Nathan Painchaud completed his PhD through a joint supervision program between INSA Lyon and the University of Sherbrooke (UdS, Canada), under the guidance of Professor Olivier Bernard (INSA Lyon), Professor Nicolas Duchateau (Université Lyon 1), and Professor Pierre-Marc Jodoin from the Department of Computer Science at the Faculty of Science, UdS.
He was awarded the Best Doctoral Thesis Prize in the Medicine and Health Sciences category for his dissertation entitled Deep Manifold Learning for Improved Characterization of Hypertension in Echocardiographic Imaging.
Hypertension affects 1.5 billion people worldwide. Assessing this complex cardiovascular condition is often challenging due to the variety of symptoms and the lack of fast and accurate analytical tools. As artificial intelligence continues to advance rapidly, several models have been developed to support physicians in interpreting cardiac ultrasound images. However, these techniques often produce inconsistent results in many cases. This is precisely where Nathan Painchaud’s work comes into play. His thesis addresses the limitations of neural networks by introducing constraints that reflect prior knowledge of what should be observed in the images — a first in the field. It also enabled unprecedented precision in the study of hypertension symptoms, revealing potential early warning signs.