Poster Y18 - A Rapid, Simple, and Effective Derivative Signal Processing Method for Impedance Flow Cytometry

Friday, October 25, 2024

3:30 PM - 4:30 PM EST

Location: Exhibit Hall E, F & G

Co-Author(s)

BW

Brendan Wurts

Student
College of Charleston, United States
CJ

Charlie Jindrich

Student
College of Charleston, United States

Presenting Author(s)

Leilei Shi, PhD

Assistant Professor
College of Charleston
Charleston, South Carolina, United States

Introduction: The processing of single-cell flow cytometry data is crucial in medical science, offering valuable insights into cellular properties and behaviors. Flow cytometry, particularly impedance flow cytometry (IFC), is a powerful technique that measures the electrical impedance of cells or particles as they pass through a microfluidic channel. This technique is vital for applications such as cell counting, sizing, and phenotyping, which are essential in fields ranging from oncology to immunology. However, the presence of noise poses significant challenges inaccurately detecting and analyzing these signals. Traditional signal processing methods often involve adjusting filter frequency bands, orders, and thresholds, which can be time-consuming and prone to inaccuracies, particularly in high-noise environments. Inspired by the natural processes observed in neurons in the brain, we developed a derivative-based signal processing method that simplifies the detection of events by focusing on the rate of change in the signal. This method leverages the highest slope in the signal to identify key events, providing a rapid and straightforward approach to signal detection in flow cytometry.

Materials and

Methods: The core of the derivative method involves derivative calculation, peak detection, and signal detection and reconstruction. To calculate the derivative, we minimize the "consistency error" between the true gradient and its estimate from neighboring grid points. Using the Taylor series expansion, we derive a linear system of equations. Solving this system provides coefficients that form a linear combination of the function values at the neighboring points, yielding an approximation for the first derivative. The peak will be determined by identifying the point at which the rate of change in the signal, represented by the derivative, reaches its maximum and minimal values. By focusing on the highest slope in the signal, this method effectively highlights significant events, such as the passage of particles or cells through the detection region. Once the peaks corresponding to significant events are identified, the next step involves reconstructing the signal by integrating these detected events. By using the derivative-based method, we ensure that the reconstructed signal retains its key features while minimizing the noise. To evaluate the performance of our algorithm, we conducted experiments using synthetic datasets to simulate the signal characteristics encountered in impedance flow cytometry, as measured from our own experiments. The use of synthetic datasets allowed us to easily manipulate noise levels and maintain a clean baseline for accurate evaluation. All experiments except the synthetic dataset generation were run on MATLAB R2023b.

Results, Conclusions, and Discussions: To evaluate the performance of our algorithm, we did a comparative analysis of three signal processing methods—derivative, adaptive, and notch filter—evaluated under a signal to noise ratio of 15 dB with a sample size of 4500 signal events. The results show that the derivative-based signal processing method demonstrates clear superiority across all performance metrics. Its highest accuracy, precision, and recall underscore its effectiveness in signal detection and reconstruction. The lowest average mean square error (MSE) and signal window MSE highlight its precision in reconstructing signals with minimal error, even in the presence of noise. Additionally, its quick processing time indicates its computational efficiency, making it highly suitable for real-time applications. The results suggest that the derivative method’s focus on the rate of change in the signal allows for more accurate peak detection and noise minimization. By leveraging the highest slope to identify key events, this method provides a rapid, simple and efficient approach for signal detection and reconstruction, making it an invaluable tool in medical and scientific applications requiring high precision and speed.

Acknowledgements (Optional): This work was supported by National Science Fundation ADAPT in SC Program (Award number 2242812).