Study and Classification of Cell Bio-Impedance Signature for Identification of Malignancy Using Artificial Neural Network
Debjyoti Chowdhury, Madhurima Chattopadhyay
Abstract
This article presents development of a microcapacitive sensor-based bio-impedance classification system for identification of malignant cells in a known volume (400 μL) of biological tissue sample. Malignancy induces various physical changes in an affected cell, among which increased intracellular water and sodium ion content are a couple of prominent ones. In this work, the change in electrical properties of a cell [here white blood cell (WBC)] that is membrane and cytoplasm permittivity due to afore-mentioned reasons is used for distinguishing malignant from normal cells based on their deviation in bio-impedance signature. This change in bio-impedance signature of the WBC is measured by exciting the mentioned microcapacitive sensor by a 2 V p-p signal. The developed system consists of an on-board impedance analyzer to carry out bio-impedance spectroscopy over a frequency range of 1-100 kHz. The bio-impedance data are then obtained from the sensor for different concentrations of cultured human malignant and normal white blood cells. The difference in impedance values between concentrations is then used as a discerning factor for identification the degree of malignancy, for which an artificial neural network (ANN) is employed. A detailed study considering detection efficiency for this ANN revealed classification accuracy up to 95.0%, upon which this ANN model is transferred to a portable single board computer (SBC) along with other associated circuitry that could carry out the same classification at a nominal power consumption of 516.45 mW.