Is white matter hyperintensity burden associated with cognitive and motor impairment in patients with parkinson's disease? A systematic review and meta-analysis.

White matter damage quantified as white matter hyperintensities (WMH) may aggravate cognitive and motor impairments, but whether and how WMH burden impacts these problems in Parkinson's disease (PD) is not fully understood. This study aimed to examine the association between WMH and cognitive and motor performance in PD through a systematic review and meta-analysis. We compared the WMH burden across the cognitive spectrum (cognitively normal, mild cognitive impairment, dementia) in PD including controls. Motor signs were compared in PD with low/negative and high/positive WMH burden. We compared baseline WMH burden of PD who did and did not convert to MCI or dementia. MEDLINE and EMBASE databases were used to conduct the literature search resulting in 50 studies included for data extraction. Increased WMH burden was found in individuals with PD compared with individuals without PD (i.e. control) and across the cognitive spectrum in PD (i.e. PD, PD-MCI, PDD). Individuals with PD with high/positive WMH burden had worse global cognition, executive function, and attention. Similarly, PD with high/positive WMH presented worse motor signs compared with individuals presenting low/negative WMH burden. Only three longitudinal studies were retrieved from our search and they showed that PD who converted to MCI or dementia, did not have significantly higher WMH burden at baseline, although no data was provided on WMH burden changes during the follow up. We conclude, based on cross-sectional studies, that WMH burden appears to increase with PD worse cognitive and motor status in PD.

PET Imaging in Dementia: Mini-Review and Canadian Perspective for Clinical Use.

PET imaging is increasingly recognized as an important diagnostic tool to investigate patients with cognitive disturbances of possible neurodegenerative origin. PET with 2-[18F]fluoro-2-deoxy-D-glucose ([18F]FDG), assessing glucose metabolism, provides a measure of neurodegeneration and allows a precise differential diagnosis among the most common neurodegenerative diseases, such as Alzheimer's disease, frontotemporal dementia or dementia with Lewy bodies. PET tracers specific for the pathological deposits characteristic of different neurodegenerative processes, namely amyloid and tau deposits typical of Alzheimer's Disease, allow the visualization of these aggregates in vivo. [18F]FDG and amyloid PET imaging have reached a high level of clinical validity and are since 2022 investigations that can be offered to patients in standard clinical care in most of Canada.This article will briefly review and summarize the current knowledge on these diagnostic tools, their integration into diagnostic algorithms as well as perspectives for future developments.

Quantifying cerebral microbleeds using quantitative susceptibility mapping from magnetization-prepared rapid gradient-echo.

T1-weighted magnetization-prepared rapid gradient-echo (MPRAGE) is commonly included in brain studies for structural imaging using magnitude images; however, its phase images can provide an opportunity to assess microbleed burden using quantitative susceptibility mapping (QSM). This potential application for MPRAGE-based QSM was evaluated using in vivo and simulated measurements. Possible factors affecting image quality were also explored. Detection sensitivity was evaluated against standard multiecho gradient echo (MEGE) QSM using 3-T in vivo data of 15 subjects with a combined total of 108 confirmed microbleeds. The two methods were compared based on the microbleed size and susceptibility measurements. In addition, simulations explored the detection sensitivity of MPRAGE-QSM at different representative magnetic field strengths and echo times using microbleeds of different size, susceptibility, and location. Results showed that in vivo microbleeds appeared to be smaller (× 0.54) and of higher mean susceptibility (× 1.9) on MPRAGE-QSM than on MEGE-QSM, but total susceptibility estimates were in closer agreement (slope: 0.97, r2 : 0.94), and detection sensitivity was comparable. In simulations, QSM at 1.5 T had a low contrast-to-noise ratio that obscured the detection of many microbleeds. Signal-to-noise ratio (SNR) levels at 3 T and above resulted in better contrast and increased detection. The detection rates for microbleeds of minimum one-voxel diameter and 0.4-ppm susceptibility were 0.55, 0.80, and 0.88 at SNR levels of 1.5, 3, and 7 T, respectively. Size and total susceptibility estimates were more consistent than mean susceptibility estimates, which showed size-dependent underestimation. MPRAGE-QSM provides an opportunity to detect and quantify the size and susceptibility of microbleeds of at least one-voxel diameter at B0  of 3 T or higher with no additional time cost, when standard T2 *-weighted images are not available or have inadequate spatial resolution. The total susceptibility measure is more robust against sequence variations and might allow combining data from different protocols.

A multi-center distributed learning approach for Parkinson's disease classification using the traveling model paradigm.

Distributed learning is a promising alternative to central learning for machine learning (ML) model training, overcoming data-sharing problems in healthcare. Previous studies exploring federated learning (FL) or the traveling model (TM) setup for medical image-based disease classification often relied on large databases with a limited number of centers or simulated artificial centers, raising doubts about real-world applicability. This study develops and evaluates a convolution neural network (CNN) for Parkinson's disease classification using data acquired by 83 diverse real centers around the world, mostly contributing small training samples. Our approach specifically makes use of the TM setup, which has proven effective in scenarios with limited data availability but has never been used for image-based disease classification. Our findings reveal that TM is effective for training CNN models, even in complex real-world scenarios with variable data distributions. After sufficient training cycles, the TM-trained CNN matches or slightly surpasses the performance of the centrally trained counterpart (AUROC of 83% vs. 80%). Our study highlights, for the first time, the effectiveness of TM in 3D medical image classification, especially in scenarios with limited training samples and heterogeneous distributed data. These insights are relevant for situations where ML models are supposed to be trained using data from small or remote medical centers, and rare diseases with sparse cases. The simplicity of this approach enables a broad application to many deep learning tasks, enhancing its clinical utility across various contexts and medical facilities.

Exploiting macro- and micro-structural brain changes for improved Parkinson's disease classification from MRI data.

Parkinson's disease (PD) is the second most common neurodegenerative disease. Accurate PD diagnosis is crucial for effective treatment and prognosis but can be challenging, especially at early disease stages. This study aimed to develop and evaluate an explainable deep learning model for PD classification from multimodal neuroimaging data. The model was trained using one of the largest collections of T1-weighted and diffusion-tensor magnetic resonance imaging (MRI) datasets. A total of 1264 datasets from eight different studies were collected, including 611 PD patients and 653 healthy controls (HC). These datasets were pre-processed and non-linearly registered to the MNI PD25 atlas. Six imaging maps describing the macro- and micro-structural integrity of brain tissues complemented with age and sex parameters were used to train a convolutional neural network (CNN) to classify PD/HC subjects. Explainability of the model's decision-making was achieved using SmoothGrad saliency maps, highlighting important brain regions. The CNN was trained using a 75%/10%/15% train/validation/test split stratified by diagnosis, sex, age, and study, achieving a ROC-AUC of 0.89, accuracy of 80.8%, specificity of 82.4%, and sensitivity of 79.1% on the test set. Saliency maps revealed that diffusion tensor imaging data, especially fractional anisotropy, was more important for the classification than T1-weighted data, highlighting subcortical regions such as the brainstem, thalamus, amygdala, hippocampus, and cortical areas. The proposed model, trained on a large multimodal MRI database, can classify PD patients and HC subjects with high accuracy and clinically reasonable explanations, suggesting that micro-structural brain changes play an essential role in the disease course.

Identifying Biases in a Multicenter MRI Database for Parkinson's Disease Classification: Is the Disease Classifier a Secret Site Classifier?

Sharing multicenter imaging datasets can be advantageous to increase data diversity and size but may lead to spurious correlations between site-related biological and non-biological image features and target labels, which machine learning (ML) models may exploit as shortcuts. To date, studies analyzing how and if deep learning models may use such effects as a shortcut are scarce. Thus, the aim of this work was to investigate if site-related effects are encoded in the feature space of an established deep learning model designed for Parkinson's disease (PD) classification based on T1-weighted MRI datasets. Therefore, all layers of the PD classifier were frozen, except for the last layer of the network, which was replaced by a linear layer that was exclusively re-trained to predict three potential bias types (biological sex, scanner type, and originating site). Our findings based on a large database consisting of 1880 MRI scans collected across 41 centers show that the feature space of the established PD model (74% accuracy) can be used to classify sex (75% accuracy), scanner type (79% accuracy), and site location (71% accuracy) with high accuracies despite this information never being explicitly provided to the PD model during original training. Overall, the results of this study suggest that trained image-based classifiers may use unwanted shortcuts that are not meaningful for the actual clinical task at hand. This finding may explain why many image-based deep learning models do not perform well when applied to data from centers not contributing to the training set.

Gray matter loss relates to dual task gait in Lewy body disorders and aging.

BACKGROUND: Within the spectrum of Lewy body disorders (LBD), both Parkinson's disease (PD) and dementia with Lewy bodies (DLB) are characterized by gait and balance disturbances, which become more prominent under dual-task (DT) conditions. The brain substrates underlying DT gait variations, however, remain poorly understood in LBD. OBJECTIVE: To investigate the relationship between gray matter volume loss and DT gait variations in LBD. METHODS: Seventy-nine participants including cognitively unimpaired PD, PD with mild cognitive impairment, PD with dementia (PDD), or DLB and 20 cognitively unimpaired controls were examined across a multi-site study. PDD and DLB were grouped together for analyses. Differences in gait speed between single and DT conditions were quantified by dual task cost (DTC). Cortical, subcortical, ventricle, and cerebellum brain volumes were obtained using FreeSurfer. Linear regression models were used to examine the relationship between gray matter volumes and DTC. RESULTS: Smaller amygdala and total cortical volumes, and larger ventricle volumes were associated with a higher DTC across LBD and cognitively unimpaired controls. No statistically significant interaction between group and brain volumes were found. Adding cognitive and motor covariates or white matter hyperintensity volumes separately to the models did not affect brain volume and DTC associations. CONCLUSION: Gray matter volume loss is associated with worse DT gait performance compared to single task gait, across cognitively unimpaired controls through and the LBD spectrum. Impairment in DT gait performance may be driven by age-related cortical neurodegeneration.

Prevention of Falls in Parkinson's Disease: Guidelines and Gaps.

Background: People living with Parkinson's disease (PD) have a high risk for falls. Objective: To examine gaps in falls prevention targeting people with PD as part of the Task Force on Global Guidelines for Falls in Older Adults. Methods: A Delphi consensus process was used to identify specific recommendations for falls in PD. The current narrative review was conducted as educational background with a view to identifying gaps in fall prevention. Results: A recent Cochrane review recommended exercises and structured physical activities for PD; however, the types of exercises and activities to recommend and PD subgroups likely to benefit require further consideration. Freezing of gait, reduced gait speed, and a prior history of falls are risk factors for falls in PD and should be incorporated in assessments to identify fall risk and target interventions. Multimodal and multi-domain fall prevention interventions may be beneficial. With advanced or complex PD, balance and strength training should be administered under supervision. Medications, particularly cholinesterase inhibitors, show promise for falls prevention. Identifying how to engage people with PD, their families, and health professionals in falls education and implementation remains a challenge. Barriers to the prevention of falls occur at individual, environmental, policy, and health system levels. Conclusion: Effective mitigation of fall risk requires specific targeting and strategies to reduce this debilitating and common problem in PD. While exercise is recommended, the types and modalities of exercise and how to combine them as interventions for different PD subgroups (cognitive impairment, freezing, advanced disease) need further study.

A Metabolomics Analysis of a Novel Phenotype of Older Adults at Higher Risk of Dementia.

BACKGROUND: Older adults presenting with dual-decline in cognition and walking speed face a 6-fold higher risk for dementia compared with those showing no decline. We hypothesized that the metabolomics profile of dual-decliners would be unique even before they show signs of decline in cognition and gait speed. OBJECTIVE: The objective of this study was to determine if plasma metabolomics signatures can discriminate dual-decliners from no decliners, purely cognitive decliners, and purely motor decliners prior to decline. METHODS: A retrospective cross-sectional study using baseline plasma for untargeted metabolomics analyses to investigate early signals of later dual-decline status in study participants (n = 76) with convenient sampling. Dual-decline was operationalized as decline in gait speed (>10 cm/s) and cognition (>2 points decline in Montreal Cognitive Assessment score) on at least two consecutive 6-monthly assessments. The participants' decliner status was evaluated 3 years after the blood sample was collected. Pair-wise comparison of detected compounds was completed using principal components and hierarchical clustering analyses. RESULTS: Analyses did not detect any cluster separation in untargeted metabolomes across baseline groups. However, follow-up analyses of specific molecules detected 4 compounds (17-Hydroxy-12-(hydroxymethyl)-10-oxo-8 oxapentacyclomethyl hexopyranoside, Fleroxacin, Oleic acid, and 5xi-11,12-Dihydroxyabieta-8(14),9(11),12-trien-20-oic acid) were at significantly higher concentration among the dual-decliners compared to non-decliners. The pure cognitive decliner group had significantly lower concentration of six compounds (1,3-nonanediol acetate, 4-(2-carboxyethyl)-2-methoxyphenyl beta-D-glucopyranosiduronic acid, oleic acid, 2E-3-[4-(sulfo-oxy)phenyl] acrylic acid, palmitelaidic acid, and myristoleic acid) compared to the non-decliner group. CONCLUSIONS: The unique metabolomics profile of dual-decliners warrants follow-up metabolomics analysis. Results may point to modifiable pathways.

Image-encoded biological and non-biological variables may be used as shortcuts in deep learning models trained on multisite neuroimaging data.

OBJECTIVE: This work investigates if deep learning (DL) models can classify originating site locations directly from magnetic resonance imaging (MRI) scans with and without correction for intensity differences. MATERIAL AND METHODS: A large database of 1880 T1-weighted MRI scans collected across 41 sites originally for Parkinson's disease (PD) classification was used to classify sites in this study. Forty-six percent of the datasets are from PD patients, while 54% are from healthy participants. After preprocessing the T1-weighted scans, 2 additional data types were generated: intensity-harmonized T1-weighted scans and log-Jacobian deformation maps resulting from nonlinear atlas registration. Corresponding DL models were trained to classify sites for each data type. Additionally, logistic regression models were used to investigate the contribution of biological (age, sex, disease status) and non-biological (scanner type) variables to the models' decision. RESULTS: A comparison of the 3 different types of data revealed that DL models trained using T1-weighted and intensity-harmonized T1-weighted scans can classify sites with an accuracy of 85%, while the model using log-Jacobian deformation maps achieved a site classification accuracy of 54%. Disease status and scanner type were found to be significant confounders. DISCUSSION: Our results demonstrate that MRI scans encode relevant site-specific information that models could use as shortcuts that cannot be removed using simple intensity harmonization methods. CONCLUSION: The ability of DL models to exploit site-specific biases as shortcuts raises concerns about their reliability, generalization, and deployability in clinical settings.

Machine learning-based prediction of longitudinal cognitive decline in early Parkinson's disease using multimodal features.

Patients with Parkinson's Disease (PD) often suffer from cognitive decline. Accurate prediction of cognitive decline is essential for early treatment of at-risk patients. The aim of this study was to develop and evaluate a multimodal machine learning model for the prediction of continuous cognitive decline in patients with early PD. We included 213 PD patients from the Parkinson's Progression Markers Initiative (PPMI) database. Machine learning was used to predict change in Montreal Cognitive Assessment (MoCA) score using the difference between baseline and 4-years follow-up data as outcome. Input features were categorized into four sets: clinical test scores, cerebrospinal fluid (CSF) biomarkers, brain volumes, and genetic variants. All combinations of input feature sets were added to a basic model, which consisted of demographics and baseline cognition. An iterative scheme using RReliefF-based feature ranking and support vector regression in combination with tenfold cross validation was used to determine the optimal number of predictive features and to evaluate model performance for each combination of input feature sets. Our best performing model consisted of a combination of the basic model, clinical test scores and CSF-based biomarkers. This model had 12 features, which included baseline cognition, CSF phosphorylated tau, CSF total tau, CSF amyloid-beta1-42, geriatric depression scale (GDS) scores, and anxiety scores. Interestingly, many of the predictive features in our model have previously been associated with Alzheimer's disease, showing the importance of assessing Alzheimer's disease pathology in patients with Parkinson's disease.

Effects of Exercise Alone or Combined With Cognitive Training and Vitamin D Supplementation to Improve Cognition in Adults With Mild Cognitive Impairment: A Randomized Clinical Trial.

Importance: Exercise, cognitive training, and vitamin D may enhance cognition in older adults with mild cognitive impairment (MCI). Objective: To determine whether aerobic-resistance exercises would improve cognition relative to an active control and if a multidomain intervention including exercises, computerized cognitive training, and vitamin D supplementation would show greater improvements than exercise alone. Design, Setting, and Participants: This randomized clinical trial (the SYNERGIC Study) was a multisite, double-masked, fractional factorial trial that evaluated the effects of aerobic-resistance exercise, computerized cognitive training, and vitamin D on cognition. Eligible participants were between ages 65 and 84 years with MCI enrolled from September 19, 2016, to April 7, 2020. Data were analyzed from February 2021 to December 2022. Interventions: Participants were randomized to 5 study arms and treated for 20 weeks: arm 1 (multidomain intervention with exercise, cognitive training, and vitamin D), arm 2 (exercise, cognitive training, and placebo vitamin D), arm 3 (exercise, sham cognitive training, and vitamin D), arm 4 (exercise, sham cognitive training, and placebo vitamin D), and arm 5 (control group with balance-toning exercise, sham cognitive training, and placebo vitamin D). The vitamin D regimen was a 10 000 IU dose 3 times weekly. Main Outcomes and Measures: Primary outcomes were changes in ADAS-Cog-13 and Plus variant at 6 months. Results: Among 175 randomized participants (mean [SD] age, 73.1 [6.6] years; 86 [49.1%] women), 144 (82%) completed the intervention and 133 (76%) completed the follow-up (month 12). At 6 months, all active arms (ie, arms 1 through 4) with aerobic-resistance exercise regardless of the addition of cognitive training or vitamin D, improved ADAS-Cog-13 when compared with control (mean difference, -1.79 points; 95% CI, -3.27 to -0.31 points; P = .02; d = 0.64). Compared with exercise alone (arms 3 and 4), exercise and cognitive training (arms 1 and 2) improved the ADAS-Cog-13 (mean difference, -1.45 points; 95% CI, -2.70 to -0.21 points; P = .02; d = 0.39). No significant improvement was found with vitamin D. Finally, the multidomain intervention (arm 1) improved the ADAS-Cog-13 score significantly compared with control (mean difference, -2.64 points; 95% CI, -4.42 to -0.80 points; P = .005; d = 0.71). Changes in ADAS-Cog-Plus were not significant. Conclusions and Relevance: In this clinical trial, older adults with MCI receiving aerobic-resistance exercises with sequential computerized cognitive training significantly improved cognition, although some results were inconsistent. Vitamin D supplementation had no effect. Our findings suggest that this multidomain intervention may improve cognition and potentially delay dementia onset in MCI. Trial Registration: ClinicalTrials.gov Identifier: NCT02808676.

Identifying key multi-modal predictors of incipient dementia in Parkinson's disease: a machine learning analysis and Tree SHAP interpretation.

Background: Persons with Parkinson's disease (PD) differentially progress to cognitive impairment and dementia. With a 3-year longitudinal sample of initially non-demented PD patients measured on multiple dementia risk factors, we demonstrate that machine learning classifier algorithms can be combined with explainable artificial intelligence methods to identify and interpret leading predictors that discriminate those who later converted to dementia from those who did not. Method: Participants were 48 well-characterized PD patients (Mbaseline age = 71.6; SD = 4.8; 44% female). We tested 38 multi-modal predictors from 10 domains (e.g., motor, cognitive) in a computationally competitive context to identify those that best discriminated two unobserved baseline groups, PD No Dementia (PDND), and PD Incipient Dementia (PDID). We used Random Forest (RF) classifier models for the discrimination goal and Tree SHapley Additive exPlanation (Tree SHAP) values for deep interpretation. Results: An excellent RF model discriminated baseline PDID from PDND (AUC = 0.84; normalized Matthews Correlation Coefficient = 0.76). Tree SHAP showed that ten leading predictors of PDID accounted for 62.5% of the model, as well as their relative importance, direction, and magnitude (risk threshold). These predictors represented the motor (e.g., poorer gait), cognitive (e.g., slower Trail A), molecular (up-regulated metabolite panel), demographic (age), imaging (ventricular volume), and lifestyle (activities of daily living) domains. Conclusion: Our data-driven protocol integrated RF classifier models and Tree SHAP applications to selectively identify and interpret early dementia risk factors in a well-characterized sample of initially non-demented persons with PD. Results indicate that leading dementia predictors derive from multiple complementary risk domains.

Subcortical volumes in cerebral amyloid angiopathy compared with Alzheimer's disease and controls.

Background: Previous reports have suggested that patients with cerebral amyloid angiopathy (CAA) may harbor smaller white matter, basal ganglia, and cerebellar volumes compared to age-matched healthy controls (HC) or patients with Alzheimer's disease (AD). We investigated whether CAA is associated with subcortical atrophy. Methods: The study was based on the multi-site Functional Assessment of Vascular Reactivity cohort and included 78 probable CAA (diagnosed according to the Boston criteria v2.0), 33 AD, and 70 HC. Cerebral and cerebellar volumes were extracted from brain 3D T1-weighted MRI using FreeSurfer (v6.0). Subcortical volumes, including total white matter, thalamus, basal ganglia, and cerebellum were reported as proportion (%) of estimated total intracranial volume. White matter integrity was quantified by the peak width of skeletonized mean diffusivity. Results: Participants in the CAA group were older (74.0 ± 7.0, female 44%) than the AD (69.7 ± 7.5, female 42%) and HC (68.8 ± 7.8, female 69%) groups. CAA participants had the highest white matter hyperintensity volume and worse white matter integrity of the three groups. After adjusting for age, sex, and study site, CAA participants had smaller putamen volumes (mean differences, -0.024% of intracranial volume; 95% confidence intervals, -0.041% to -0.006%; p = 0.005) than the HCs but not AD participants (-0.003%; -0.024 to 0.018%; p = 0.94). Other subcortical volumes including subcortical white matter, thalamus, caudate, globus pallidus, cerebellar cortex or cerebellar white matter were comparable between all three groups. Conclusion: In contrast to prior studies, we did not find substantial atrophy of subcortical volumes in CAA compared to AD or HCs, except for the putamen. Differences between studies may reflect heterogeneity in CAA presenting syndromes or severity.

Brain iron content in cerebral amyloid angiopathy using quantitative susceptibility mapping.

Introduction: Cerebral amyloid angiopathy (CAA) is a small vessel disease that causes covert and symptomatic brain hemorrhaging. We hypothesized that persons with CAA would have increased brain iron content detectable by quantitative susceptibility mapping (QSM) on magnetic resonance imaging (MRI), and that higher iron content would be associated with worse cognition. Methods: Participants with CAA (n = 21), mild Alzheimer's disease with dementia (AD-dementia; n = 14), and normal controls (NC; n = 83) underwent 3T MRI. Post-processing QSM techniques were applied to obtain susceptibility values for regions of the frontal and occipital lobe, thalamus, caudate, putamen, pallidum, and hippocampus. Linear regression was used to examine differences between groups, and associations with global cognition, controlling for multiple comparisons using the false discovery rate method. Results: No differences were found between regions of interest in CAA compared to NC. In AD, the calcarine sulcus had greater iron than NC (ß = 0.99 [95% CI: 0.44, 1.53], q < 0.01). However, calcarine sulcus iron content was not associated with global cognition, measured by the Montreal Cognitive Assessment (p > 0.05 for all participants, NC, CAA, and AD). Discussion: After correcting for multiple comparisons, brain iron content, measured via QSM, was not elevated in CAA compared to NC in this exploratory study.

Dual-task gait and white matter hyperintensities in Lewy body diseases: An exploratory analysis.

Background: Parkinson's disease (PD) and dementia with Lewy bodies (DLB) are part of a spectrum of Lewy body disorders, who exhibit a range of cognitive and gait impairments. Cognitive-motor interactions can be examined by performing a cognitive task while walking and quantified by a dual task cost (DTC). White matter hyperintensities (WMH) on magnetic resonance imaging have also been associated with both gait and cognition. Our goal was to examine the relationship between DTC and WMH in the Lewy body spectrum, hypothesizing DTC would be associated with increased WMH volume. Methods: Seventy-eight participants with PD, PD with mild cognitive impairment (PD-MCI), PD with dementia or DLB (PDD/DLB), and 20 cognitively unimpaired participants were examined in a multi-site study. Gait was measured on an electronic walkway during usual gait, counting backward, animal fluency, and subtracting sevens. WMH were quantified from magnetic resonance imaging using an automated pipeline and visual rating. A median split based on DTC was performed. Models included age as well as measures of global cognition and cardiovascular risk. Results: Compared to cognitively unimpaired participants, usual gait speed was lower and DTC was higher in PD-MCI and PDD/DLB. Low DTC participants had higher usual gait speed. WMH burden was greater in high counting DTC participants. Frontal WMH burden remained significant after adjusting for age, cardiovascular risk and global cognition. Conclusion: Increased DTC was associated with higher frontal WMH burden in Lewy body disorders after adjusting for age, cardiovascular risk, and global cognition. Higher DTC was associated with age.

Explainable classification of Parkinson's disease using deep learning trained on a large multi-center database of T1-weighted MRI datasets.

INTRODUCTION: Parkinson's disease (PD) is a severe neurodegenerative disease that affects millions of people. Early diagnosis is important to facilitate prompt interventions to slow down disease progression. However, accurate PD diagnosis can be challenging, especially in the early disease stages. The aim of this work was to develop and evaluate a robust explainable deep learning model for PD classification trained from one of the largest collections of T1-weighted magnetic resonance imaging datasets. MATERIALS AND METHODS: A total of 2,041 T1-weighted MRI datasets from 13 different studies were collected, including 1,024 datasets from PD patients and 1,017 datasets from age- and sex-matched healthy controls (HC). The datasets were skull stripped, resampled to isotropic resolution, bias field corrected, and non-linearly registered to the MNI PD25 atlas. The Jacobian maps derived from the deformation fields together with basic clinical parameters were used to train a state-of-the-art convolutional neural network (CNN) to classify PD and HC subjects. Saliency maps were generated to display the brain regions contributing the most to the classification task as a means of explainable artificial intelligence. RESULTS: The CNN model was trained using an 85%/5%/10% train/validation/test split stratified by diagnosis, sex, and study. The model achieved an accuracy of 79.3%, precision of 80.2%, specificity of 81.3%, sensitivity of 77.7%, and AUC-ROC of 0.87 on the test set while performing similarly on an independent test set. Saliency maps computed for the test set data highlighted frontotemporal regions, the orbital-frontal cortex, and multiple deep gray matter structures as most important. CONCLUSION: The developed CNN model, trained on a large heterogenous database, was able to differentiate PD patients from HC subjects with high accuracy with clinically feasible classification explanations. Future research should aim to investigate the combination of multiple imaging modalities with deep learning and on validating these results in a prospective trial as a clinical decision support system.

Dementia Risk Prediction in a Longitudinal Geriatric Parkinson's Disease Cohort: Evaluation and Application of the Montreal Parkinson Risk of Dementia Scale.

Background: Parkinson's disease (PD) increases risk for dementia and cascading adverse outcomes. The eight-item Montreal Parkinson Risk of Dementia Scale (MoPaRDS) is a rapid, in-office dementia screening tool. We examine predictive validity and other characteristics of the MoPaRDS in a geriatric PD cohort by testing a series of alternative versions and modelling risk score change trajectories. Methods: Participants were 48 initially non-demented PD patients (Mage = 71.6 years, range = 65-84) from a three-year, three-wave prospective Canadian cohort study. A dementia diagnosis at Wave 3 was used to stratify two baseline groups: PD with Incipient Dementia (PDID) and PD with No Dementia (PDND). We aimed to predict dementia three years prior to diagnosis using baseline data for eight indicators that harmonized with the original report, plus education. Results: Three MoPaRDS items (age, orthostatic hypotension, mild cognitive impairment [MCI]) discriminated the groups both independently and as a composite three-item scale (area under the curve [AUC] = 0.88). The eight-item MoPaRDS reliably discriminated PDID from PDND (AUC = 0.81). Education did not improve predictive validity (AUC = 0.77). Performance of the eight-item MoPaRDS varied across sex (AUCfemales = 0.91; AUCmales = 0.74), whereas the three-item configuration did not (AUCfemales = 0.88; AUCmales = 0.91). Risk scores of both configurations increased over time. Conclusions: We report new data on the application of the MoPaRDS as a dementia prediction tool for a geriatric PD cohort. Results support the viability of the full MoPaRDS, and indicate that an empirically determined brief version is a promising complement.

Elicited clinician knowledge did not improve dementia risk prediction in individuals with mild cognitive impairment.

OBJECTIVES: This study aims to develop and validate a Bayesian risk prediction model that combines research cohort data with elicited expert knowledge to predict dementia progression in people with mild cognitive impairment (MCI). STUDY DESIGN AND SETTING: This is a prognostic risk prediction modeling study based on cohort data (Alzheimer's disease neuroimaging initiative [ADNI]; n = 365) of research participants with MCI and elicited expert data. Bayesian Cox models were used to combine expert knowledge and ADNI data to predict dementia progression in people with MCI. Posterior distributions were obtained based on Gibbs sampler and the predictive performance was evaluated using ten-fold cross-validation via c-index, integrated calibration index (ICI), and integrated brier score (IBS). RESULTS: 365 people with MCI were included, mean age was 73 years (SD = 7.5), and 39% developed dementia within 3 years. When expert knowledge was incorporated, the c-index, ICI, and IBS values were 0.74 (95% CI 0.70-0.79), 0.06 (95% CI 0.05-0.08), and 0.17 (95% CI 0.14-0.19), respectively. These were similar to the model without expert knowledge data. CONCLUSION: The addition of expert knowledge did not improve model accuracy in this ADNI sample to predict dementia progression in individuals with MCI.