Identifying the Best Machine Learning Algorithms for Brain Tumor Segmentation, Progression Assessment, and Overall Survival Prediction in the BRATS Challenge

By S. Bakas, Mauricio Reyes, A. Jakab, Stefan Bauer, Markus Rempfler, Alessandro Crimi, R. Shinohara, Christoph Berger, S. Ha, Martin Rozycki and 41 more. arXiv.org, 2018

2,033 citations recorded by the source corpus.

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 tumoris 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 thestate-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 tota lresection. 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.

Subfields: Genetics, Neurology, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/2cebf0212380f83b7171fb5660f842c8d7043f60

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Genetics

In Neurology

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Federated learning in medicine: facilitating multi-institutional collaborations without sharing patient data (2020, methodology): “We used the BraTS 2017 training dataset 4–6, 21 , 22 to The radiographically abnormal regions of each image were annotated and approved by multiple clinical experts at each contributing institution following a pre-defined annotation protocol.”
3D MRI brain tumor segmentation using autoencoder regularization (2018, methodology): “rt methods for the segmentation of brain tumors by providing a 3D MRI dataset with ground truth tumor segmentation labels annotated arXiv:1810.11654v3 [cs.CV] 19 Nov 2018 2 A. Myronenko by physicians [5,18,4,2,3]. This year, BraTS 2018 training dataset included 285 cases (210 HGG and 75 LGG), each w”
CaraNet: context axial reverse attention network for segmentation of small medical objects (2021, methodology): “The brain tumor dataset was created from BraTS 2018 [19] database by slicing 2D images from the “T1ce” source with “NET” type labels.”
Diffusion Models for Implicit Image Segmentation Ensembles (2021, methodology): “In this work, we focus on the BRATS2020 brain tumor segmentation challenge (Menze et al., 2014; Bakas et al., 2017, 2018).”
The Medical Segmentation Decathlon (2021, methodology): “We therefore decided to use a ranking approach similar to the evaluation of the popular BraTS challenge, 14 which was based on a Wilcoxon-signed rank pairwise statistical test between algorithms.”
Transformers in Medical Image Analysis: A Review (2022, background): “[115] Brain; Pelvic IXI; BRATS [116], [117], [118]; pelvic MRI-CT datase [119] PSNR, SSIM Multi-model Image synthesis”
TransBTS: Multimodal Brain Tumor Segmentation Using Transformer (2021, methodology): “The second 3D MRI dataset is provided by the Brain Tumor Segmentation Challenge (BraTS) 2020 [11,2,3].”
U-Net and Its Variants for Medical Image Segmentation: A Review of Theory and Applications (2020, neutral): “Example from Brain Tumor Segmentation Challenge (BRATS) 2017 [276]–[278].”
Privacy-preserving Federated Brain Tumour Segmentation (2019, methodology): “For details of the imaging and annotation protocols, we refer the readers to Bakas et al. [2].”
Unified Focal loss: Generalising Dice and cross entropy-based losses to handle class imbalanced medical image segmentation (2021, background): “BraTS20 dataset is currently the largest, publicly available and fully-annotated dataset for medical image segmentation (Nazir et al., 2021), and comprises of 494 multimodal scans of patients with either low-grade glioma or high-grade glioblastoma (Menze et al., 2014; Bakas et al., 2017, 2018).”
Medical Image Segmentation Review: The Success of U-Net (2022, methodology): “This hybrid network captures the local and global information from 3D data and demonstrates the improved performance within two Brain Tumor Segmentation (BraTS) 2019-2020 [91], [135], [136] datasets over the previous CNN U-Net structures.”
AbdomenCT-1K: Is Abdominal Organ Segmentation a Solved Problem? (2020, background): “Since 2015, deep convolutional neural network (CNN)-based methods [36], which neither rely on hand crafted features nor rely on anatomical correspondences, have been successfully introduced into abdominal organ segmentation and reach SOTA performances [37], [38].”
INet: Convolutional Networks for Biomedical Image Segmentation (2021, methodology): “The BraTS2013 dataset [30] provides four MRI modalities, including T1-weighted (T1w), T1w contrast-enhanced (T1C), T2-weighted (T2w), and ﬂuid attenuation inversion recovery (FLAIR).”
The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification (2021, methodology): “The speciﬁc methods fused were the DeepMedic [28], DeepScan [27], and nnU-Net [26], all trained on the BraTS 2020 dataset [11–13].”
A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises (2020, background): “The Brain Tumor Segmentation Challenge (BraTS) [130] has provided the community with an accessible data set and a way to benchmark various approaches against one another.”
Optimization for Medical Image Segmentation: Theory and Practice When Evaluating With Dice Score or Jaccard Index (2020, background): “Implementation details are in [27], with ranks substituted with scores (in line with the hypothesis).”
The state of the art in kidney and kidney tumor segmentation in contrast-enhanced CT imaging: Results of the KiTS19 challenge (2019, background): “Further, because this problem is relatively esoteric compared with its 2D counterpart, the prospect for leveraging transfer learning (Bengio, 2012) from huge, well-established computer vision benchmarks like ImageNet (Deng et al., 2009) or MSCOCO (Lin et al., 2014) is severely limited.”