Nicolas Delinte

Although cochlear implantation (CI) has been used for decades to restore hearing in sensorineural deafness, accurately predicting individual outcomes remains challenging—particularly in adults with postlingual deafness, a population that has been relatively understudied. In our previous work, we showed that speech comprehension after 6 months of CI use is better predicted when preoperative grey matter volume and cortical thickness in regions supporting language and cognition are considered alongside preoperative residual hearing and audiovisual abilities. In the present study, we investigated whether preoperative diffusion-weighted imaging (DWI) metrics in subcortical white matter (WM) could account for individual variability in cochlear implant adaptation and enhance predictive accuracy. Thereto, we collected preoperative DWI and speech performance under auditory-only and audiovisual conditions in 65 adults with postlingual bilateral deafness. Improvement in speech comprehension with CI was evaluated at 6 and 12 months post-implantation. A whole-brain analysis revealed significant positive correlations between preoperative fractional anisotropy (FA) within a cluster in the left arcuate fasciculus and speech perception at 6 months in both auditory-only and audiovisual conditions. Incorporating FA from the arcuate fasciculus cluster into the multiple linear regression model developed in our previous study significantly improved prediction accuracy for CI outcomes at 6 months. We also identified FA clusters in the right inferior fronto-occipital fasciculus and the corpus callosum that showed positive relationships with audiovisual comprehension and mid-term cochlear implant performance. In contrast, mean FA values averaged across entire tracts did not show similarly strong associations with speech perception, as shown after tract reconstruction. Notably, sex modulated the relationship between age and FA values across multiple white matter clusters. These findings underscore the role of localized WM microstructural properties in shaping speech adaptation after CI. Identifying WM-based predictors of CI performance may aid clinicians in forecasting individual outcomes and tailoring rehabilitation strategies, particularly for patients at risk of delayed or limited benefit. Further research is needed to clarify the origins of regional FA variability.

Objective: This study aims to support early diagnosis of Alzheimer's disease and detection of amyloid accumulation by leveraging the microstructural information available in multi-shell diffusion MRI (dMRI) data, using a vision transformer-based deep learning framework. Methods: We present a classification pipeline that employs the Swin Transformer, a hierarchical vision transformer model, on multi-shell dMRI data for the classification of Alzheimer's disease and amyloid presence. Key metrics from DTI and NODDI were extracted and projected onto 2D planes to enable transfer learning with ImageNet-pretrained models. To efficiently adapt the transformer to limited labeled neuroimaging data, we integrated Low-Rank Adaptation. We assessed the framework on diagnostic group prediction (cognitively normal, mild cognitive impairment, Alzheimer's disease dementia) and amyloid status classification. Results: The framework achieved competitive classification results within the scope of multi-shell dMRI-based features, with the best balanced accuracy of 95.2% for distinguishing cognitively normal individuals from those with Alzheimer's disease dementia using NODDI metrics. For amyloid detection, it reached 77.2% balanced accuracy in distinguishing amyloid-positive mild cognitive impairment/Alzheimer's disease dementia subjects from amyloid-negative cognitively normal subjects, and 67.9% for identifying amyloidpositive individuals among cognitively normal subjects. Grad-CAM-based explainability analysis identified clinically relevant brain regions, including the parahippocampal gyrus and hippocampus, as key contributors to model predictions. Conclusion/Significance: This study demonstrates the promise of diffusion MRI and transformer-based architectures for early detection of Alzheimer's disease and amyloid pathology, supporting biomarker-driven diagnostics in data-limited biomedical settings.

Alcohol use disorder (AUD) is a complex condition including affective, cognitive and motivational dimensions. Although AUD is known to induce diffuse brain damage, including grey matter shrinkage and ventricular enlargement, the microstructural changes it induces in white matter remain incompletely understood. This study leverages multi-shell diffusion MRI and multi-fixel models to (i) undertake whole-brain and tract-specific analyses to investigate the microstructure of white matter (WM) tracts affected by AUD, (ii) evaluate whether these differences persist in early abstinence, and (iii) correlate these results with clinical measures evaluated by validated psychological questionnaires. We recruited a final cohort of 37 AUD patients, admitted for alcohol withdrawal and selected for their ongoing alcohol consumption at the time of admission, and a demographically matched control group of 19 healthy subjects. Both groups underwent MRI scans at baseline and 18 days later, with assessments of depression, obsession-compulsion, and anxiety conducted in both sessions for the AUD patients and once for the control group. The imaging results confirmed the presence in AUD participants of clusters microstructural alterations in the fornix, corpus callosum, cingulum, uncinate fasciculus and anterior thalamic radiations. These white matter tracts presented global and localized microstructural changes in axial diffusivity and fractional anisotropy, which are linked to axonal damage and inflammation. There was no significant improvement in the diffusion metrics after almost three weeks of abstinence, although clinical measures did improve significantly. Depression scores were significantly elevated in the patients at admission and decreased with time. Depression scores before withdrawal showed correlations with microstructural metrics across the right anterior thalamic radiations, the isthmus of the corpus callosum, and the right uncinate fasciculus. Lower fractional anisotropy and higher radial diffusivity were predictive of higher depression scores. Overall, these findings highlight the long-term vulnerability of WM tracts affected by AUD and the link between tract microstructure, brain function and behavior.

Apathy is marked by disruptions in effort-based decision-making for rewards and alterations in fronto–basal ganglia (BG) circuits. While changes in primary motor cortex (M1) activity during effort and reward valuation have been reported, prior work on apathy and effort-based decision-making has overlooked the connections between fronto–BG structures and M1. Here, we addressed this gap by investigating structural and effective connectivity within fronto-M1, fronto-BG-M1, and intra-M1 circuits in 45 healthy participants using tractography and paired-pulse transcranial magnetic stimulation. These measures were related to apathy scores and computational parameters of effort and reward valuation modelled from decision-making behavior. Apathy scores were associated with both structural and effective connectivity in orbitofrontal cortex-originating circuits. Additionally, effort and reward valuation were preferentially linked to supplementary motor area-related circuits and intra-M1 GABAergic circuits, respectively. These findings reveal dissociable circuit-level contributions to different dimensions of motivated behavior and identify potential neuromodulation targets for treating apathy.

Objective: This study aims to support early diagnosis of Alzheimer's disease and detection of amyloid accumulation by leveraging the microstructural information available in multi-shell diffusion MRI (dMRI) data, using a vision transformer-based deep learning framework. Methods: We present a classification pipeline that employs the Swin Transformer, a hierarchical vision transformer model, on multi-shell dMRI data for the classification of Alzheimer's disease and amyloid presence. Key metrics from DTI and NODDI were extracted and projected onto 2D planes to enable transfer learning with ImageNet-pretrained models. To efficiently adapt the transformer to limited labeled neuroimaging data, we integrated Low-Rank Adaptation. We assessed the framework on diagnostic group prediction (cognitively normal, mild cognitive impairment, Alzheimer's disease dementia) and amyloid status classification. Results: The framework achieved competitive classification results within the scope of multi-shell dMRI-based features, with the best balanced accuracy of 95.2% for distinguishing cognitively normal individuals from those with Alzheimer's disease dementia using NODDI metrics. For amyloid detection, it reached 77.2% balanced accuracy in distinguishing amyloid-positive mild cognitive impairment/Alzheimer's disease dementia subjects from amyloid-negative cognitively normal subjects, and 67.9% for identifying amyloid-positive individuals among cognitively normal subjects. Grad-CAM-based explainability analysis identified clinically relevant brain regions, including the parahippocampal gyrus and hippocampus, as key contributors to model predictions. Conclusion: This study demonstrates the promise of diffusion MRI and transformer-based architectures for early detection of Alzheimer's disease and amyloid pathology, supporting biomarker-driven diagnostics in data-limited biomedical settings.