Fast and low-GPU-memory abdomen CT organ segmentation: The FLARE challenge
Jun Ma
(1)
,
Yao Zhang
(2)
,
Song Gu
(3)
,
Xingle An
(4)
,
Zhihe Wang
(5)
,
Cheng Ge
(6)
,
Congcong Wang
(7, 8)
,
Fan Zhang
(9)
,
Yu Wang
(9)
,
Yinan Xu
(10)
,
Shuiping Gou
(10)
,
Franz Thaler
(11, 12)
,
Christian Payer
(12)
,
Darko Štern
(11)
,
Edward Henderson
(13, 14)
,
Donal Mcsweeney
(13, 14)
,
Andrew Green
(13, 14)
,
Price Jackson
(15)
,
Lachlan Mcintosh
(16)
,
Quoc-Cuong Nguyen
(17)
,
Abdul Qayyum
(18)
,
Pierre-Henri Conze
(19, 20)
,
Ziyan Huang
(21)
,
Ziqi Zhout
(22)
,
Deng-Ping Fan
(23, 24)
,
Huan Xiong
(24, 25)
,
Guoqiang Dong
(26)
,
Qiongjie Zhu
(26)
,
Jian He
(26)
,
Xiaoping Yang
(27)
1
NJUST -
Nanjing University of Science and Technology
2 ICT - CAS Institute of Computing Technology
3 Nanjing Anke Medical Technology
4 Infervision Technology
5 Shenzhen Haichuang Medical
6 Jiangsu University of Technology [Changzhou]
7 TUST - Tianjin University of Science and Technology
8 NTNU - Norwegian University of Science and Technology
9 Fosun Aitrox Information Technology
10 Xidian University
11 Medical University of Graz
12 TU Graz - Graz University of Technology [Graz]
13 University of Manchester [Manchester]
14 The Christie NHS Foundation Trust [Manchester, Royaume-Uni]
15 Peter MacCallum Cancer Centre
16 Peter MacCallum Cancer Center
17 University of Information Technology, Vietnam National University - HCMC
18 Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance
19 IMT Atlantique - ITI - Département lmage et Traitement Information
20 LaTIM - Laboratoire de Traitement de l'Information Medicale
21 Shanghai Jiao Tong University [Shanghai]
22 Shenzhen University
23 NKU - Nankai University
24 IIAI - Inception Institute of Artificial Intelligence [Abu Dhabi]
25 HIT - Harbin Institute of Technology
26 Medical School (University of Nanjing)
27 NJU - Nanjing University
2 ICT - CAS Institute of Computing Technology
3 Nanjing Anke Medical Technology
4 Infervision Technology
5 Shenzhen Haichuang Medical
6 Jiangsu University of Technology [Changzhou]
7 TUST - Tianjin University of Science and Technology
8 NTNU - Norwegian University of Science and Technology
9 Fosun Aitrox Information Technology
10 Xidian University
11 Medical University of Graz
12 TU Graz - Graz University of Technology [Graz]
13 University of Manchester [Manchester]
14 The Christie NHS Foundation Trust [Manchester, Royaume-Uni]
15 Peter MacCallum Cancer Centre
16 Peter MacCallum Cancer Center
17 University of Information Technology, Vietnam National University - HCMC
18 Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance
19 IMT Atlantique - ITI - Département lmage et Traitement Information
20 LaTIM - Laboratoire de Traitement de l'Information Medicale
21 Shanghai Jiao Tong University [Shanghai]
22 Shenzhen University
23 NKU - Nankai University
24 IIAI - Inception Institute of Artificial Intelligence [Abu Dhabi]
25 HIT - Harbin Institute of Technology
26 Medical School (University of Nanjing)
27 NJU - Nanjing University
Abdul Qayyum
- Function : Author
- PersonId : 800447
- ORCID : 0000-0003-3102-1595
Pierre-Henri Conze
- Function : Author
- PersonId : 174700
- IdHAL : pierre-henri-conze
- ORCID : 0000-0003-2214-3654
- IdRef : 238573486
Huan Xiong
- Function : Author
- PersonId : 1047154
Abstract
Automatic segmentation of abdominal organs in CT scans plays an important role in clinical practice. However, most existing benchmarks and datasets only focus on segmentation accuracy, while the model efficiency and its accuracy on the testing cases from different medical centers have not been evaluated. To comprehensively benchmark abdominal organ segmentation methods, we organized the first Fast and Low GPU memory Abdominal oRgan sEgmentation (FLARE) challenge, where the segmentation methods were encouraged to achieve high accuracy on the testing cases from different medical centers, fast inference speed, and low GPU memory consumption, simultaneously. The winning method surpassed the existing state-of-the-art method, achieving a faster inference speed and reducing the GPU memory consumption by 60% with comparable accuracy. We provide a summary of the top methods, make their code and Docker containers publicly available, and give practical suggestions on building accurate and efficient abdominal organ segmentation models. The FLARE challenge remains open for future submissions through a live platform for benchmarking further methodology developments at https://flare.grand-challenge.org/.