Simultaneous synthesis of FLAIR and segmentation of white matter hypointensities from T1 MRIs

• DOI: 10.48550/arxiv.1808.06519
• Publication Date: 2018-01-01
• Authors: Mauricio Orbes‐Ar...; M. Jorge Cardoso; Lauge Sørensen; Marc Modat; Sébastien Ourselin; Mads Nielsen; Akshay Pai
• Keywords: Fluid-attenuated inversion recovery; Imputation (statistics); Modality (human–computer interaction)

Segmenting vascular pathologies such as white matter lesions in Brain magnetic resonance images (MRIs) require acquisition of multiple sequences T1-weighted (T1-w) --on which appear hypointense-- and fluid attenuated inversion recovery (FLAIR) sequence --where hyperintense--. However, most the existing retrospective datasets do not consist FLAIR sequences. Existing missing modality imputation methods separate process imputation, segmentation. In this paper, we propose a method to link both segmentation using convolutional neural networks. We show that by jointly optimizing network network, only produces more realistic synthetic from T1-w images, but also improves WMH only.