Separating spatially distributed response to stimulation from background. I. Optical imaging

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Estimating small stimulus-induced responses is challenging when masked by fluctuating background noise.
• Optical imaging of the cortex is a common experimental situation where this problem arises.
• Availability of background measurements in the absence of stimulation is a key factor.

Purpose of the Study:

• To develop and compare methods for accurately estimating stimulus-induced responses in the presence of significant background noise.
• To evaluate the effectiveness of Karhunen-Loève based procedures for signal detection in noisy experimental data.
• To provide insights into assessing the quality of estimated responses.

Main Methods:

• Two related methods based on the Karhunen-Loève procedure were investigated.
• Method 1: Identifying an indicator function most parallel to response data and orthogonal to background data.
• Method 2: Projecting out the subspace spanned by background data from the response data.

Main Results:

• Numerical simulations using optical imaging data demonstrated the superiority of the first method.
• The indicator function approach generally yielded better estimation of the stimulus-induced response.
• The study explored connections between the two methods and quality assessment techniques.

Conclusions:

• The Karhunen-Loève procedure provides effective tools for estimating weak signals in noisy environments.
• The indicator function method is recommended for its superior performance in simulated optical imaging scenarios.
• Further analysis of method connections and quality assessment is valuable for experimental applications.