Innovative Machine Learning Techniques for Biomedical Imaging

Abstract

Machine Learning (ML) is a powerful paradigm to solve several inverse problems arising in biomedical imaging with very high computational efficiency. As a matter of fact, learning-by-examples (LBE) strategies can be successfully exploited to predict the status of the domain under investigation (DoI) starting from measured data with almost real-time performance. Some recent advances of ML as applied to brain stroke detection, classification, and localization, as well as to human chest monitoring are presented. An illustrative example concerned with a novel LBE strategy for the real-time prediction of the lungs dimensions from electrical impedance tomography (EIT) measurements is given, as well.