ORGANISATION/COMPANYUniversité Clermont Auvergne
RESEARCH FIELDComputer science
RESEARCHER PROFILEFirst Stage Researcher (R1)
APPLICATION DEADLINE28/06/2020 00:00 - Europe/Brussels
LOCATIONFrance › AUBIERE
TYPE OF CONTRACTTemporary
HOURS PER WEEK35 H
OFFER STARTING DATE01/10/2020
IS THE JOB RELATED TO STAFF POSITION WITHIN A RESEARCH INFRASTRUCTURE?Yes
Subject: Robust graph-based deep learning; application to (bio)medical
Supervisor: Prof. Antoine Vacavant
Laboratory: Institut Pascal, UMR6602 UCA/SIGMA/CNRS
Email and phone: firstname.lastname@example.org ; (+33) 4 71 09 90 82
Abstract (up to 10 lines):
Numerous deep learning (DL) approaches have been proposed for
image analysis tasks in computer-aided (bio)medical applications.
Images are generally considered as examples that are learned directly
to the system, then able to predict specific decisions (e.g. image
segmentation or classification). Nevertheless, image data can be
preferably represented with discrete structures as graphs. This PhD
deals with the development of novel DL methods based on graphs, by
robust approaches, i.e. they are able to resist to inherent image data
noise. Within a multidisciplinary team, composed of scientific members
and medical doctors, several applications can be considered, mostly
related to the 3D reconstruction of vascular networks.
Skills: Good mathematical background; solid skills in programming;
good knowledge of machine/deep learning and image
analysis/processing preferable; experience in medical or biomedical
Keywords: Deep learning; Graphs; Robustness; Image analysis;
Biomedical and medical applications
Description (up to 1 page):
An extremely common methodology for analyzing image data today in
computer-aided (bio)medical applications is the development of deep
neural networks and other related deep learning (DL) architectures
. These networks generally take directly images as inputs during
learning phase, and are then able to predict specific decisions (e.g.
image segmentation or classification) with new image samples.
Recently, a relevant work has been done towards generalizing DL to
graphs [2,3]. Considering graphs is an important task with many
potential applications, such as analyzing data on networks, computer
vision, natural language processing, medical applications, etc.
Additionally, image data can be preferably represented with discrete
structures as graphs. As an illustration of the interest in learning
graphs by neural structures, we can note that 2 libraries, named NSL
(Neural Structured Learning) and Deep Graph Library, have been
developed for this purpose, based on the reputed Tensor Flow
This PhD deals with the development of novel DL methods based on
graphs, by robust approaches, i.e. they are able to resist to inherent
image data noise . We would like to investigate further Graph
Convolutional Networks (GCN)  and their application to 3D vascular
segmentation and reconstruction, which is a very recent concern [7,8].
It is even possible to combine GCN and standard Convolutional Neural
Networks (CNN) to learn and represent both structural and image
intensity information into a single DL framework.
The PhD candidate will join the CaVITI multidisciplinary team (Cardio-
Vascular Interventional Therapy and Imaging), hosted at Le Puy-en-
Velay. It is composed of scientific members (skilled in computer
science, signal/image processing, machine learning, etc.) and medical
doctors (specialized in interventional radiology, hepatology, liver
surgery, etc.) from University Hospital (CHU) of Clermont-Ferrand. This
PhD will be associated to ongoing research projects coordinated by the
host team, in particular the ANR R-Vessel-X  and PHC/Polonium
DeepVesselNets projects. Main applications would be the 3D
reconstruction of liver vessels from MRI (Magnetic Resonance Imaging)
sequences or from microscopic imaging of small animals (synchrotron
acquisitions or μMRI), but the methodologies developed in this PhD
could be applied to other organs, e.g. brain vasculature or lung
References (up to ½ page):
 D. Shen et al. Deep Learning in Medical Image Analysis. Annual
Review of Biomedical Engineering, vol. 19, pp. 221-248, 2017.
 T.N. Kipf and M. Welling. Semi-supervised classification with
graph convolutional networks. In International Conference on
Learning Representations 2017.
 F. Wu et al. Simplifying Graph Convolutional Networks. In
International Conference on Machine Learning, 2019.
 A. Fawzi et al. The robustness of deep networks: A geometrical
perspective. IEEE Signal Processing Magazine, vol. 34, pp. 50-62, 2017.
 H. Yang et al. CPR-GCN: Conditional Partial-Residual Graph
Convolutional Network in Automated Anatomical Labeling of Coronary
Arteries. arXiv:2003.08560, 2020.
 J.M. Wolterink et al. Graph Convolutional Networks for Coronary
Artery Segmentation in Cardiac CT Angiography. In International
Workshop on Graph Learning in Medical Imaging, 2019.
How to candidate?
Contact the supervisor
REQUIRED EDUCATION LEVELOther: Master Degree or equivalent
EURAXESS offer ID: 528425
The responsibility for the jobs published on this website, including the job description, lies entirely with the publishing institutions. The application is handled uniquely by the employer, who is also fully responsible for the recruitment and selection processes.
Please contact email@example.com if you wish to download all jobs in XML.