A Novel Dielectric Fingerprinting Tool for Histopathology Assessment Leveraging AI and RF: A Feasibility Study Using Gastrointestinal Tissues

Abstract

Digital representation of tissues allows the examination of tissue morphology in new ways enabling patient stratification for effective treatments. Current slide-scanning techniques capture the visible details of the tissue as whole-slide images and digitally record them in the form of spatial and color relationships. Specialized experimental techniques like dielectric spectroscopy can also be used to investigate a tissue's response to an applied electric field. This study used the dielectric spectroscopy method to collect the complex permittivity of healthy and abnormal biopsy tissues excised during Gastroenterology procedures. A single pole Cole-Cole model is fitted to the measurements dataset to extract the Cole-Cole parameters which are used as features in the machine learning binary classification model. The model's performance demonstrates the feasibility of using microwave-based spectroscopy measurements to create a digital dielectric fingerprint for tissues under investigation.