Shah and Nameetha Shah and Mazhar Shaikh and B. Uma Shankar and Zeina Shboul and Haipeng Shen and Dinggang Shen and Linlin Shen and Haocheng Shen and Varun Shenoy and Feng Shi and Hyung Eun Shin and Hai Shu and Diana Sima and M Sinclair and Orjan Smedby and James M. Snyder and Mohammadreza Soltaninejad and Guidong Song and Mehul Soni and Jean Stawiaski and Shashank Subramanian and Li Sun and Roger Sun and Jiawei Sun and Kay Sun and Yu Sun and Guoxia Sun and Shuang Sun and Yannick R Suter and Laszlo Szilagyi and Sanjay Talbar and Dacheng Tao and Dacheng Tao and Zhongzhao Teng and Siddhesh Thakur and Meenakshi H Thakur and Sameer Tharakan and Pallavi Tiwari and Guillaume Tochon and Tuan Tran and Yuhsiang M. Tsai and Kuan-Lun Tseng and Tran Anh Tuan and Vadim Turlapov and Nicholas Tustison and Maria Vakalopoulou and Sergi Valverde and Rami Vanguri and Evgeny Vasiliev and Jonathan Ventura and Luis Vera and Tom Vercauteren and C. A. Verrastro and Lasitha Vidyaratne and Veronica Vilaplana and Ajeet Vivekanandan and Guotai Wang and Qian Wang and Chiatse J. Wang and Weichung Wang and Duo Wang and Ruixuan Wang and Yuanyuan Wang and Chunliang Wang and Guotai Wang and Ning Wen and Xin Wen and Leon Weninger and Wolfgang Wick and Shaocheng Wu and Qiang Wu and Yihong Wu and Yong Xia and Yanwu Xu and Xiaowen Xu and Peiyuan Xu and Tsai-Ling Yang and Xiaoping Yang and Hao-Yu Yang and Junlin Yang and Haojin Yang and Guang Yang and Hongdou Yao and Xujiong Ye and Changchang Yin and Brett Young-Moxon and Jinhua Yu and Xiangyu Yue and Songtao Zhang and Angela Zhang and Kun Zhang and Xuejie Zhang and Lichi Zhang and Xiaoyue Zhang and Yazhuo Zhang and Lei Zhang and Jianguo Zhang and Xiang Zhang and Tianhao Zhang and Sicheng Zhao and Yu Zhao and Xiaomei Zhao and Liang Zhao and Yefeng Zheng and Liming Zhong and Chenhong Zhou and Xiaobing Zhou and Fan Zhou and Hongtu Zhu and Jin Zhu and Ying Zhuge and Weiwei Zong and Jayashree Kalpathy-Cramer and Keyvan Farahani and Christos Davatzikos and Koen van Leemput and Bjoern Menze},
journal= {arXiv preprint arXiv:1811.02629},
year = {2019}
}
Comments
The International Multimodal Brain Tumor Segmentation (BraTS) Challenge