Medical imaging and deep learning
Improving how medical images are acquired and processed : deep learning methods for the benefit of patients and clinicians.
Research areas
- 3D ultrasound imaging Sparse arrays, coded sequences and image reconstruction.
- Cerebral emboli in transcranial Doppler Semi-automatic annotation, multimodal models and frugal AI.
- Personalised ventilation in intensive care Chest CT analysis for patients with acute respiratory distress syndrome (ARDS), through to clinical transfer.
- Vascular malformations in MRI Volumetric segmentation, unsupervised learning and generative models.
Recent publications
- M. Shekarnabi, E. Roux, J.-C. Richard, S. Bayat, M. Orkisz. Deformation-aided lung segmentation: Introducing the DALS network for efficient learning. Next Research, 2025.
- M. Tamraoui, A. Bernard, E. Roux, H. Liebgott. Complementary Coded Multiplane Wave Sequence for SNR Increase in Ultrafast Power Doppler Ultrasound Imaging. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2025.
- A. Fraissenon, A. Kugusheva, P. Clarysse, E. Roux. Image pre-processing impact on generative model performance for Unsupervised Venous Malformation Segmentation. Computer Methods and Programs in Biomedicine, accepted.
Teaching
Mathematics and applied physics in the GEII bachelor’s programme, medical devices in the TEM vocational bachelor’s degree, electricity and electronics in the second-year Health Sciences degree, and embedded AI for third-year students.
Science outreach
Declics meetings with high-school students, “Trace ton son” workshops at the Fête de la science, hosting school interns, and the CooLearning and Memaudio software.