Early detection of Alzheimer’s disease from aMCI patients: validation of NAP4A technological platform via biomedical signals acquisition and processing

Alzheimer’s disease (AD) is an idiopathic, primary, degenerative form of neurological disorder that accounts from 60 to 80 percent of worldwide dementia cases. This pathology usually manifests as slight loss of memory and other neurocognitive skills, becoming gradually more acute. Anamnestic mild cognitive impairment (aMCI) refers to a condition of degeneration of cognitive functions: studies reported that in 65% of cases, it represents a transitional state from normal ageing to Alzheimer’s disease. Its symptoms are not yet clearly identified, and the degree of deterioration is difficult to determine. Nowadays dementia diagnosis requires time consuming and qualitative screening tests and mostly expensive medical images exams like brain magnetic resonance. These factors, along with the large number of affected individuals, make it evident and crucial to identify its onset in an early stage. This thesis aims to validate the NAP4A biosensors platform, a tool for supporting clinicians in early detection of AD in aMCI patients, developed at the San Raffaele Hospital (MI) by the Center for Advanced Technology in Health and Wellbeing in collaboration with the Neuroimaging Research Unit. The experimental protocol NAP4A2020, approved by OSR Ethical Committee, requires sensors connected with the platform for the synchronized acquisition of electroencephalographic, photoplethysmography, galvanic skin response, eye-tracker, and webcam signals. The low-risk longitudinal study involves both control and patients’ group, enrolled according to the inclusion and exclusion criteria reported in the protocol cited above. The whole experimental procedure is performed with the heterogeneous team of researchers, and it includes two main phases, both involving neurophysiological parameters acquisition. Initially, CANTAB digital battery of tests is proposed to the participant, assessing cognitive areas performances in different tasks execution. Next, two series of visual stimuli consisting of personal and so called ‘NEFFIE’ images are administered, chosen for their peculiarity of stimulating attention and emotions. During the first carousel, the subject expresses the rate of arousal felt for each image; with the last one, memory skill is evaluated, asking if each precise picture was previously seen. Features extraction and selection processes from signals recorded are then implemented in MATLAB ® environment. Finally, an explorative and preliminary analysis of control group numerical parameters has been performed, verifying their adherence with literature indications, and highlighting variables of interest for next case-control comparison. Improvements and critical issues have been collected till today and they will help in optimizing NAP4A platform for future applications.