Enhanced fNIRS-Based MCI Detection via Resting-State and Task-State Integration With Spatial-Temporal Feature...

Area of Science:

• Neuroscience
• Medical Technology
• Machine Learning

Background:

• Mild cognitive impairment (MCI) is a precursor to dementia, necessitating early detection for intervention.
• Functional near-infrared spectroscopy (fNIRS) offers a non-invasive method for brain activity monitoring.
• Current machine learning methods for MCI detection using fNIRS suffer from underutilized data and high dimensionality.

Purpose of the Study:

• To develop a spatio-temporal feature engineering framework for improved MCI classification using fNIRS.
• To address the limitations of underutilizing complementary resting-state and task-state fNIRS data.
• To reduce feature dimensionality for more robust and generalizable MCI detection models.

Main Methods:

• Independent component analysis (ICA) was used to derive spatial filters from resting-state fNIRS signals.
• A universal population-level filter set was created by clustering subject-specific filters to isolate spatial features from task-state signals.
• Temporal feature selection identified discriminative task-evoked time points, reducing dimensionality for MCI detection.

Main Results:

• The framework achieved 90.91% accuracy in classifying cognitively normal individuals versus those with MCI.
• A significant feature dimensionality reduction of 91.07% was attained.
• Analysis revealed universal spatial filters linked to MCI biomarkers and critical temporal decision points during cognitive tasks.

Conclusions:

• The proposed framework effectively integrates resting-state and task-state fNIRS data for enhanced MCI detection.
• Dimensionality reduction leads to higher accuracy and interpretability in fNIRS-based MCI classification.
• This advancement holds potential for generalizable MCI detection and efficient clinical data expansion.