Marco Zanet is an Early Stage Researcher based at TissueGnostics GmbG. His research project, as part of Helical network, is focused on developing machine learning (ML) and deep learning (DL) algorithms as support to other clinical tools that assist clinicians in diagnosing active vasculitis and predicting outcome.
Main goal of this research project is to choose/develop best possible ML and/or DL techniques which will help in identifying the source of extracellular vesicles and establish them as diagnostic and predictive tool in ANCA-associated vasculitis. Adopting this alternative strategy, both publicly available and in-house developed medical datasets will be used to define tissue morphological changes (descriptors) that can be used as predictors of outcome in renal ANCA vasculitis.
Based on existing descriptors and algorithms, this project aims to define morphological changes in renal biopsies from patients with ANCA vasculitis that are suited to automated morphometric analysis and subsequent validation using existing clinical outcome data. Datasets, both medical and non-medical, will be validated and augmented to help in improving ML and DL algorithms.
Main data used in this project are images of the kidney tissue. First objective is to develop an algorithm which will detect and segment glomeruli on WSI’s (Whole Slide Images). After glomeruli are detected machine learning techniques will be used to detect features in development of ANCA-associated vasculitis.