Background and Purpose— Cerebral small vessel disease is characterized by a wide range of focal and global brain changes. We used a magnetic resonance imaging segmentation tool to quantify multiple types of small vessel disease–related brain changes and examined their individual and combined predictive value on cognitive and functional abilities. Methods— Magnetic resonance imaging scans of 560 older individuals from LADIS (Leukoaraiosis and Disability Study) were analyzed using automated atlas- and convolutional neural network–based segmentation methods yielding volumetric measures of white matter hyperintensities, lacunes, enlarged perivascular spaces, chronic cortical infarcts, and global and regional brain atrophy. The subjects were followed up with annual neuropsychological examinations for 3 years and evaluation of instrumental activities of daily living for 7 years. Results— The strongest predictors of cognitive performance and functional outcome over time were the total volumes of white matter hyperintensities, gray matter, and hippocampi ( P <0.001 for global cognitive function, processing speed, executive functions, and memory and P <0.001 for poor functional outcome). Volumes of lacunes, enlarged perivascular spaces, and cortical infarcts were significantly associated with part of the outcome measures, but their contribution was weaker. In a multivariable linear mixed model, volumes of white matter hyperintensities, lacunes, gray matter, and hippocampi remained as independent predictors of cognitive impairment. A combined measure of these markers based on Z scores strongly predicted cognitive and functional outcomes ( P <0.001) even above the contribution of the individual brain changes. Conclusions— Global burden of small vessel disease–related brain changes as quantified by an image segmentation tool is a powerful predictor of long-term cognitive decline and functional disability. A combined measure of white matter hyperintensities, lacunar, gray matter, and hippocampal volumes could be used as an imaging marker associated with vascular cognitive impairment.

No abstract

Purpose Severity of white matter lesion (WML) is typically evaluated on magnetic resonance images (MRI), yet the more accessible, faster, and less expensive method is computed tomography (CT). Our objective was to study whether WML can be automatically segmented from CT images using a convolutional neural network (CNN). The second aim was to compare CT segmentation with MRI segmentation. Methods The brain images from the Helsinki University Hospital clinical image archive were systematically screened to make CT-MRI image pairs. Selection criteria for the study were that both CT and MRI images were acquired within 6 weeks. In total, 147 image pairs were included. We used CNN to segment WML from CT images. Training and testing of CNN for CT was performed using 10-fold cross-validation, and the segmentation results were compared with the corresponding segmentations from MRI. Results A Pearson correlation of 0.94 was obtained between the automatic WML volumes of MRI and CT segmentations. The average Dice similarity index validating the overlap between CT and FLAIR segmentations was 0.68 for the Fazekas 3 group. Conclusion CNN-based segmentation of CT images may provide a means to evaluate the severity of WML and establish a link between CT WML patterns and the current standard MRI-based visual rating scale.

Background and purpose: Cerebral small vessel disease (SVD) is characterized by a wide range of focal and global brain changes. We used automated MRI segmentation to quantify multiple types of SVD brain changes and examined their individual and combined predictive value on cognitive and functional abilities.Methods: MRI scans of 560 subjects of the Leukoaraiosis and Disability Study (LADIS) were analyzed using automated atlas-and convolutional neural network-based segmentation methods yielding volumetric measures of white matter hyperintensities (WMH), lacunes, cortical infarcts, enlarged perivascular spaces and regional brain atrophy. The subjects were followed up with annual neuropsychological examinations for 3 years and evaluation of instrumental activities of daily living for 7 years. Results:The strongest predictors of cognitive performance and functional outcome over time were total volumes of WMH, grey matter (GM) and hippocampi (p<0.001 for global cognitive function, processing speed, executive functions and memory; and p<0.001 for poor functional outcome).Volumes of lacunes, cortical infarcts and EVPS were significantly associated with part of the outcome measures, but their contribution was weaker. In a multivariable linear mixed model, volumes of WMH, lacunes, GM and hippocampi remained as independent predictors of cognitive impairment. A combined measure of these markers based on z-scores strongly predicted cognitive and functional outcomes (p<0.001) even above the contribution of the individual brain changes. Conclusions:Global burden of SVD-related brain changes as quantified by automated image segmentation is a powerful predictor of long-term cognitive decline and functional disability. A combined measure of WMH, lacunar, GM and hippocampal volumes could be used as an imaging identification model of vascular cognitive impairment.