Table 1.

MS lesion segmentation methods evaluated at the MICCAI 2016 challenge.

Team	Authors	Segmentation approach	Platform	Sequences used
1	J. Beaumont O. Commowick	Graph cut segmentation initialized by a robust EM23,24	CPU	T1-w, T2-w, FLAIR (preprocessed)
2	J. Beaumont O. Commowick	Multi-modal abnormalities detection from normalized images on an atlas25,26	CPU	T2-w, FLAIR (preprocessed)
3	S. Doyle F. Forbes	HMRF segmentation framework with a weighted data model27,28	CPU	T1-w, FLAIR (raw)
4	J. Knight A. Khademi	Segmentation by edge-based model of partial volume/pure tissue gray levels29,30	CPU	FLAIR (raw)
5	A. Mahbod C. Wang	Supervised artificial neural network with intensity and spatial based features31,32	CPU	FLAIR (preprocessed)
6	R. McKinley T. Gundersen	Ensemble of three 2D fully Convolutional Neural Networks with skip connections33	GPU	FLAIR (preprocessed)
7	J. Muschelli E. Sweeney	Random Forest (RF) on normalized multi-modal features34	CPU	T1-w, T2-w, PD, FLAIR (raw)
8	E. Roura X. Lladó	Outlier segmentation based on brain tissue labeling and post-processing rules35,36	CPU	T1-w, FLAIR (raw)
9	M. Santos A. Silva-Filho	Multilayer perceptron with cost functions oriented to competition evaluation metrics37,38	CPU	T1-w, T2-w, FLAIR (preprocessed)
10	X. Tomas-Fernandez S.K. Warfield	Lesions and brain tissue segmentation through simultaneous estimation of spatially and population varying intensity distributions39,40	CPU	T1-w, T2-w, FLAIR (raw)
11	H. Urien I. Bloch	Hierarchical segmentation using max-tree, spatial context and anatomical constraints41,42	CPU	T1-w, T1-w Gd, T2-w, PD, FLAIR (raw, preprocessed)
12	S. Valverde M. Cabezas	Cascade of two 7-layer convolutional neural networks of 3D patches43	GPU	T1-w, T2-w, PD, FLAIR (preprocessed)
13	F.J. Vera-Olmos N. Malpica	Grey matter filter as input to a RF classifier corrected with Markov Random Field processing44	CPU	T1-w, T2-w, PD, FLAIR (preprocessed)