Age group classification based on optical measurement of brain pulsation using machine learning.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Medical Imaging

Background:

• Non-invasive wearable optical techniques like functional near-infrared spectroscopy (fNIRS) offer potential for monitoring cerebral vascular health in aging populations.
• Real-time hemodynamic monitoring is crucial for understanding age-related changes in brain function.

Purpose of the Study:

• To explore age-related differences in cerebral hemodynamics using single-channel fNIRS and machine learning.
• To evaluate the efficacy of fNIRS-derived pulse features for distinguishing between young and elderly adults.

Main Methods:

• Thirty-six healthy adults (younger and elderly groups) were assessed using single-channel fNIRS during fMRI.
• Brain pulses were extracted at 830 nm, and four feature sets were derived using a pulse decomposition algorithm.
• Machine learning algorithms (SVM, Random Forest) and feature selection methods (mRMR, PCA) were employed.

Main Results:

• The best mean balanced accuracies for the feature sets exceeded 75%, indicating age-related information within the pulse features.
• Learning curves demonstrated stable classification performance as sample size increased.
• Single-channel fNIRS successfully identified age-related differences in cerebral hemodynamics.

Conclusions:

• Single-channel fNIRS, combined with machine learning, shows significant potential for non-invasive analysis of aging-related cerebral vascular conditions.
• This approach could pave the way for wearable devices to monitor brain health in aging individuals.
• The extracted pulse features contain valuable information for differentiating age groups.