Abstract
Background and Aims: New clinically viable approaches are needed to realise the potential of non-invasive imaging to monitor progression of the severity of cerebral small vessel disease (SVD) in patients 1. Field-cycling imaging (FCI) is an emerging whole-body MRI technology that provides unique access to underlying tissue features by varying the magnetic field during acquisition, at strengths 10,000 times lower than conventional fixed-field MRI 2. The aim of this preliminary work was to investigate the feasibility of FCI combined with automated segmentation to assess SVD severity.
Methods: After ethical approvals, 9 data sets were included from the first 6 patients recruited, who attended an initial 3T MRI (Philips 3T dStream) and FCI scan (N=6) and repeated scans after 30 days (N=3). Tissue label masks were created for brain tissue, ventricle, and small vessel disease, using a novel in-house automated approach to process multi-field FCI data. An existing segmentation approach was used to process 3T MRI data 3.SVD fraction was calculated as the ratio between SVD and brain matter volume. Agreement between FCI and 3T co-registered segmentation results was assessed by Dice coefficient and Pearson correlation.
Results: Dice coefficients (mean, range) were obtained for brain matter inclusive of SVD (0.89, 0.86–0.93), ventricle (0.91, 0.81–0.95), and SVD only (0.52, 0.27–0.73), (Fig.1). A significant Pearson correlation was obtained between SVD fractions (r=0.861, p=0.003).
Conclusions: These promising preliminary results are a crucial first step towards demonstrating the clinical feasibility of FCI to assess SVD severity.
Methods: After ethical approvals, 9 data sets were included from the first 6 patients recruited, who attended an initial 3T MRI (Philips 3T dStream) and FCI scan (N=6) and repeated scans after 30 days (N=3). Tissue label masks were created for brain tissue, ventricle, and small vessel disease, using a novel in-house automated approach to process multi-field FCI data. An existing segmentation approach was used to process 3T MRI data 3.SVD fraction was calculated as the ratio between SVD and brain matter volume. Agreement between FCI and 3T co-registered segmentation results was assessed by Dice coefficient and Pearson correlation.
Results: Dice coefficients (mean, range) were obtained for brain matter inclusive of SVD (0.89, 0.86–0.93), ventricle (0.91, 0.81–0.95), and SVD only (0.52, 0.27–0.73), (Fig.1). A significant Pearson correlation was obtained between SVD fractions (r=0.861, p=0.003).
Conclusions: These promising preliminary results are a crucial first step towards demonstrating the clinical feasibility of FCI to assess SVD severity.
Original language | English |
---|---|
Article number | EP201/#2592 |
Pages (from-to) | 209 |
Journal | International Journal of Stroke |
Volume | 18 |
Issue number | 3_suppl |
Early online date | 10 Oct 2023 |
DOIs | |
Publication status | Published - 10 Oct 2023 |
Event | World Stroke Congress 2023 - Toronto, Canada Duration: 10 Oct 2023 → 12 Oct 2023 https://2023.worldstrokecongress.org/ |
Fingerprint
Dive into the research topics of 'Assessing severity of cerebral small vessel disease using field-cycling MRI and automated segmentation'. Together they form a unique fingerprint.Equipment
-
Aberdeen Biomedical Imaging Centre
Waiter, G. (Manager) & Morris, T. (Facilities Co-ordinator)
Aberdeen Biomedical Imaging CentreResearch Facilities: Facility