Threshold optimization in separating cortical and extracerebral hemodynamics using principal component analysis.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Distinguishing cortical activity from extracerebral hemodynamics is crucial for accurate neuroimaging.
• Principal Component Analysis (PCA) is a statistical method used to reduce extracerebral hemodynamic influences.
• Current PCA methods often use fixed thresholds, failing to address inter-individual variability and experimental design differences, leading to inaccurate signal correction.

Purpose of the Study:

• To optimize the thresholding method within PCA for differentiating cortical and extracerebral hemodynamic signals.
• To enhance the accuracy of PCA-based analyses for cortical activity detection.
• To improve upon existing methods for hemodynamic signal correction in neuroimaging.

Main Methods:

• Four analysis approaches were compared: no correction (NC), short separation regression (SSR), PCA with optimized threshold (PCA_opt), and PCA with maximum threshold (PCA_max).
• Data was collected from older participants performing a verbal n-back task.
• Bayesian t-tests were used to evaluate the equivalence between SSR and PCA_opt.

Main Results:

• No correction (NC) showed the strongest cortical activation, while PCA_max showed the weakest.
• SSR and PCA_opt yielded intermediate results.
• Bayesian t-tests indicated strong evidence (BF₀₁ > 3.0) for equivalence between SSR and PCA_opt across most channels, with no channels showing strong evidence for differences (BF₁₀ > 3.0).

Conclusions:

• Optimizing the PCA threshold is a practical and effective strategy for separating cortical and extracerebral hemodynamics.
• This optimized PCA approach provides appropriate signal correction, even when short-separation channels are unavailable.
• The findings support the utility of PCA_opt as a reliable method for analyzing cortical activity.