Parallel Acceleration on Removal of Optical Mapping Baseline Wandering.

Abstract

Optical mapping measurements on hearts stained with fluorescent dyes is imagining method widely accepted and recognized as a tool to study complex spatial-temporal dynamics of cardiac electro-physiology. One shortcoming of the method is baseline wandering in obtained fluorescence signals as signals relevant to transmembrane potential (V_m) change and free intracellular calcium concentration ([Ca]_i ⁺²), the two most used dyes, are calculated as a relative signal change in respect to the fluorescence baseline. These changes are small fractional changes often smaller than 10 %. Baseline fluorescence drifts due to dye photo-bleaching, heart contraction/movement artifacts, and stability of the excitation light source over time. Depending on experimental instrumentation, recording duration, signal to noise levels and study aims of the optical imagining, many research groups adopted their own techniques tailored to a specific experimental data. Here we present a technique based on finite impulse response (FIR) filters with paralleled acceleration implemented on GPUs and multi-core CPU, in MATLAB.