Pulmonary Embolism Risk Stratification from CTPA and Medical Records: Vascular Graphs Are Not All You Need
We benchmark different approaches to combine medical records and cardiac biomarkers with rich pulmonary vascular information.
Professional Summary
I am currently a postdoctoral researcher at the CREATIS laboratory (INSA Lyon, France), where I work on multimodal learning on graphs and clinical data for pulmonary embolism risk stratification models.
Prior to my post-doc, I obtained my joint Ph.D. between the Université de Sherbrooke (Canada) and Institut National des Sciences Appliquées de Lyon (France), working on deep representation learning for cardiac medical image analysis.
Education
Assistant Professor Qualification
Conseil National des Universités (CNU)
Ph.D. Computer Science (Joint Supervision)
Université de Sherbrooke (Canada) / INSA Lyon (France)
M.Sc. Computer Science
Université de Sherbrooke (Canada)
B.Sc. Computer Science
Université de Sherbrooke (Canada)
Interests
Representation Learning
Graph Learning
Population Representation
Quantitative Medical Image Analysis
Multimodal Medical Data Fusion
📚 My Research
I’m a junior researcher working on applications of AI to healthcare data. I am mostly interested in how to combine the complex multimodal data, especially the rich imaging data, to improve disease characterization.
From my background in computer science, I’ve also kept a strong interest in the technical side of things. I try my best to be active on GitHub and to contribute to the open-source projects that I use, either professionally or personally.
Don’t hesitate to reach out to collaborate 😃
Featured Publications
Fusing Echocardiography Images and Medical Records for Continuous Patient Stratification
Drawing on novel Transformer models applied to tabular data, we propose a method that considers all descriptors extracted from medical records and echocardiograms to learn the …
Echocardiography Segmentation With Enforced Temporal Consistency
In this paper, we propose a framework to learn the 2D+time apical long-axis cardiac shape such that the segmented sequences can benefit from temporal and anatomical consistency …
Cardiac Segmentation With Strong Anatomical Guarantees
In this paper, we present a framework for producing cardiac image segmentation maps that are guaranteed to respect pre-defined anatomical criteria, while remaining within the …
Featured Talks
Attention and Transformers
Lecture on attention and Transformers given at the Deep Learning for Medical Imaging (DLMI) 2025 spring school.
Deep manifold learning for improved high blood pressure characterization using echocardiography
PhD Thesis Defense of Joint Thesis between Université de Sherbrooke and INSA Lyon.