Text

theses

Bioelectrical signals such as electrocardiogram (ECG) and flow cytometry signals are often affected by both low-frequency and high frequency perturbations during data acquisition. It is necessary to remove these interferences from the acquired data so that the pertinent information needed from these signals can be obtained. Methods for detrending and denoising in ECG are well established, but in the context of impedance cytometry data, there lacks a body of literature available that provides guidance as to a robust methodology for processing the data. For the first time in the context of impedance cytometry, to the best of our knowledge, in this work we systematically studied and compared the performance of different algorithms for detrending and denoising, and developed a procedure to analyze impedance cytometry data with minimal error. This work can serve as guidance to select the optimal algorithm based on the following parameters: Standard Deviation, Correlation Coefficient, Power Spectral Density (PSD), Root Mean Square Error (RMSE) and Root Mean Square Difference (RMSD). The approaches discussed conventional filtering techniques such as: Butterworth Filtering, Chebyshev Filtering, as well as the thresholding techniques in the Discrete Wavelet Transform (DWT). The procedure of selecting the mother wavelet basis functions and the four different thresholding methods are discussed and compared. The performance of the optimized algorithm is compared with the pClamp10 commercially available cytometry analysis software resulting in 19.2% improvement in amplitude preservation, 18.4% improvement in area preservation and more than 50% improvement in the peak-search error rate.

ETD_6827

M.S.

Includes bibliographical references

by Xinnan Cao

doi:10.7282/T3CV4KQ3

ETD graduate

The author owns the copyright to this work.