Virtual Cancer Biopsy: An Integrated Machine Learning Framework for Tumor Malignancy Prediction, Surgical Operability Assessment, and Prognosis Estimation

Abstract

Cancer is one of the major causes of morbidity and mortality in the world, and therefore, early cancer detection is a crucial factor when it comes to increasing the survival period. The traditional diagnostic methods are often time-consuming, costly and invasive, thus putting some patients in pre-treatment risks. We are introducing what we call the Virtual Cancer Biopsy (VCB) which is a web-based application that uses machine learning instead of needles that are invasive. VCB uses logistic regression to predict the country of tumors as being malignant or not and uses k nearest neighbor modelling as a prognosis predictor of cancerous tumors. Instead of using simple ML code that is custom-built, the system is built on the Flask library and the front-end is integrated with JavaScript, which is compatible with other browsers. All the workflow is contained in a Streamlit system providing a user-friendly interface that suits clinical needs. In addition to speeding up the diagnostics process, VCB provides doctors with readable visuals and interactive tools, which are changing the way things are categorized.