ABSTRACT
INTRODUCTION: Cerebral small vessel disease (SVD) and amyloid beta (Aß) pathology frequently co-exist. The impact of concurrent pathology on the pattern of hippocampal atrophy, a key substrate of memory impacted early and extensively in dementia, remains poorly understood. METHODS: In a unique cohort of mixed Alzheimer's disease and moderate-severe SVD, we examined whether total and regional neuroimaging measures of SVD, white matter hyperintensities (WMH), and Aß, as assessed by 18F-AV45 positron emission tomography, exert additive or synergistic effects on hippocampal volume and shape. RESULTS: Frontal WMH, occipital WMH, and Aß were independently associated with smaller hippocampal volume. Frontal WMH had a spatially distinct impact on hippocampal shape relative to Aß. In contrast, hippocampal shape alterations associated with occipital WMH spatially overlapped with Aß-vulnerable subregions. DISCUSSION: Hippocampal degeneration is differentially sensitive to SVD and Aß pathology. The pattern of hippocampal atrophy could serve as a disease-specific biomarker, and thus guide clinical diagnosis and individualized treatment strategies for mixed dementia.
Subject(s)
Alzheimer Disease , Amyloid beta-Peptides , Cerebral Small Vessel Diseases , Hippocampus , Positron-Emission Tomography , Humans , Hippocampus/pathology , Hippocampus/diagnostic imaging , Cerebral Small Vessel Diseases/pathology , Cerebral Small Vessel Diseases/diagnostic imaging , Male , Aged , Female , Alzheimer Disease/pathology , Alzheimer Disease/diagnostic imaging , Amyloid beta-Peptides/metabolism , White Matter/pathology , White Matter/diagnostic imaging , Atrophy/pathology , Magnetic Resonance Imaging , Aged, 80 and over , Neuroimaging , Cohort StudiesABSTRACT
BACKGROUND: Neuropsychiatric symptoms (NPS) are a core feature of most neurodegenerative and cerebrovascular diseases. White matter hyperintensities and brain atrophy have been implicated in NPS. We aimed to investigate the relative contribution of white matter hyperintensities and cortical thickness to NPS in participants across neurodegenerative and cerebrovascular diseases. METHODS: Five hundred thirteen participants with one of these conditions, i.e. Alzheimer's Disease/Mild Cognitive Impairment, Amyotrophic Lateral Sclerosis, Frontotemporal Dementia, Parkinson's Disease, or Cerebrovascular Disease, were included in the study. NPS were assessed using the Neuropsychiatric Inventory - Questionnaire and grouped into hyperactivity, psychotic, affective, and apathy subsyndromes. White matter hyperintensities were quantified using a semi-automatic segmentation technique and FreeSurfer cortical thickness was used to measure regional grey matter loss. RESULTS: Although NPS were frequent across the five disease groups, participants with frontotemporal dementia had the highest frequency of hyperactivity, apathy, and affective subsyndromes compared to other groups, whilst psychotic subsyndrome was high in both frontotemporal dementia and Parkinson's disease. Results from univariate and multivariate results showed that various predictors were associated with neuropsychiatric subsyndromes, especially cortical thickness in the inferior frontal, cingulate, and insula regions, sex(female), global cognition, and basal ganglia-thalamus white matter hyperintensities. CONCLUSIONS: In participants with neurodegenerative and cerebrovascular diseases, our results suggest that smaller cortical thickness and white matter hyperintensity burden in several cortical-subcortical structures may contribute to the development of NPS. Further studies investigating the mechanisms that determine the progression of NPS in various neurodegenerative and cerebrovascular diseases are needed.
Subject(s)
Cerebrovascular Disorders , Cognitive Dysfunction , Frontotemporal Dementia , Parkinson Disease , White Matter , Humans , Female , White Matter/diagnostic imaging , Cognitive Dysfunction/psychology , Cerebrovascular Disorders/complications , Cerebrovascular Disorders/diagnostic imaging , Magnetic Resonance ImagingABSTRACT
BACKGROUND: Acute change in gait speed while performing a mental task [dual-task gait cost (DTC)], and hyperintensity magnetic resonance imaging signals in white matter are both important disability predictors in older individuals with history of stroke (poststroke). It is still unclear, however, whether DTC is associated with overall hyperintensity volume from specific major brain regions in poststroke. METHODS: This is a cohort study with a total of 123 older (69 ± 7 years of age) participants with history of stroke were included from the Ontario Neurodegenerative Disease Research Initiative. Participants were clinically assessed and had gait performance assessed under single- and dual-task conditions. Structural neuroimaging data were analyzed to measure both, white matter hyperintensity (WMH) and normal appearing volumes. Percentage of WMH volume in frontal, parietal, occipital, and temporal lobes as well as subcortical hyperintensities in basal ganglia + thalamus were the main outcomes. Multivariate models investigated associations between DTC and hyperintensity volumes, adjusted for age, sex, years of education, global cognition, vascular risk factors, APOE4 genotype, residual sensorimotor symptoms from previous stroke and brain volume. RESULTS: There was a significant positive global linear association between DTC and hyperintensity burden (adjusted Wilks' λ = .87, P = .01). Amongst all WMH volumes, hyperintensity burden from basal ganglia + thalamus provided the most significant contribution to the global association (adjusted ß = .008, η2 = .03; P = .04), independently of brain atrophy. CONCLUSIONS: In poststroke, increased DTC may be an indicator of larger white matter damages, specifically in subcortical regions, which can potentially affect the overall cognitive processing and decrease gait automaticity by increasing the cortical control over patients' locomotion.
Subject(s)
Neurodegenerative Diseases , Stroke , White Matter , Humans , Aged , White Matter/diagnostic imaging , White Matter/pathology , Cohort Studies , Neurodegenerative Diseases/pathology , Brain/diagnostic imaging , Brain/pathology , Gait , Stroke/complications , Stroke/diagnostic imaging , Stroke/pathology , Magnetic Resonance ImagingABSTRACT
BACKGROUND AND PURPOSE: The pathophysiology of Parkinson's disease (PD) negatively affects brain network connectivity, and in the presence of brain white matter hyperintensities (WMHs) cognitive and motor impairments seem to be aggravated. However, the role of WMHs in predicting accelerating symptom worsening remains controversial. The objective was to investigate whether location and segmental brain WMH burden at baseline predict cognitive and motor declines in PD after 2 years. METHODS: Ninety-eight older adults followed longitudinally from Ontario Neurodegenerative Diseases Research Initiative with PD of 3-8 years in duration were included. Percentages of WMH volumes at baseline were calculated by location (deep and periventricular) and by brain region (frontal, temporal, parietal, occipital lobes and basal ganglia + thalamus). Cognitive and motor changes were assessed from baseline to 2-year follow-up. Specifically, global cognition, attention, executive function, memory, visuospatial abilities and language were assessed as were motor symptoms evaluated using the Movement Disorder Society Unified Parkinson's Disease Rating Scale Part III, spatial-temporal gait variables, Freezing of Gait Questionnaire and Activities Specific Balance Confidence Scale. RESULTS: Regression analysis adjusted for potential confounders showed that total and periventricular WMHs at baseline predicted decline in global cognition (p < 0.05). Also, total WMH burden predicted the decline of executive function (p < 0.05). Occipital WMH volumes also predicted decline in global cognition, visuomotor attention and visuospatial memory declines (p < 0.05). WMH volumes at baseline did not predict motor decline. CONCLUSION: White matter hyperintensity burden at baseline predicted cognitive but not motor decline in early to mid-stage PD. The motor decline observed after 2 years in these older adults with PD is probably related to the primary neurodegenerative process than comorbid white matter pathology.
Subject(s)
Cognitive Dysfunction , Gait Disorders, Neurologic , Neurodegenerative Diseases , Parkinson Disease , White Matter , Humans , Aged , White Matter/pathology , Neurodegenerative Diseases/pathology , Ontario , Magnetic Resonance Imaging/methods , Cognition/physiology , Cognitive Dysfunction/pathologyABSTRACT
White matter (WM) injury is frequently observed along with dementia. Positron emission tomography with amyloid-ligands (Aß-PET) recently gained interest for detecting WM injury. Yet, little is understood about the origin of the altered Aß-PET signal in WM regions. Here, we investigated the relative contributions of diffusion MRI-based microstructural alterations, including free water and tissue-specific properties, to Aß-PET in WM and to cognition. We included a unique cohort of 115 participants covering the spectrum of low-to-severe white matter hyperintensity (WMH) burden and cognitively normal to dementia. We applied a bi-tensor diffusion-MRI model that differentiates between (i) the extracellular WM compartment (represented via free water), and (ii) the fiber-specific compartment (via free water-adjusted fractional anisotropy [FA]). We observed that, in regions of WMH, a decrease in Aß-PET related most closely to higher free water and higher WMH volume. In contrast, in normal-appearing WM, an increase in Aß-PET related more closely to higher cortical Aß (together with lower free water-adjusted FA). In relation to cognitive impairment, we observed a closer relationship with higher free water than with either free water-adjusted FA or WM PET. Our findings support free water and Aß-PET as markers of WM abnormalities in patients with mixed dementia, and contribute to a better understanding of processes giving rise to the WM PET signal.
Subject(s)
Alzheimer Disease , Dementia , Vascular Diseases , White Matter , Humans , White Matter/diagnostic imaging , White Matter/metabolism , Diffusion Tensor Imaging/methods , Cognition/physiology , Water/metabolism , Dementia/diagnostic imaging , Alzheimer Disease/diagnostic imaging , Alzheimer Disease/metabolismABSTRACT
It remains unclear to what extent cerebrovascular burden relates to amyloid beta (Aß) deposition, neurodegeneration, and cognitive dysfunction in mixed disease populations with small vessel disease and Alzheimer's disease (AD) pathology. In 120 subjects, we investigated the association of vascular burden (white matter hyperintensity [WMH] volumes) with cognition. Using mediation analyses, we tested the indirect effects of WMH on cognition via Aß deposition (18 F-AV45 positron emission tomography [PET]) and neurodegeneration (cortical thickness or 18 F fluorodeoxyglucose PET) in AD signature regions. We observed that increased total WMH volume was associated with poorer performance in all tested cognitive domains, with the strongest effects observed for semantic fluency. These relationships were mediated mainly via cortical thinning, particularly of the temporal lobe, and to a lesser extent serially mediated via Aß and cortical thinning of AD signature regions. WMH volumes differentially impacted cognition depending on lobar location and Aß status. In summary, our study suggests mainly an amyloid-independent pathway in which vascular burden affects cognitive function via localized neurodegeneration. HIGHLIGHTS: Alzheimer's disease often co-exists with vascular pathology. We studied a unique cohort enriched for high white matter hyperintensities (WMH). High WMH related to cognitive impairment of semantic fluency and executive function. This relationship was mediated via temporo-parietal atrophy rather than metabolism. This relationship was, to lesser extent, serially mediated via amyloid beta and atrophy.
Subject(s)
Alzheimer Disease , Cognitive Dysfunction , White Matter , Humans , Alzheimer Disease/pathology , Amyloid beta-Peptides/metabolism , Cerebral Cortical Thinning/pathology , Magnetic Resonance Imaging , Cognition , Cognitive Dysfunction/metabolism , Positron-Emission Tomography , Amyloid/metabolism , Atrophy/pathology , White Matter/pathologyABSTRACT
BACKGROUND AND PURPOSE: The thalamus is a key brain hub that is globally connected to many cortical regions. Previous work highlights thalamic contributions to multiple cognitive functions, but few studies have measured thalamic volume changes or cognitive correlates. This study investigates associations between thalamic volumes and post-stroke cognitive function. METHODS: Participants with non-thalamic brain infarcts (3-42 months) underwent MRI and cognitive testing. Focal infarcts and thalami were traced manually. In cases with bilateral infarcts, the side of the primary infarct volume defined the hemisphere involved. Brain parcellation and volumetrics were extracted using a standardized and previously validated neuroimaging pipeline. Age and gender-matched healthy controls provided normal comparative thalamic volumes. Thalamic atrophy was considered when the volume exceeded 2 standard deviations greater than the controls. RESULTS: Thalamic volumes ipsilateral to the infarct in stroke patients (n=55) were smaller than left (4.4 ± 1.4 vs. 5.4 ± 0.5 cc, p < 0.001) and right (4.4 ± 1.4 vs. 5.5 ± 0.6 cc, p < 0.001) thalamic volumes in the controls. After controlling for head-size and global brain atrophy, infarct volume independently correlated with ipsilateral thalamic volume (ß= -0.069, p=0.024). Left thalamic atrophy correlated significantly with poorer cognitive performance (ß = 4.177, p = 0.008), after controlling for demographics and infarct volumes. CONCLUSIONS: Our results suggest that the remote effect of infarction on ipsilateral thalamic volume is associated with global post-stroke cognitive impairment.
Subject(s)
Cognitive Dysfunction , Stroke , Humans , Stroke/complications , Stroke/diagnostic imaging , Stroke/pathology , Thalamus/diagnostic imaging , Brain Infarction/complications , Brain Infarction/diagnostic imaging , Magnetic Resonance Imaging/methods , Cognitive Dysfunction/diagnostic imaging , Cognitive Dysfunction/etiology , Atrophy/pathologyABSTRACT
Alterations in tissue microstructure in normal-appearing white matter (NAWM), specifically measured by diffusion tensor imaging (DTI) fractional anisotropy (FA), have been associated with cognitive outcomes following stroke. The purpose of this study was to comprehensively compare conventional DTI measures of tissue microstructure in NAWM to diverse vascular brain lesions in people with cerebrovascular disease (CVD) and to examine associations between FA in NAWM and cerebrovascular risk factors. DTI metrics including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) were measured in cerebral tissues and cerebrovascular anomalies from 152 people with CVD participating in the Ontario Neurodegenerative Disease Research Initiative (ONDRI). Ten cerebral tissue types were segmented including NAWM, and vascular lesions including stroke, periventricular and deep white matter hyperintensities, periventricular and deep lacunar infarcts, and perivascular spaces (PVS) using T1-weighted, proton density-weighted, T2-weighted, and fluid attenuated inversion recovery MRI scans. Mean DTI metrics were measured in each tissue region using a previously developed DTI processing pipeline and compared between tissues using multivariate analysis of covariance. Associations between FA in NAWM and several CVD risk factors were also examined. DTI metrics in vascular lesions differed significantly from healthy tissue. Specifically, all tissue types had significantly different MD values, while FA was also found to be different in most tissue types. FA in NAWM was inversely related to hypertension and modified Rankin scale (mRS). This study demonstrated the differences between conventional DTI metrics, FA, MD, AD, and RD, in cerebral vascular lesions and healthy tissue types. Therefore, incorporating DTI to characterize the integrity of the tissue microstructure could help to define the extent and severity of various brain vascular anomalies. The association between FA within NAWM and clinical evaluation of hypertension and disability provides further evidence that white matter microstructural integrity is impacted by cerebrovascular function.
ABSTRACT
PURPOSE: Magnetic resonance imaging (MRI) scanner-specific geometric distortions may contribute to scanner induced variability and decrease volumetric measurement precision for multi-site studies. The purpose of this study was to determine whether geometric distortion correction increases the precision of brain volumetric measurements in a multi-site multi-scanner study. METHODS: Geometric distortion variation was quantified over a one-year period at 10 sites using the distortion fields estimated from monthly 3D T1-weighted MRI geometrical phantom scans. The variability of volume and distance measurements were quantified using synthetic volumes and a standard quantitative MRI (qMRI) phantom. The effects of geometric distortion corrections on MRI derived volumetric measurements of the human brain were assessed in two subjects scanned on each of the 10 MRI scanners and in 150 subjects with cerebrovascaular disease (CVD) acquired across imaging sites. RESULTS: Geometric distortions were found to vary substantially between different MRI scanners but were relatively stable on each scanner over a one-year interval. Geometric distortions varied spatially, increasing in severity with distance from the magnet isocenter. In measurements made with the qMRI phantom, the geometric distortion correction decreased the standard deviation of volumetric assessments by 35% and distance measurements by 42%. The average coefficient of variance decreased by 16% in gray matter and white matter volume estimates in the two subjects scanned on the 10 MRI scanners. CONCLUSION: Geometric distortion correction using an up-to-date correction field is recommended to increase precision in volumetric measurements made from MRI images.
Subject(s)
Brain , Magnetic Resonance Imaging , Brain/diagnostic imaging , Humans , Magnetic Resonance Imaging/methods , Phantoms, ImagingABSTRACT
BACKGROUND: Although previously thought to be asymptomatic, recent studies have suggested that magnetic resonance imaging-visible perivascular spaces (PVS) in the basal ganglia (BG-PVS) of patients with Parkinson's disease (PD) may be markers of motor disability and cognitive decline. In addition, a pathogenic and risk profile difference between small (≤3-mm diameter) and large (>3-mm diameter) PVS has been suggested. OBJECTIVE: The aim of this study was to examine associations between quantitative measures of large and small BG-PVS, global cognition, and motor/nonmotor features in a multicenter cohort of patients with PD. METHODS: We performed a cross-sectional study examining the association between large and small BG-PVS with Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS-UPDRS) Parts I-IV and cognition (Montreal Cognitive Assessment) in 133 patients with PD enrolled in the Ontario Neurodegenerative Disease Research Initiative study. RESULTS: Patients with PD with small BG-PVS demonstrated an association with MDS-UPDRS Parts I (P = 0.008) and II (both P = 0.02), whereas patients with large BG-PVS demonstrated an association with MDS-UPDRS Parts III (P < 0.0001) and IV (P < 0.001). BG-PVS were not correlated with cognition. CONCLUSIONS: Small BG-PVS are associated with motor and nonmotor aspects of experiences in daily living, while large BG-PVS are associated with the motor symptoms and motor complications. © 2022 International Parkinson and Movement Disorder Society.
Subject(s)
Disabled Persons , Motor Disorders , Neurodegenerative Diseases , Parkinson Disease , Basal Ganglia/diagnostic imaging , Basal Ganglia/pathology , Cross-Sectional Studies , Humans , Magnetic Resonance Imaging , Neurodegenerative Diseases/pathology , Parkinson Disease/complicationsABSTRACT
White matter hyperintensities (WMHs) are frequently observed on structural neuroimaging of elderly populations and are associated with cognitive decline and increased risk of dementia. Many existing WMH segmentation algorithms produce suboptimal results in populations with vascular lesions or brain atrophy, or require parameter tuning and are computationally expensive. Additionally, most algorithms do not generate a confidence estimate of segmentation quality, limiting their interpretation. MRI-based segmentation methods are often sensitive to acquisition protocols, scanners, noise-level, and image contrast, failing to generalize to other populations and out-of-distribution datasets. Given these concerns, we propose a novel Bayesian 3D convolutional neural network with a U-Net architecture that automatically segments WMH, provides uncertainty estimates of the segmentation output for quality control, and is robust to changes in acquisition protocols. We also provide a second model to differentiate deep and periventricular WMH. Four hundred thirty-two subjects were recruited to train the CNNs from four multisite imaging studies. A separate test set of 158 subjects was used for evaluation, including an unseen multisite study. We compared our model to two established state-of-the-art techniques (BIANCA and DeepMedic), highlighting its accuracy and efficiency. Our Bayesian 3D U-Net achieved the highest Dice similarity coefficient of 0.89 ± 0.08 and the lowest modified Hausdorff distance of 2.98 ± 4.40 mm. We further validated our models highlighting their robustness on "clinical adversarial cases" simulating data with low signal-to-noise ratio, low resolution, and different contrast (stemming from MRI sequences with different parameters). Our pipeline and models are available at: https://hypermapp3r.readthedocs.io.
Subject(s)
Leukoaraiosis , White Matter , Aged , Bayes Theorem , Humans , Image Processing, Computer-Assisted , Leukoaraiosis/pathology , Magnetic Resonance Imaging/methods , Uncertainty , White Matter/diagnostic imaging , White Matter/pathologyABSTRACT
RATIONALE: We evaluated K-means clustering to classify amyloid brain PETs as positive or negative. PATIENTS AND METHODS: Sixty-six participants (31 men, 35 women; age range, 52-81 years) were recruited through a multicenter observational study: 19 cognitively normal, 25 mild cognitive impairment, and 22 dementia (11 Alzheimer disease, 3 subcortical vascular cognitive impairment, and 8 Parkinson-Lewy Body spectrum disorder). As part of the neurocognitive and imaging evaluation, each participant had an 18F-flutemetamol (Vizamyl, GE Healthcare) brain PET. All studies were processed using Cortex ID software (General Electric Company, Boston, MA) to calculate SUV ratios in 19 regions of interest and clinically interpreted by 2 dual-certified radiologists/nuclear medicine physicians, using MIM software (MIM Software Inc, Cleveland, OH), blinded to the quantitative analysis, with final interpretation based on consensus. K-means clustering was retrospectively used to classify the studies from the quantitative data. RESULTS: Based on clinical interpretation, 46 brain PETs were negative and 20 were positive for amyloid deposition. Of 19 cognitively normal participants, 1 (5%) had a positive 18F-flutemetamol brain PET. Of 25 participants with mild cognitive impairment, 9 (36%) had a positive 18F-flutemetamol brain PET. Of 22 participants with dementia, 10 (45%) had a positive 18F-flutemetamol brain PET; 7 of 11 participants with Alzheimer disease (64%), 1 of 3 participants with vascular cognitive impairment (33%), and 2 of 8 participants with Parkinson-Lewy Body spectrum disorder (25%) had a positive 18F-flutemetamol brain PET. Using clinical interpretation as the criterion standard, K-means clustering (K = 2) gave sensitivity of 95%, specificity of 98%, and accuracy of 97%. CONCLUSIONS: K-means clustering may be a powerful algorithm for classifying amyloid brain PET.
Subject(s)
Aniline Compounds , Benzothiazoles , Brain/diagnostic imaging , Image Processing, Computer-Assisted/methods , Positron-Emission Tomography , Aged , Aged, 80 and over , Amyloid/metabolism , Brain/metabolism , Cluster Analysis , Female , Humans , Male , Middle Aged , Neurocognitive Disorders/diagnostic imaging , Neurocognitive Disorders/metabolism , Retrospective StudiesABSTRACT
Successful segmentation of the total intracranial vault (ICV) and ventricles is of critical importance when studying neurodegeneration through neuroimaging. We present iCVMapper and VentMapper, robust algorithms that use a convolutional neural network (CNN) to segment the ICV and ventricles from both single and multi-contrast MRI data. Our models were trained on a large dataset from two multi-site studies (N = 528 subjects for ICV, N = 501 for ventricular segmentation) consisting of older adults with varying degrees of cerebrovascular lesions and atrophy, which pose significant challenges for most segmentation approaches. The models were tested on 238 participants, including subjects with vascular cognitive impairment and high white matter hyperintensity burden. Two of the three test sets came from studies not used in the training dataset. We assessed our algorithms relative to four state-of-the-art ICV extraction methods (MONSTR, BET, Deep Extraction, FreeSurfer, DeepMedic), as well as two ventricular segmentation tools (FreeSurfer, DeepMedic). Our multi-contrast models outperformed other methods across many of the evaluation metrics, with average Dice coefficients of 0.98 and 0.96 for ICV and ventricular segmentation respectively. Both models were also the most time efficient, segmenting the structures in orders of magnitude faster than some of the other available methods. Our networks showed an increased accuracy with the use of a conditional random field (CRF) as a post-processing step. We further validated both segmentation models, highlighting their robustness to images with lower resolution and signal-to-noise ratio, compared to tested techniques. The pipeline and models are available at: https://icvmapp3r.readthedocs.io and https://ventmapp3r.readthedocs.io to enable further investigation of the roles of ICV and ventricles in relation to normal aging and neurodegeneration in large multi-site studies.
Subject(s)
Cerebral Ventricles/diagnostic imaging , Image Processing, Computer-Assisted , Neural Networks, Computer , Aged , Atrophy , Cerebral Ventricles/pathology , Humans , Magnetic Resonance Imaging , NeuroimagingABSTRACT
The Ontario Neurodegenerative Research Initiative (ONDRI) is a 3 years multi-site prospective cohort study that has acquired comprehensive multiple assessment platform data, including 3T structural MRI, from neurodegenerative patients with Alzheimer's disease, mild cognitive impairment, Parkinson's disease, amyotrophic lateral sclerosis, frontotemporal dementia, and cerebrovascular disease. This heterogeneous cross-section of patients with complex neurodegenerative and neurovascular pathologies pose significant challenges for standard neuroimaging tools. To effectively quantify regional measures of normal and pathological brain tissue volumes, the ONDRI neuroimaging platform implemented a semi-automated MRI processing pipeline that was able to address many of the challenges resulting from this heterogeneity. The purpose of this paper is to serve as a reference and conceptual overview of the comprehensive neuroimaging pipeline used to generate regional brain tissue volumes and neurovascular marker data that will be made publicly available online.
ABSTRACT
PURPOSE: The aim of this study was to evaluate random forests (RFs) to identify ROIs on F-florbetapir and F-FDG PET associated with Montreal Cognitive Assessment (MoCA) score. MATERIALS AND METHODS: Fifty-seven subjects with significant white matter disease presenting with either transient ischemic attack/lacunar stroke or mild cognitive impairment from early Alzheimer disease, enrolled in a multicenter prospective observational trial, had MoCA and F-florbetapir PET; 55 had F-FDG PET. Scans were processed using the MINC toolkit to generate SUV ratios, normalized to cerebellar gray matter (F-florbetapir PET), or pons (F-FDG PET). SUV ratio data and MoCA score were used for supervised training of RFs programmed in MATLAB. RESULTS: F-Florbetapir PETs were randomly divided into 40 training and 17 testing scans; 100 RFs of 1000 trees, constructed from a random subset of 16 training scans and 20 ROIs, identified ROIs associated with MoCA score: right posterior cingulate gyrus, right anterior cingulate gyrus, left precuneus, left posterior cingulate gyrus, and right precuneus. Amyloid increased with decreasing MoCA score. F-FDG PETs were randomly divided into 40 training and 15 testing scans; 100 RFs of 1000 trees, each tree constructed from a random subset of 16 training scans and 20 ROIs, identified ROIs associated with MoCA score: left fusiform gyrus, left precuneus, left posterior cingulate gyrus, right precuneus, and left middle orbitofrontal gyrus. F-FDG decreased with decreasing MoCA score. CONCLUSIONS: Random forests help pinpoint clinically relevant ROIs associated with MoCA score; amyloid increased and F-FDG decreased with decreasing MoCA score, most significantly in the posterior cingulate gyrus.
Subject(s)
Amyloid/metabolism , Brain/diagnostic imaging , Brain/metabolism , Fluorodeoxyglucose F18 , Image Processing, Computer-Assisted/methods , Machine Learning , Positron-Emission Tomography , Aged , Aniline Compounds , Ethylene Glycols , Female , Humans , Male , Middle AgedABSTRACT
Hippocampal volumetry is a critical biomarker of aging and dementia, and it is widely used as a predictor of cognitive performance; however, automated hippocampal segmentation methods are limited because the algorithms are (a) not publicly available, (b) subject to error with significant brain atrophy, cerebrovascular disease and lesions, and/or (c) computationally expensive or require parameter tuning. In this study, we trained a 3D convolutional neural network using 259 bilateral manually delineated segmentations collected from three studies, acquired at multiple sites on different scanners with variable protocols. Our training dataset consisted of elderly cases difficult to segment due to extensive atrophy, vascular disease, and lesions. Our algorithm, (HippMapp3r), was validated against four other publicly available state-of-the-art techniques (HippoDeep, FreeSurfer, SBHV, volBrain, and FIRST). HippMapp3r outperformed the other techniques on all three metrics, generating an average Dice of 0.89 and a correlation coefficient of 0.95. It was two orders of magnitude faster than some of the tested techniques. Further validation was performed on 200 subjects from two other disease populations (frontotemporal dementia and vascular cognitive impairment), highlighting our method's low outlier rate. We finally tested the methods on real and simulated "clinical adversarial" cases to study their robustness to corrupt, low-quality scans. The pipeline and models are available at: https://hippmapp3r.readthedocs.ioto facilitate the study of the hippocampus in large multisite studies.
Subject(s)
Dementia/diagnostic imaging , Dementia/pathology , Hippocampus/diagnostic imaging , Hippocampus/pathology , Image Interpretation, Computer-Assisted , Magnetic Resonance Imaging , Neural Networks, Computer , Neuroimaging , Aged , Atrophy , Female , Humans , Image Interpretation, Computer-Assisted/methods , Image Interpretation, Computer-Assisted/standards , Magnetic Resonance Imaging/methods , Magnetic Resonance Imaging/standards , Male , Neuroimaging/methods , Neuroimaging/standardsABSTRACT
PURPOSE: To evaluate random forests (RFs) as a supervised machine learning algorithm to classify amyloid brain PET as positive or negative for amyloid deposition and identify key regions of interest for stratification. METHODS: The data set included 57 baseline F-florbetapir (Amyvid; Lilly, Indianapolis, IN) brain PET scans in participants with severe white matter disease, presenting with either transient ischemic attack/lacunar stroke or mild cognitive impairment from early Alzheimer disease, enrolled in a multicenter prospective observational trial. Scans were processed using the MINC toolkit to generate SUV ratios, normalized to cerebellar gray matter, and clinically read by 2 nuclear medicine physicians with interpretation based on consensus (35 negative, 22 positive). SUV ratio data and clinical reads were used for supervised training of an RF classifier programmed in MATLAB. RESULTS: A 10,000-tree RF, each tree using 15 randomly selected cases and 20 randomly selected features (SUV ratio per region of interest), with 37 cases for training and 20 cases for testing, had sensitivity = 86% (95% confidence interval [CI], 42%-100%), specificity = 92% (CI, 64%-100%), and classification accuracy = 90% (CI, 68%-99%). The most common features at the root node (key regions for stratification) were (1) left posterior cingulate (1039 trees), (2) left middle frontal gyrus (1038 trees), (3) left precuneus (857 trees), (4) right anterior cingulate gyrus (655 trees), and (5) right posterior cingulate (588 trees). CONCLUSIONS: Random forests can classify brain PET as positive or negative for amyloid deposition and suggest key clinically relevant, regional features for classification.