Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : XLIM, university of Poitiers
Durée : 12 months
Contact : olfa.ben.ahmed@univ-poitiers.fr
Date limite de publication : 2024-01-31

Contexte :
We are looking for an outstanding and highly motivated Postdoctoral researcher in artificial intelligence, data science, or related fields to work on Meta-learning for low-prevalence disease detection. This postdoctoral position is part of the ANR JCJC MIMIC research project.

Sujet :
Context:

Deep learning-based approaches have seen an impressively good performance in the computer vision domain. However, huge, labeled datasets are needed to train on. Collecting such extensive annotated data is time and resource-consuming and it is not feasible for real-world applications, especially in the medical domain [1]. Developing deep learning approaches in the medical domain presents several challenges namely the scarcity and heterogeneity of data. In this position, the selected candidate will work on proposing data-efficient medical image analysis methods by addressing the current limitations of deep learning models in the medical domain.

Objectives:

This position aims to develop efficient deep-learning models using a small available medical imaging dataset for low-prevalence disease detection. We will investigate the recent development of meta-learning approaches [2] to facilitate quick adaptation of deep neural networks trained on data samples of common diseases for the identification of diseases with much less annotated data. Specifically, the selected candidate will focus on developing supervised meta-learning approaches taking into account the specificity of medical images. The work includes also leveraging the limited number of labeled images, along with potentially available unlabeled images to enhance the performance of the trained meta-learner.
The proposed methods will be tested and evaluated on medical diagnosis tasks representing real-world scenarios of low-prevalence disease detection, assessing the models’ ability to detect disease from small amounts of data.

Data used to implement different methods will be issued from our archives of imaging data collected in previous and current projects that involve Poitiers University Hospital. In addition, we will use several SOTA publicly available medical imaging datasets.
References :

[1] Chen, Xuxin, et al. “Recent advances and clinical applications of deep learning in medical image analysis.” Medical Image Analysis 79 (2022): 102444.
[2] Ouahab, Achraf, Olfa Ben-Ahmed, and Christine Fernandez-Maloigne. “A Self-attentive Meta-learning Approach for Image-Based Few-Shot Disease Detection.” MICCAI Workshop on Resource-Efficient Medical Image Analysis. Cham: Springer Nature Switzerland, 2022.

Profil du candidat :

• Ph.D. in Computer science and signal processing / Applied mathematics/ Artificial Intelligence, data sciences
• Strong skills in deep learning, mathematics, science, and data analysis…
• Programming experience in Python
• Experience in the medical imaging field would be a plus
• Experience in meta-learning will be a plus
• Excellent oral and written communication skills

Formation et compétences requises :
Salary :
Remuneration and social benefits are based on the collective wage agreement for public-sector employees at the national French level, considering previous years of experience. Salary between €2905 and €4081 gross monthly, depending on experience.

Start date and duration: The exact starting date is flexible and will be arranged with the candidate, and it should take place between February and April 2024. The position is funded for 12 months with a possible extension.

How to apply: Send your CV with a publications list, 2 names of references, and a motivation letter to olfa.ben.ahmed@univ-poitiers.fr

Adresse d’emploi :
Xlim site de Futuroscope, university of Poitiers