Multiclass characterization of frontotemporal dementia variants via multimodal brain network computational inference

A structural MRI marker predicts individual differences in impulsivity and classifies patients with behavioral-variant frontotemporal dementia from matched controls

Impulsivity and higher preference for sooner over later rewards (i.e., delay discounting) are transdiagnostic markers of many psychiatric and neurodegenerative disorders. Yet, their neurobiological basis is still debated. Here, we aimed at 1) identifying a structural MRI signature of delay discounting in healthy adults, and 2) validating it in patients with behavioral variant frontotemporal dementia (bvFTD)-a neurodegenerative disease characterized by high impulsivity. We used a machine-learning algorithm to predict individual differences in delay discounting rates based on whole-brain grey matter density maps in healthy male adults (Study 1, N=117). This resulted in a cross-validated prediction-outcome correlation of r=0.35 (p=0.0028). We tested the validity of this brain signature in an independent sample of 166 healthy adults (Study 2) and its clinical relevance in 24 bvFTD patients and 18 matched controls (Study 3). In Study 2, responses of the brain signature did not correlate significantly with discounting rates, but in both Studies 1 and 2, they correlated with psychometric measures of trait urgency-a measure of impulsivity. In Study 3, brain-based predictions correlated with discounting rates, separated bvFTD patients from controls with 81% accuracy, and were associated with the severity of disinhibition among patients. Our results suggest a new structural brain pattern-the Structural Impulsivity Signature (SIS)-which predicts individual differences in impulsivity from whole-brain structure, albeit with small-to-moderate effect sizes. It provides a new brain target that can be tested in future studies to assess its diagnostic value in bvFTD and other neurodegenerative and psychiatric conditions characterized by high impulsivity.

Educational disparities in brain health and dementia across Latin America and the United States

BACKGROUND

Education influences brain health and dementia. However, its impact across regions, specifically Latin America (LA) and the United States (US), is unknown.

METHODS

A total of 1412 participants comprising controls, patients with Alzheimer's disease (AD), and frontotemporal lobar degeneration (FTLD) from LA and the US were included. We studied the association of education with brain volume and functional connectivity while controlling for imaging quality and variability, age, sex, total intracranial volume (TIV), and recording type.

RESULTS

Education influenced brain measures, explaining 24%-98% of the geographical differences. The educational disparities between LA and the US were associated with gray matter volume and connectivity variations, especially in LA and AD patients. Education emerged as a critical factor in classifying aging and dementia across regions.

DISCUSSION

The results underscore the impact of education on brain structure and function in LA, highlighting the importance of incorporating educational factors into diagnosing, care, and prevention, and emphasizing the need for global diversity in research.

HIGHLIGHTS

Lower education was linked to reduced brain volume and connectivity in healthy controls (HCs), Alzheimer's disease (AD), and frontotemporal lobar degeneration (FTLD). Latin American cohorts have lower educational levels compared to the those in the United States. Educational disparities majorly drive brain health differences between regions. Educational differences were significant in both conditions, but more in AD than FTLD. Education stands as a critical factor in classifying aging and dementia across regions.

A synergetic turn in cognitive neuroscience of brain diseases

Despite significant improvements in our understanding of brain diseases, many barriers remain. Cognitive neuroscience faces four major challenges: complex structure-function associations; disease phenotype heterogeneity; the lack of transdiagnostic models; and oversimplified cognitive approaches restricted to the laboratory. Here, we propose a synergetics framework that can help to perform the necessary dimensionality reduction of complex interactions between the brain, body, and environment. The key solutions include low-dimensional spatiotemporal hierarchies for brain-structure associations, whole-brain modeling to handle phenotype diversity, model integration of shared transdiagnostic pathophysiological pathways, and naturalistic frameworks balancing experimental control and ecological validity. Creating whole-brain models with reduced manifolds combined with ecological measures can improve our understanding of brain disease and help identify novel interventions. Synergetics provides an integrated framework for future progress in clinical and cognitive neuroscience, pushing the boundaries of brain health and disease toward more mature, naturalistic approaches.

Differential diagnosis of frontotemporal dementia subtypes with explainable deep learning on structural MRI

Background

Frontotemporal dementia (FTD) represents a collection of neurobehavioral and neurocognitive syndromes that are associated with a significant degree of clinical, pathological, and genetic heterogeneity. Such heterogeneity hinders the identification of effective biomarkers, preventing effective targeted recruitment of participants in clinical trials for developing potential interventions and treatments. In the present study, we aim to automatically differentiate patients with three clinical phenotypes of FTD, behavioral-variant FTD (bvFTD), semantic variant PPA (svPPA), and nonfluent variant PPA (nfvPPA), based on their structural MRI by training a deep neural network (DNN).

Methods

Data from 277 FTD patients (173 bvFTD, 63 nfvPPA, and 41 svPPA) recruited from two multi-site neuroimaging datasets: the Frontotemporal Lobar Degeneration Neuroimaging Initiative and the ARTFL-LEFFTDS Longitudinal Frontotemporal Lobar Degeneration databases. Raw T1-weighted MRI data were preprocessed and parcellated into patch-based ROIs, with cortical thickness and volume features extracted and harmonized to control the confounding effects of sex, age, total intracranial volume, cohort, and scanner difference. A multi-type parallel feature embedding framework was trained to classify three FTD subtypes with a weighted cross-entropy loss function used to account for unbalanced sample sizes. Feature visualization was achieved through post-hoc analysis using an integrated gradient approach.

Results

The proposed differential diagnosis framework achieved a mean balanced accuracy of 0.80 for bvFTD, 0.82 for nfvPPA, 0.89 for svPPA, and an overall balanced accuracy of 0.84. Feature importance maps showed more localized differential patterns among different FTD subtypes compared to groupwise statistical mapping.

Conclusion

In this study, we demonstrated the efficiency and effectiveness of using explainable deep-learning-based parallel feature embedding and visualization framework on MRI-derived multi-type structural patterns to differentiate three clinically defined subphenotypes of FTD: bvFTD, nfvPPA, and svPPA, which could help with the identification of at-risk populations for early and precise diagnosis for intervention planning.

Visual deep learning of unprocessed neuroimaging characterises dementia subtypes and generalises across non-stereotypic samples

BACKGROUND

Dementia's diagnostic protocols are mostly based on standardised neuroimaging data collected in the Global North from homogeneous samples. In other non-stereotypical samples (participants with diverse admixture, genetics, demographics, MRI signals, or cultural origins), classifications of disease are difficult due to demographic and region-specific sample heterogeneities, lower quality scanners, and non-harmonised pipelines.

METHODS

We implemented a fully automatic computer-vision classifier using deep learning neural networks. A DenseNet was applied on raw (unpreprocessed) data from 3000 participants (behavioural variant frontotemporal dementia-bvFTD, Alzheimer's disease-AD, and healthy controls; both male and female as self-reported by participants). We tested our results in demographically matched and unmatched samples to discard possible biases and performed multiple out-of-sample validations.

FINDINGS

Robust classification results across all groups were achieved from standardised 3T neuroimaging data from the Global North, which also generalised to standardised 3T neuroimaging data from Latin America. Moreover, DenseNet also generalised to non-standardised, routine 1.5T clinical images from Latin America. These generalisations were robust in samples with heterogenous MRI recordings and were not confounded by demographics (i.e., were robust in both matched and unmatched samples, and when incorporating demographic variables in a multifeatured model). Model interpretability analysis using occlusion sensitivity evidenced core pathophysiological regions for each disease (mainly the hippocampus in AD, and the insula in bvFTD) demonstrating biological specificity and plausibility.

INTERPRETATION

The generalisable approach outlined here could be used in the future to aid clinician decision-making in diverse samples.

FUNDING

The specific funding of this article is provided in the acknowledgements section.

Model-based whole-brain perturbational landscape of neurodegenerative diseases

The treatment of neurodegenerative diseases is hindered by lack of interventions capable of steering multimodal whole-brain dynamics towards patterns indicative of preserved brain health. To address this problem, we combined deep learning with a model capable of reproducing whole-brain functional connectivity in patients diagnosed with Alzheimer's disease (AD) and behavioral variant frontotemporal dementia (bvFTD). These models included disease-specific atrophy maps as priors to modulate local parameters, revealing increased stability of hippocampal and insular dynamics as signatures of brain atrophy in AD and bvFTD, respectively. Using variational autoencoders, we visualized different pathologies and their severity as the evolution of trajectories in a low-dimensional latent space. Finally, we perturbed the model to reveal key AD- and bvFTD-specific regions to induce transitions from pathological to healthy brain states. Overall, we obtained novel insights on disease progression and control by means of external stimulation, while identifying dynamical mechanisms that underlie functional alterations in neurodegeneration.