Pietro Nardelli, PhD: Deep Learning for Pulmonary Vascular Disease

Pietro Nardelli, PhD
Instructor in Radiology
Department of Radiology
Brigham and Women’s Hospital
Harvard Medical School

Abstract

Pulmonary vascular disease (PVD) is a term used to indicate any condition that affects the blood vessels along the route between the heart and the lung. In its most severe from, PVD results in pulmonary hypertension (PH), a heterogeneous condition that occurs most frequently as complicating comorbidity of very common disease in the western world, including COPD, heart failure, and blood clots in the lung. In this seminar, I will introduce newly developed methods based on artificial intelligence (AI) for robust vascular and perfusion assessment to quantify changes of the vascular morphology of the lung. These methods are allowing us to perform end-to-end automatic imaging phenotyping of pulmonary hypertension. Methods for automatic artery-vein classification, measurement of vascular blood and external muscular wall, as well as detection of vascular perfusion from single energy CT images will be presented. Finally, I will show some of the applications of these new approaches to quantify the changes in the parenchymal pulmonary arterial and venous vasculature in patients with PH.

Short Bio

Pietro is an Instructor in Radiology at the Brigham and Women’s Hospital. Originally trained in biomedical engineering, he obtained his PhD in Electrical and Electronic Engineering from the University College Cork, Ireland. He has a specific expertise in medical imaging and Artificial Intelligence applied to medicine. In 2016, he joined the Applied Chest Imaging Laboratory run by Dr. San José Estépar and Dr. George R. Washko to develop innovative computational algorithms based on Deep Learning for the extraction of novel, image-based biomarkers of pulmonary vascular disease. His research includes the assessment of the morphology of peripheral airways and vessels from CT images, the role of pulmonary vasculature on the origin of pulmonary hypertension, and the prediction of outcomes directly from imaging with artificial intelligence techniques. Pietro was recently awarded the NIH BioData Catalyst Fellowship grant with the aim of developing an AI-based tool for pulmonary vasculature morphology assessment from CTs for pulmonary hypertension characterization.