Neurology Colloquium
Automated Methods for the Quantification and Detection of Enlarged Perivascular Spaces in Brain MRI
- 17. Juni 2019, 13:00 Uhr
- Campus FME, Universitätsklinik für Neurologie (Haus 60b, Flur 3, Raum 3784), Leipziger Str. 44, 39120 Magdeburg
- Florian Dubost (Departments of Radiology and Medical Informatics, Erasmus MC, Rotterdam);
- Host: Stefanie Schreiber & Steffen Oeltze-Jafra (Department of Neurology/DZNE/CBBS)
-
Perivascular spaces (PVS) are very common in aging and have emerged as a new imaging marker for cerebral small vessel disease and beta-amyloid drainage. Visual and manual scoring of PVS is a tedious and observer-dependent task. Florian Dubost proposes and evaluated an automated method to quantify and detect PVS in several brain regions. His approach is a deep learning algorithm based on convolutional neural network regression, and is contingent on successful brain structure segmentation.He concludes that this method may replace visual scoring and facilitate large epidemiological and clinical studies of PVS.