Explainable AI Points to White Matter Hyperintensities for Alzheimer's Disease Identification: a Preliminary Study.

Deep Learning approaches are powerful tools in a great variety of classification tasks. However, they are limitedly accepted or trusted in clinical frameworks due to their typical "black box" outline: their architecture is well-known, but processes employed in classification are often inaccessible to humans. With this work, we explored the problem of "Explainable AI" (XAI) in Alzheimer's disease (AD) classification tasks. Data from a neuroimaging cohort (n = 251 from OASIS-3) of early-stage AD dementia and healthy controls (HC) were analysed. The MR scans were initially fed to a pre-trained DL model, which achieved good performance on the test set (AUC: 0.82, TPR: 0.78, TNR: 0.81). Results were then investigated by means of an XAI approach (Occlusion Sensitivity method) that provided measures of relevance (RV) as outcome. We compared RV values obtained within healthy tissues with those underlying white matter hyperintensity (WMH) lesions. The analysis was conducted on 4 different groups of data, obtained by stratifying correct and misclassified images according to the health condition of participants (AD/HC). Results highlighted that the DL model found favourable leveraging lesioned brain areas for AD identification. A statistically significant difference ( ) between WMH and healthy tissue contributions was indeed observed for AD recognition, differently from the HC case ( p=0.27). Clinical Relevance - This study, though preliminary, suggested that DL models might be trained to use known clinical information and reinforced the role of WMHs as neuroimaging biomarker for AD dementia. The outlined findings have a significant clinical relevance as they prepare the ground for a progressive increase in the level of trust laid in DL approaches.

Multi-Scale Evaluation of Sleep Quality Based on Motion Signal from Unobtrusive Device.

Sleep disorders are a growing threat nowadays as they are linked to neurological, cardiovascular and metabolic diseases. The gold standard methodology for sleep study is polysomnography (PSG), an intrusive and onerous technique that can disrupt normal routines. In this perspective, m-Health technologies offer an unobtrusive and rapid solution for home monitoring. We developed a multi-scale method based on motion signal extracted from an unobtrusive device to evaluate sleep behavior. Data used in this study were collected during two different acquisition campaigns by using a Pressure Bed Sensor (PBS). The first one was carried out with 22 subjects for sleep problems, and the second one comprises 11 healthy shift workers. All underwent full PSG and PBS recordings. The algorithm consists of extracting sleep quality and fragmentation indexes correlating to clinical metrics. In particular, the method classifies sleep windows of 1-s of the motion signal into: displacement (DI), quiet sleep (QS), disrupted sleep (DS) and absence from the bed (ABS). QS proved to be positively correlated (0.72±0.014) to Sleep Efficiency (SE) and DS/DI positively correlated (0.85±0.007) to the Apnea-Hypopnea Index (AHI). The work proved to be potentially helpful in the early investigation of sleep in the home environment. The minimized intrusiveness of the device together with a low complexity and good performance might provide valuable indications for the home monitoring of sleep disorders and for subjects' awareness.

Development and Testing of SPIDER-NET: An Interactive Tool for Brain Connectogram Visualization, Sub-Network Exploration and Graph Metrics Quantification.

Brain connectomics consists in the modeling of human brain as networks, mathematically represented as numerical connectivity matrices. However, this representation may result in difficult interpretation of the data. To overcome this limitation, graphical representation by connectograms is currently used via open-source tools, which, however, lack user-friendly interfaces and options to explore specific sub-networks. In this context, we developed SPIDER-NET (Software Package Ideal for Deriving Enhanced Representations of brain NETworks), an easy-to-use, flexible, and interactive tool for connectograms generation and sub-network exploration. This study aims to present SPIDER-NET and to test its potential impact on pilot cases. As a working example, structural connectivity (SC) was investigated with SPIDER-NET in a group of 17 healthy controls (HCs) and in two subjects with stroke injury (Case 1 and Case 2, both with a focal lesion affecting part of the right frontal lobe, insular cortex and subcortical structures). 165 parcels were determined from individual structural magnetic resonance imaging data by using the Destrieux atlas, and defined as nodes. SC matrices were derived with Diffusion Tensor Imaging tractography. SC matrices of HCs were averaged to obtain a single group matrix. SC matrices were then used as input for SPIDER-NET. First, SPIDER-NET was used to derive the connectogram of the right hemisphere of Case 1 and Case 2. Then, a sub-network of interest (i.e., including gray matter regions affected by the stroke lesions) was interactively selected and the associated connectograms were derived for Case 1, Case 2 and HCs. Finally, graph-based metrics were derived for whole-brain SC matrices of Case 1, Case 2 and HCs. The software resulted effective in representing the expected (dis) connectivity pattern in the hemisphere affected by the stroke lesion in Cases 1 and 2. Furthermore, SPIDER-NET allowed to test an a priori hypothesis by interactively extracting a sub-network of interest: Case 1 showed a sub-network connectivity pattern different from Case 2, reflecting the different clinical severity. Global and local graph-based metrics derived with SPIDER-NET were different between cases with stroke injury and HCs. The tool proved to be accessible, intuitive, and interactive in brain connectivity investigation and provided both qualitative and quantitative evidence.