Validation and comparison of two automated methods to quantify brain white matter hyperintensities of presumed vascular origin

Background: White matter hyperintensities (WMH) are a common imaging finding, indicative of cerebral small vessel disease. Lesion segmentation algorithms have been developed to overcome issues arising from visual rating scales. In this study, two automated methods were evaluated against visual and manual segmentation. We aimed to determine the most robust algorithm provided by the open-source Lesion Segmentation Toolbox (LST). Methods: We compared WMH data from visual rating (Scheltens’ scale) with those derived from algorithms provided within LST. We then compared spatial and volumetric WMH data derived from manually-delineated lesion maps with WMH data and lesion maps provided by LST algorithms. Results: We identified optimal initial thresholds for algorithms provided by LST when compared with visual rating (LGA: initial _K and lesion probability thresholds = 0.5; LPA lesion probability threshold = 0.65). LGA was found to perform best in comparisons with manual segmentation. Discussion: When compared to Scheltens’ score and to manual segmentation, LGA appeared to be the most suitable algorithm. We found LGA to be a userfriendly, effective WMH segmentation method for a research environment.