Determining the concentration of red and white blood cells using dielectric properties

Abstract

This project investigates an innovative method to determine the red and white blood cell concentrations in a blood sample using the dielectric properties of blood at microwave frequencies. The dielectric properties of biological tissue characterise the interaction of an electric field applied to the biotissue under test. In recent years, microwaves and their use for imaging and diagnosis of medical conditions are continuously being studied as non-ionising, portable and low-cost alternatives to standard methods used in clinical practice. The concentrations ofred blood cells and white blood cells in a sample of whole blood can vary due to illnesses or diseases. This study is a proof-of-concept, where the dielectric properties of samples containing red blood cells and plasma are investigated. The dielectric properties of samples of different concentration of red blood cells in plasma are measured. The data is then used to create artificial neural networks where a model is trained to relate the measured properties to the known concentrations of red blood cells. The results of the study show that a trained neural network can predict the red blood cell concentrations, of samples that were not used to train the model, with an average error of 1.37 %.