Autoencoder-based Image Recommendation for Lung Cancer Characterization

Abstract

In this project, we aim to develop an AI system that recommends a set of relative (past) cases to guide the decision-making of the clinician. Objective: The ambition is to develop an AI-based learning model for lung cancer characterization in order to assist in clinical routine. Considering the complexity of the biological phenomenat hat occur during cancer development, relationships between these and visual manifestations captured by CT have been explored in recent years; however, given the lack of robustness of current deep learning methods, these correlations are often found spurious and get lost when facing data collected from shifted distributions: different institutions, demographics or even stages of cancer development.