Deep Learning for Classifying and Cognitive Profiling of Subcortical Vascular Cognitive Impairment

It is challenging to predict which patients who meet criteria for subcortical ischemic vascular disease (SIVD) will ultimately progress to subcortical vascular cognitive impairment (SVCI). We collected clinical information, neuropsychological assessments, T1 imaging, diffusion tensor imaging, and resting-state functional magnetic resonance imaging from 83 patients with SVCI and 53 age-matched patients with SIVD without cognitive impairment. We built an unsupervised machine learning model to isolate patients with SVCI. The model was validated using multimodal data from an external cohort comprising 45 patients with SVCI and 32 patients with SIVD without cognitive impairment. The accuracy, sensitivity, and specificity of the unsupervised machine learning model were 86.03%, 79.52%, and 96.23% and 80.52%, 71.11%, and 93.75% for internal and external cohort, respectively. We developed an accurate and accessible clinical tool which requires only data from routine imaging to predict patients at risk of progressing from SIVD to SVCI. Our unsupervised machine learning model provides an accurate and accessible clinical tool to predict patients at risk of progressing from subcortical ischemic vascular disease (SIVD) to subcortical vascular cognitive impairment (SVCI) and requires only data from imaging routinely used during the diagnosis of suspected SVCI. The model yields good accuracy, sensitivity, and specificity and is portable to other cohorts and to clinical practice to distinguish patients with SIVD at risk for progressing to SVCI. The model combines assessment of diffusion tensor imaging and functional magnetic resonance imaging measures in patients with SVCI to analyze whether the "disconnection hypothesis" contributes to functional and structural changes and to the clinical presentation of SVCI.

BackgroundIt is challenging to distinguish which subcortical ischemic vascular disease (SIVD) patients will present with cognitive impairment. A blood‐based biomarker to distinguish SIVD patients with cognitive impairment would be superior to neuropsychological measures and neuroimaging measures in terms of cost, time, and feasibility for repeated measures. Metabolomics profiling studies could help identify blood‐based biomarkers for SIVD patients with cognitive impairment.Method100 SIVD patients without cognitive impairment,70 SIVD patients with cognitive impairment, 30 DLB patients and 30 AD patients were recruited from three Chinese research hospitals. 34 SIVD patients without cognitive impairment and 34 SIVD patients were follow‐up for 2 years. Patients’ serum samples were assessed using non‐targeted metabolomics and lipidomics analyses.ResultSix metabolites including TG(52:2)‐FA18:1, DHCer(d18:0_24:0), ChE(22:4), Cer(d18:1_18:0), SM(d18:1_26:0) and Hex2Cer(d18:1_24:0) were identified as informative biomarkers for cognitive impairment in SIVD patients. The performance was stable after 2 years follow up and in differential diagnosis among AD and DLB patients.ConclusionOur study has developed an accurate and accessible means based on blood samples to identify SIVD patients at high risk of cognitive impairment. Dysregulation of sphingolipid metabolism was involved in the neurobiological pathways that contributed to cognitive impairment in SIVD patients.

Objective: Changes in the normal asymmetry of the human brain often mean pathology. Current studies on the correlation between asymmetry and cognitive impairment have focused on Alzheimer's disease (AD) and AD-related mild cognitive impairment (MCI). The purpose of this study was to investigate changes in gray matter asymmetry and their relationship with cognitive impairment in patients with subcortical ischemic vascular disease (SIVD) by using voxel-based morphological measurements.Methods: Fifty-nine SIVD patients with (subcortical vascular cognitive impairment, SVCI, N = 30) and without (pre-SVCI, N = 29) cognitive impairment and 30 normal controls (NC, N = 30) underwent high-resolution structural MRI and neuropsychological examinations. The differences in gray matter asymmetry among the three groups were estimated by using one-way ANOVA. Moreover, partial correlation analysis was performed to explore the relationships between the asymmetry index (AI) values and cognitive assessments controlled for age, sex, and education.Results: The gray matter asymmetries in the fusiform and parahippocampal gyruses of the SVCI group were significantly different from those of the NC group and the pre-SVCI group, while no differences were found between the NC group and the pre-SVCI group in the same areas. More specifically, in the fusiform and parahippocampal gyruses, the SVCI group displayed a dramatic rightward asymmetry, whereas the NC group and pre-SVCI group exhibited a marked leftward asymmetry. The results of the correlation analysis showed that the “mean AI” in significant cluster was strongly correlated with the changes in cognitive outcomes.Conclusion: This study demonstrated different lateralization in the fusiform and parahippocampal gyruses of SIVD patients with cognitive impairment compared to healthy subjects and SIVD patients without cognitive decline. Our findings may contribute to better understanding the possible mechanism of cognitive impairment in patients with SIVD, and they suggest the possibility of using gray matter asymmetry as a biomarker for disease progression.

BackgroundIt is challenging to predict which patients who meet criteria for subcor‐ tical ischemic vascular disease (SIVD) will ultimately progress to subcortical vascular cognitive impairment (SVCI).MethodWe collected clinical information, neuropsychological assessments, T1 imag‐ ing, diffusion tensor imaging, and resting‐state functional magnetic resonance imaging from 83 patients with SVCI and 53 age‐matched patients with SIVD without cogni‐ tive impairment. We built an unsupervised machine learning model to isolate patients with SVCI. The model was validated using multimodal data from an external cohort comprising 45 patients with SVCI and 32 patients with SIVD without cognitive impairment.ResultThe accuracy, sensitivity, and specificity of the unsupervised machine learning model were 86.03%, 79.52%, and 96.23% and 80.52%, 71.11%, and 93.75% for internal and external cohort, respectively.ConclusionWe developed an accurate and accessible clinical tool which requires only data from routine imaging to predict patients at risk of progressing from SIVD to SVCI.

To explore the static and dynamic characteristics of intrinsic brain activity (IBA) in subcortical ischemic vascular disease (SIVD) patients with or without cognitive impairment. In total, 90 participants were recruited, including 32 SIVD patients with cognitive impairment (SIVD-CI, N = 32), 26 SIVD patients with no cognitive impairment (SIVD-NCI, N = 26), and 32 healthy controls (HC, N = 32) matched for age, gender, and education. All subjects underwent resting-state functional magnetic resonance imaging (rs-fMRI) scanning and neuropsychological tests. Amplitude of low-frequency fluctuation (ALFF) was calculated to reflect static alterations of regional IBA. Sliding window analysis was conducted in order to explore the dynamic characteristics. Both SIVD-CI and SIVD-NCI group showed significantly decreased ALFF in left angular gyrus (ANG), whereas SIVD-CI group showed increased ALFF in right superior frontal gyrus (SFG), compared with HCs. Furthermore, SIVD-CI group showed significantly decreased ALFF dynamics (dALFF) in right precuneus (PreCu) and left dorsal anterior cingulate cortex (dACC), compared with HC and SIVD-NCI groups (Gaussian random field-corrected, voxel-level P < 0.001, cluster-level P < 0.05). No dynamic changes were detected between SIVD-NCI group and HC group. The mean ALFF value in left ANG of SIVD-CI group was correlated with the score of delayed memory scale. ANG may be a vulnerable brain region in SIVD patients. Temporal dynamic analysis could serve as a sensitive and promising method to investigate IBA alterations in SIVD patients.

The presence and pattern of iron accumulation in subcortical vascular mild cognitive impairment (svMCI) and their effects on cognition have rarely been investigated. We aimed to examine brain iron deposition in svMCI subjects using quantitative susceptibility mapping (QSM). Moreover, we aimed to investigate the correlation between brain iron deposition and the severity of cognitive impairment as indicated by z-scores. We recruited 20 subcortical ischemic vascular disease (SIVD) patients who fulfilled the criteria for svMCI. The control group comprised 19 SIVD patients without cognitive impairment. The SIVD and control groups were matched based on age, gender, and years of education. Both groups underwent QSM using a 3.0T MRI system. Susceptibility maps were reconstructed from in vivo data, which were acquired with a three-dimensional spoiled gradient recalled sequence. Then, regions of interest were drawn manually on the map of each subject. The inter-group differences of susceptibility values were explored in deep gray matter nuclei, including the bilateral pulvinar nucleus of the thalamus, head of caudate nucleus, globus pallidus, putamen, hippocampus, substantia nigra, and red nucleus. The correlations between regional iron deposition and composite z-score, memory z-score, language z-score, attention-executive z-score and visuospatial z-score were assessed using partial correlation analysis, with patient age and gender as covariates. Compared with the control, the svMCI group had elevated susceptibility values within the bilateral hippocampus and right putamen. Furthermore, the susceptibility value in the right hippocampus was negatively correlated with memory z-score and positively correlated with language z-score. The susceptibility value in the right putamen was negatively correlated with attention-executive z-score in the svMCI group. However, composite z-score were unrelated to susceptibility values. Our results suggest that brain iron deposition has clinical relevance as a biomarker for cognition. In addition, our results highlight the importance of iron deposition in understanding svMCI-associated cognitive deficits in addition to conventional MRI markers.

Subcortical ischemic vascular disease (SIVD) is a key contributor to vascular cognitive impairment, yet early-stage SIVD is underexplored despite its link to cognitive decline. This study investigates structural and functional changes in SIVD patients with and without cognitive impairment. Twenty-five individuals of subcortical ischemic vascular disease with normal cognition, 25 individuals with subcortical ischemic vascular cognitive impairment (SVCI), and 23 normal controls were enrolled. Voxel-based morphometry was used to assess differences in gray matter volume (GMV) across 3 groups by one-way analysis of variance. Pearson correlation coefficients were computed between voxel-based morphometry-identified seeds and all brain voxels derived from the resting-state functional magnetic resonance imaging analyses to evaluate the alterations of functional connectivity (FC). Spearman correlation analyses were performed to examine the relationships between altered GMV, abnormal FC, and cognitive assessments. Compared with normal controls, the SVCI group showed significant reductions in GMV in prefrontal regions (orbitofrontal cortex, gyrus rectus, inferior frontal gyrus), temporal lobe (olfactory cortex, superior temporal gyrus), insula, and subcortical areas (thalamus, caudate). The subcortical ischemic vascular disease with normal cognition group exhibited GMV loss in the right orbitofrontal cortex. FC analyses revealed hypoconnectivity in cortical-subcortical circuits as well as ipsilateral and contralateral cognitive-related regions. Several GMV and FC alterations in the SVCI group were correlated with neuropsychological assessments, including the Stroop Color-Word Test II and Trail Making Test B. Early-stage SIVD patients exhibit structural and functional abnormalities linked to cognitive deficits. The right orbitofrontal cortex may serve as a potential imaging biomarker for early detection of SIVD.

Subcortical ischemic vascular disease (SIVD) is considered to be the most important and common cause of vascular cognitive impairment (VCI). If patients with subcortical ischemic vascular cognitive impairment (SIVCI) and subcortical vascular cognitive impairment (sVCI) can be found early, it is possible that vascular dementia (VaD) can be identified before occurrence and even reverse the process. Recent studies have shown that resting-state functional magnetic resonance imaging (rsfMRI) may provide the objective basis for the diagnosis of SIVCI. This article reviews the application of rsfMRI in the diagnosis of SIVCI. Key words: Cognition Disorders; Dementia, Vascular; Brain Ischemia; Magnetic Resonance Imaging; Cerebral Cortex; Nerve Net

Vascular cognitive impairment (VCI) describes all forms of cognitive impairment caused by any type of cerebrovascular disease. Early identification of VCI is quite difficult due to the lack of both...

Both Alzheimer's disease (AD) and cerebral small vessel disease (CSVD) manifest in cognitive impairment and gait disorders. The precise similarities and differences in gait characteristics and underlying neuroimaging mechanisms remain unclear. A total of 399 participants were enrolled: 132 with probable AD, including 98 with mild cognitive impairment due to AD (AD-MCI) and 34 with AD dementia, and 185 with CSVD and 82 healthy controls. CSVD patients with cognitive impairment, including subcortical vascular mild cognitive impairment (svMCI) and subcortical vascular dementia, were grouped as subcortical vascular cognitive impairment (SVCI). Voxel-based morphology analysis assessed grey matter volume (GMV), while cerebral blood flow (CBF) was derived from 3D-arterial spin labelling data. Gait metrics included the timed up and go (TUG) test, dual-task TUG (DTUG) test, Berg balance scale (BBS), dual-task cost (DTC), step length, gait speed, cadence and coefficient of variation of gait. The relationships among structural and perfusion variations, gait metrics and cognitive function were examined. SVCI patients exhibited greater gait impairments and variability than those with AD, while AD patients experienced higher DTC (p < 0.05). These differences were most evident in the MCI stage. In AD, gait speed correlated with GMV in the left middle occipital gyrus (F = 6.149), middle temporal gyrus (F = 4.595), right precuneus (F = 5.174) and other regions (all p < 0.025). In SVCI, gait speed was linked to thalamic GMV (F = 6.004, p < 0.025). Altered CBF in the parietal lobe and precuneus was associated with DTUG (F = 5.672), gait speed (F = 4.347) and BBS (F = 4.153) in AD, while cerebellar CBF related to TUG (F = 6.042), DTUG (F = 4.857) and BBS (F = 7.097) in SVCI (all p < 0.025). In AD-MCI, memory mediated the effect of hippocampal volume on DTC (indirect effect: -2.432, 95% CI [-5.503, -0.438]), while executive function (indirect effect: -2.920, 95% CI [-7.227, -0.695]) and processing speed (indirect effect: -2.286, 95% CI [-5.174, -0.484]) mediated the effect on DTUG. In svMCI, executive function mediated the effect of thalamic volume on step length (indirect effect: 2.309, 95% CI [0.486, 4.685]) and gait speed (indirect effect: 2.029, 95% CI [0.142, 4.588]), while processing speed mediated the effect on step length (indirect effect: 1.777, 95% CI [0.311, 4.021]). Different gait disorder characteristics and mechanisms were observed in AD and CSVD patients. In AD, gait is associated with volume/perfusion in posterior brain regions, whereas in SVCI, it relates to thalamic volume and cerebellar perfusion. Cognitive impairment mediates the effect of hippocampal and thalamic volumes on gait in AD-MCI and svMCI, respectively.

Abnormal reductions in cortical cerebral blood flow (CBF) have been identified in subcortical vascular cognitive impairment (SVCI). However, little is known about the pattern of CBF reduction in relation with the degree of cognitive impairment. CBF measured with three-dimensional (3D) Arterial Spin Labeling (ASL) perfusion magnetic resonance imaging (MRI) helps detect functional changes in subjects with SVCI. We aimed to compare CBF maps in subcortical ischemic vascular disease (SIVD) subjects with and without cognitive impairment and to detect the relationship of the regions of CBF reduction in the brain with the degree of cognitive impairment according to the z-score. A total of 53 subjects with SVCI and 23 matched SIVD subjects without cognitive impairment (controls), underwent a whole-brain 3D ASL MRI in the resting state. Regional CBF (rCBF) was compared voxel wise by using an analysis of variance design in a statistical parametric mapping program, with patient age and sex as covariates. Correlations were calculated between the rCBF value in the whole brain and the z-score in the 53 subjects with SVCI. Compared with the control subjects, SVCI group demonstrated diffuse decreased CBF in the brain. Significant positive correlations were determined in the rCBF values in the left hippocampus, left superior temporal pole gyrus, right superior frontal orbital lobe, right medial frontal orbital lobe, right middle temporal lobe, left thalamus and right insula with the z-scores in SVCI group. The noninvasively quantified resting CBF demonstrated altered CBF distributions in the SVCI brain. The deficit brain perfusions in the temporal and frontal lobe, hippocampus, thalamus and insula was related to the degree of cognitive impairment. Its relationship to cognition indicates the clinical relevance of this functional marker. Thus, our results provide further evidence for the mechanisms underlying the cognitive deficit in patients with SVCI.

Subcortical ischemic vascular disease (SIVD) is one of the important causes of cognitive dysfunction, altered amyloid-beta (Aβ) and cerebral perfusion may be involved in the pathophysiological mechanism of SIVD and are closely related to cognitive function. We aimed to investigate altered serum Aβ and cerebral perfusion in patients with SIVD and their correlation with cognitive function. Seventy-four healthy controls (HCs) and 74 SIVD patients, including 38 SIVD patients with no cognitive impairment (SIVD-NCI) and 36 SIVD patients with mild cognitive impairment (SIVD-MCI) underwent the measurement of serum Aβ40 and Aβ42 levels, pseudo-continuous arterial spin labeling MRI scanning, and cognitive evaluation. Compared to the healthy controls (HCs), the level of serum Aβ40 and Aβ40/42 ratio increased and Aβ42 decreased in SIVD patients. The serum Aβ40 level and Aβ40/42 ratio in patients with SIVD-MCI were significantly higher than those in the HCs and SIVD-NCI, and the level of Aβ42 in the SIVD-MCI was lower than the HCs. In addition, the serum Aβ40/42 ratio provided high diagnostic accuracy for SIVD and SIVD-MCI, it was further identified as an independent risk factor for cognitive impairment. Patients with SIVD-NCI and SIVD-MCI exhibited both increased and decreased cerebral blood flow (CBF) in regional. The Aβ40/42 ratio was associated with global CBF, while altered global and regional CBF was associated with cognitive deficits. In addition, white matter hyperintensities volume (WMHV) correlated with Aβ40/42 ratio, CBF, and cognition. The relationship between Aβ40/42 ratio and cognition was partially mediated by altered CBF. Based on these results, we conclude that the serum Aβ40/42 ratio may be a potential biomarker that can complement current methods for the prediction and diagnosis of cognitive impairment in SIVD patients. In addition, serum Aβ may play a role in cognitive function by regulating CBF, which provides new insights into the intervention, treatment, and prevention of cognitive impairment in SIVD.

Patients with subcortical ischemic vascular disease (SIVD) exhibit a high risk of cognitive impairment that might be caused by neurologic deficits and vascular injuries. However, the mechanism remains unknown. In current study, 24 normal controls (NC) and 54 SIVD patients, including 26 SIVD patients with no cognitive impairment (SIVD-NCI) and 28 SIVD patients with mild cognitive impairment (SIVD-MCI) underwent the resting-state functional MRI (rs-fMRI) and neuropsychological assessments. We combined regional homogeneity (ReHo) and cerebral blood flow (CBF) by using the global ReHo-CBF correlations coefficient and the ReHo/CBF ratio to detect the inner link between neuronal activity and vascular responses. Correlations between the ReHo/CBF ratio and neuropsychological assessments were explored in patients with SIVD. As a result, we identified significantly decreased global ReHo-CBF coupling in the SIVD-NCI group and SIVD- MCI group with respect to the NC. The SIVD-MCI group showed more serious decoupling of the global ReHo-CBF correlation. We also found a significantly abnormal ReHo/CBF ratio predominantly located in cognitive-related brain regions, including the left insula, right middle temporal gyrus, right precuneus, left precentral gyrus, and left inferior parietal lobule but not the supramarginal and angular gyri. The SIVD-MCI group showed more severe disorders of neurovascular coupling than the other two groups. Moreover, the ReHo/CBF ratio in the left precentral gyrus of the SIVD-NCI group exhibited a positive correlation with the MMSE scores. These findings suggested that patients with SIVD show abnormal neurovascular coupling at the early stage of the disease and during disease development. It might be associated with disease severity and cognitive impairment. Neurovascular decoupling in brain may be a possible neuropathological mechanism of SIVD.

Few studies exist on ERPs and patients with subcortical vascular cognitive impairment (SVCI). This latter is a quite homogeneous subtype of vascular dementia whose cognitive profile is quite different from that of Alzheimer disease (AD). The present study aims at comparing the ERPs profile both in patients with SVCI and in patients with AD. ERPs and psychometric tests were collected from 39 healthy elderly controls, 51 patients with SVCI and 43 patients with AD. Subjects mentally count high pitched target tones that were randomly intermixed with low pitched frequent tones. We measured ERPs latencies (N1, P2, N2 and P3), and interpeak latencies (N1-P3, N1-P2, N1-N2). Grand averaged potentials in SVCI showed a significant increase of P3 latency. AD patients showed a prolongation of N1, P2, N2, P3 latencies. As far as interpeak latencies are concerned, SVCI patients showed a significant prolongation of N1-P3, AD patients had a significant increase of N1-N2, and N1-P3 intervals. When all patients were considered as a single group, correlation of neuropsychological tests scores showed a significant negative relationship between P300 latency and, respectively, Mini Mental Status Examination, auditive and visual span forward. In both groups, ERPs latency sensitivity, was low, whilst specificity values were quite high. Our finding suggest that these two dementing diseases have different electrophysiologic features that may be related to their specific underlying pathogenetic mechanism; in particular, we hypothesise that, differently from AD, P300 latency prolongation characterizes the early stage of SVCI. So, this ERPs approach could be helpful to detect early alterations of the attentional/working-memory functions in patients with subcortical ischaemic vascular disease.

Although the association between apolipoprotein E (APOE) genotype and disease progression is well characterized in patients with Alzheimer’s disease, such a relationship is unknown in patients with subcortical vascular cognitive impairment. We evaluated whether APOE genotype is associated with disease progression in subcortical vascular mild cognitive impairment (svMCI) patients. We prospectively recruited 72 svMCI patients (19 APOE4 carriers, 42 APOE3 homozygotes, and 11 APOE2 carriers). Patients were annually followed-up with brain MRI and neuropsychological tests for three years and underwent a second Pittsburgh compound B (PiB)-PET at a mean interval of 32.3 months. Amyloid-ß burden was quantified by PiB standardized uptake value ratio (SUVR), and the amount of small vessel disease was quantified by number of lacune and small vessel disease score on MRI. We also measured cortical thickness. During the three years of follow-up, compared to the APOE3 homozygotes, there was less increase in PiB SUVR among APOE2 carriers (p = 0.023), while the APOE genotype did not show significant effects on small vessel disease progression. APOE2 carriers also showed less cortical thinning (p = 0.023) and a slower rate of cognitive decline (p = 0.009) compared to those with APOE3 homozygotes. Our findings suggest that, in svMCI patients, APOE2 has protective effects against amyloid-ß accumulation, cortical thinning, and cognitive decline.