Classification of Prefrontal Cortex Activity Based on Functional Near-Infrared Spectroscopy Data upon Olfactory...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• The sense of smell is crucial for human experience.
• Olfactory imaging techniques are vital for understanding brain activity related to smell.
• Functional near-infrared spectroscopy (fNIRS) offers a non-invasive method to measure brain activity.

Purpose of the Study:

• To assess the efficacy of support vector machines (SVMs) in classifying olfactory stimuli using fNIRS data.
• To compare the performance of different SVM kernel functions and signal types (HRF, PPG) for olfactory detection.
• To determine the optimal signal processing strategy for accurate odor classification.

Main Methods:

• Utilized fNIRS to collect brain activity data from the prefrontal cortex during olfactory stimulation (odor vs. air).
• Extracted hemodynamic response function (HRF) signals from oxyhemoglobin (oxyHb) and deoxyhemoglobin (deoxyHb) variations.
• Employed photoplethysmography (PPG) signals and analyzed three SVM kernel functions (linear, quadratic, cubic).

Main Results:

• The quadratic kernel function with oxyHb single-signal data showed high efficiency.
• A combination of HRF and PPG signals with the cubic kernel function yielded the most efficient multi-signal data.
• SVM analysis of HRF signals demonstrated superior classification of odor and air status from fNIRS data.

Conclusions:

• SVM analysis of fNIRS data provides a robust method for classifying human olfactory stimuli.
• Quadratic and cubic SVM kernel functions enable accurate classification of olfactory stimulation, even with individual participant data.
• This approach advances the potential for objective olfactory assessment and brain-computer interfaces.