Segmentation of white matter hyperintensities using convolutional neural networks with global spatial information in routine clinical brain MRI with none or mild vascular pathology.

•2D CNN outperforms other machine learning algorithms in segmenting WMH on brains with none or mild vascular pathology.•15 supervised and semi-supervised machine learning schemes for WMH segmentation evaluated.•Novel scheme to incorporate global spatial information to a CNN for WMH segmentation.•All machine learning schemes applied are optimised to their best performance prior to their evaluation.