Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Early Detection of Cognitive Impairment in Elderly using a Passive FPVS-EEG BCI and Machine Learning -- Extended Version

Tomasz M. Rutkowski, Stanisław Narębski, Mihoko Otake-Matsuura, Tomasz Komendziński

TL;DR

This work tackles the lack of objective functional biomarkers for early cognitive impairment by using a passive FPVS-BCI to probe working memory in elderly adults. An end-to-end CNN (EEGNet) is trained on minimally preprocessed EEG data collected during a short ($2$ minutes) passive FPVS FastBall paradigm, avoiding explicit responses. The study demonstrates that high-gamma activity in the $30-125$ Hz range yields the highest cross-validated accuracy ($78.2 ext{ extpercent}$) for classifying CI levels into normal ($26 extsim30$), mild CI ($18 extsim25$), and moderate CI ($10 extsim17$) among $n=23$ participants, with all bands above chance ($33.3 ext{ extpercent}$). If validated in larger cohorts, this approach could enable rapid, non-invasive screening for cognitive decline in clinical and community settings.

Abstract

Early dementia diagnosis requires biomarkers sensitive to both structural and functional brain changes. While structural neuroimaging biomarkers have progressed significantly, objective functional biomarkers of early cognitive decline remain a critical unmet need. Current cognitive assessments often rely on behavioral responses, making them susceptible to factors like effort, practice effects, and educational background, thereby hindering early and accurate detection. This work introduces a novel approach, leveraging a lightweight convolutional neural network (CNN) to infer cognitive impairment levels directly from electroencephalography (EEG) data. Critically, this method employs a passive fast periodic visual stimulation (FPVS) paradigm, eliminating the need for explicit behavioral responses or task comprehension from the participant. This passive approach provides an objective measure of working memory function, independent of confounding factors inherent in active cognitive tasks, and offers a promising new avenue for early and unbiased detection of cognitive decline.

Early Detection of Cognitive Impairment in Elderly using a Passive FPVS-EEG BCI and Machine Learning -- Extended Version

TL;DR

This work tackles the lack of objective functional biomarkers for early cognitive impairment by using a passive FPVS-BCI to probe working memory in elderly adults. An end-to-end CNN (EEGNet) is trained on minimally preprocessed EEG data collected during a short ( minutes) passive FPVS FastBall paradigm, avoiding explicit responses. The study demonstrates that high-gamma activity in the Hz range yields the highest cross-validated accuracy () for classifying CI levels into normal (), mild CI (), and moderate CI () among participants, with all bands above chance (). If validated in larger cohorts, this approach could enable rapid, non-invasive screening for cognitive decline in clinical and community settings.

Abstract

Early dementia diagnosis requires biomarkers sensitive to both structural and functional brain changes. While structural neuroimaging biomarkers have progressed significantly, objective functional biomarkers of early cognitive decline remain a critical unmet need. Current cognitive assessments often rely on behavioral responses, making them susceptible to factors like effort, practice effects, and educational background, thereby hindering early and accurate detection. This work introduces a novel approach, leveraging a lightweight convolutional neural network (CNN) to infer cognitive impairment levels directly from electroencephalography (EEG) data. Critically, this method employs a passive fast periodic visual stimulation (FPVS) paradigm, eliminating the need for explicit behavioral responses or task comprehension from the participant. This passive approach provides an objective measure of working memory function, independent of confounding factors inherent in active cognitive tasks, and offers a promising new avenue for early and unbiased detection of cognitive decline.

Paper Structure

This paper contains 6 sections, 4 figures.

Table of Contents

  1. INTRODUCTION
  2. METHODS
  3. FSVP Stimulus in "FastBall" Paradigm Setting
  4. EEG Recording and Machine Learning Model
  5. RESULTS
  6. CONCLUSIONS

Figures (4)

  • Figure 1: Distribution of MoCA scores among participants, showing median, quartiles, and range.
  • Figure 2: Distribution of participants across CI groups. These group labels were used for machine learning model training and subsequent prediction.
  • Figure 3: The FPVS image sequence employed in this study, featuring food photographs from the CROCUFID database toet2019crocufid and based on the "FastBall" paradigm fastball
  • Figure 4: Ten-fold cross-validation balanced accuracies across EEG frequency bands, with overlaid median, quartiles, and range. Chance level: $33.3\%.$