Bridging Parametric And Learning-Based Generative Models To Improve Texture Modelling In Modality-Agnostic Brain Segmentation H/F

INRIA

Bridging parametric and learning-based generative models to improve texture modelling in modality-agnostic brain segmentation
Le descriptif de l'offre ci-dessous est en Anglais
Type de contrat : CDD

Niveau de diplôme exigé : Bac +4 ou équivalent

Fonction : Stagiaire de la recherche

A propos du centre ou de la direction fonctionnelle

Inria is the French National Institute for Research in Digital Science, of which the Inria Côte d'Azur University Center is a part. With strong expertise in computer science and applied mathematics, the research projects of the Inria Côte d'Azur University Center cover all aspects of digital science and technology and generate innovation. Based mainly in Sophia Antipolis, but also in Nice and Montpellier, it brings together 47 research teams and nine support services. It is active in the fields of artificial intelligence, data science, IT system security, robotics, network engineering, natural risk prevention, ecological transition, digital biology, computational neuroscience, health data, and more. The Inria Center at Université Côte d'Azur is a major player in terms of scientific excellence, thanks to the results it has achieved and its collaborations at both European and international level.

Contexte et atouts du poste

This scientific project aims at developing an AI algorithm for the segmentation of brain MRI, with a particular focus on pathologies. It will be conducted under the supervision of:

- Benjamin Billot is a researcher in the Epione team at Inria, and he is the first author of the SynthSeg-related publications (more than 1,000 citations).
- Marco Lorenzi is a Directeur de recherche in the Epione team at Inria and is a chair of AI in Medicine at the 3IA Côte d'Azur. His research focuses on brain modelling and medical image analysis.
- Olivier Humbert is a professor in biophysics and nuclear medicine at Université Côte d'Azur (TIRO team), a chair of AI in Medicine at the 3IA Côte d'Azur, and a practitioner at Hôpital Lacassagne.

Mission confiée

This project will explore novel directions to improve the robustness of state-of-the-art segmentation methods for brain images. In particular, we will focus on the SynthSeg framework, where robustness is achieved by training a network with extremely diverse data sampled from a parametric generative model whose parameters are fully randomised, a technique known as domain randomisation.

This project seeks to extend the domain randomisation framework of SynthSeg. Specifically, we will combine the current parametric model with a learning-based generative model tasked to learn a distribution of textures for healthy and diseased tissues on large public datasets. Texture will be learnt as the residuals between real scans and their reconstruction by the parametric model. We will compare several architectures ranging from supervised networks to latent diffusion models.

Expected results: By accurately simulating texture, we expect the trained network to show an improved ability to characterise healthy and diseased brain tissues, and thus to provide more accurate segmentations. By developing a contrast-agnostic pathology-aware segmentation method, this work would greatly facilitate the deployment and applications of AI-driven solutions in the clinic.

Principales activités

- Download and preprocess existing large public datasets.
- Retrain SynthSeg to get familiar with the framework: https://github.com/BBillot/SynthSeg,
- Compare several architectures (time permitting) to learn a distribution of brain textures.
- Interact with Inria students and researchers, and participate to the scientific life of the team.

Compétences

- Master's degree or engineering degree with computer science, image analysis and/or applied mathematics profile.
- Strong interest for medical applications.
- Knowledge of deep learning.
- Knowledge in digital image processing and medical imaging.
- Good programming skills in Python.
- Good writing skills.
- Good relational and communication skills to interact with professionals from various backgrounds.

Avantages

- Subsidized meals
- Partial reimbursement of public transport costs
- Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
- Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
- Professional equipment available (videoconferencing, loan of computer equipment, etc.)
- Social, cultural and sports events and activities
- Access to vocational training
- Social security coverage

Rémunération

Traineeship grant depending on attendance hours