Study of Tissue Heterogeneity and Classification using AI Techniques

Abstract

The idea behind our project is to design an algorithm that utilizes artificial intelligence to detect tissue heterogeneity in patients without the need to carry out an invasive biopsy. We aim to make the cancer prognosis process based solely on the study of the scanned medical images such as MRI or CT. The algorithm will be written in Python and will utilize large data sets of radiomics biomarkers extracted from medical images of different modalities through a software called LIFEx. Radiomics biomarkers are huge amounts of quantitative features extracted from medical images that characterize tumor phenotypes like texture and shape. The objective that we want our algorithm to achieve is to classify the cancer stage. In this project, we will focus on cervix cancer as it is of great interest to our collaborators who are providing us with private data. Another benefit to our algorithm is that it will offer a noninvasive method for cancer diagnosis and will hence bypass biopsies as they are associated with many additional health risks and costs. This project will contribute to changing the way doctors diagnose cancer and make it a more efficient process using our robust, reliable detection of tissue heterogeneity.