Link to resource in Edinburgh Datashare
Authors
Valdés Hernández MdC, Ballerini L, Glatz A, Aribisala BS, Bastin ME, Dickie DA, Duarte Coello R, Munoz Maniega S, Wardlaw JM
Description
This dataset contains the matlab code to parcellate the brain, and segment brain tissues, white matter hyperintensities, and perivascular spaces from magnetic resonance imaging (MRI) structural sequences acquired using research protocols. It assumes isotropic or nearly-isotropic T2-weighted of voxel size approximately 1 mm3 as the image space where the fluid attenuated inversion recovery (FLAIR) and T1-weighted images are co-registered beforehand. It also assumes the image data is organised in BIDS (https://bids.neuroimaging.io/) format, and that the contents of the intracranial volume have been previously extracted (although this is not a requirement). The code is written in MATLAB, and invokes third-party software for parcellating the brain and generating tissue priors. The results from the brain parcellation, are used further to clean the initial results of tissue and lesion segmentation procedures, and generate the regions of interest for assessing the enlarged perivascular spaces (PVS). Rather than an end-to-end encapsulated software, the code is presented as a step-by-step pipeline starting by the brain parcellation (step 03), with comments justifying and explaining every block of operations, as well as alternatives, to facilitate its use. It was developed for the Mild Stroke Study 3 (PMID: 33817338), a clinical study aimed at studying contributing factors and clinical implications of small vessel disease dynamics in adults that suffered a mild-to-moderate stroke. As such, the pipeline deals with a large range and variety of vascular abnormalities present in a large number of adult brain MRI scans, and can be widely applicable. However, it is intended as a guidance that will need to be further tuned for a specific cohort, to facilitate reproducibility of our brain research results.
Keywords
Citation
Social media tags and titles
Featured resource: Step-by-step pipeline for segmenting enlarged perivascular spaces from 3D T2-weighted MRI.