H-SynEx: Using synthetic images and ultra-high resolution ex vivo MRI for hypothalamus subregion segmentation

arXiv:2401.17104v1 [eess.IV] 30 Jan 2024

In this article, we train a model using synthetic images derived from label maps built from ultra-high resolution ex vivo MRI. We hypothesize that employing synthetic images will help address various MRI contrasts while constructing the label maps from ex vivo images will provide more details of the hypothalamic anatomy, enhancing the automated segmentation quality. We aim to develop an automated method for hypothalamic subregion segmentation, which is robust against variations in MRI contrast and resolution of the input images – including retrospective clinical data, which often present large slice spacing.

Full paper here: https://doi.org/10.48550/arXiv.2401.17104