Research‎ > ‎

PhD thesis proposal

MEDICAL IMAGE PROCESSING IN OPHTALMOLOGY: Analysis if retinal vessels in adaptive optics images for the study of SVD diseases

  • Funded by « Investissements d’avenir, Recherche hospitalo-universitaire-vague 2 (RHU-2) », french Ministry for research.
  • Grade : M2, with a specialty in image processing and analysis
  • Doctoral school  EDITE (see the PhD proposal on EDITE's website) 
  • Supervisors: Prof. Florence Rossant (ISEP) ; Prof. Isabelle Bloch (Télécom ParisTech), Prof. Michel Pâques (CIC de l’hôpital des Quinze-Vingts)
    • Télécom ParisTech - Dept TSI - 46 rue Barrault - 75013 Paris
    • ISEP, Dept. SITe, 10 rue de Vanves 92130, Issy les Moulineaux
    • CIC, Hôpital des Quinze-Vingts, 28 rue de Charenton, 75012 PARIS
  • Starting date: last trimestre of 2017
  • Required skills: image processing and analysis, segmentation, mathematical morphology, filtering, image deconvolution, image registration, Matlab and C/C++ programming.
  • Other skills: good level in english, team work.
  • Application must include:
    • Detailed CV
    • Grades in M1 et M2
    • Motivation letter
    • 1 or 2 recommandation letters
and must be sent to Florence Rossant (Florence.Rossant[at] and Isabelle Bloch (Isabelle.Bloch[at] in pdf format. 

Project summary and PhD thesis subject

Cerebral small vessel diseases of the brain (SVDs) are involved in one-third of ischemic strokes and contribute significantly to cognitive decline and disability, occurring with aging in the general population. The main objectives of the TRT_cSVD project are: 1) to identify new genes in SVDs; 2) to identify innovative biomarkers and clinical tools for therapeutic evaluation in patients; 3) to develop innovative evaluation and rehabilitation approaches targeting early deficits and disability and, 4) to validate our findings in early-stage clinical studies. TRT_cSVD collaborative project gathers in a transdisciplinary consortium 25 partners including clinical and imaging research teams, industrial partners.
The Clinical Investigation Center of the Quinze-Vingts Hospital, ISEP (Institut Supérieur d’Electronique de Paris) and Telecom ParisTech have collaborated for almost 10 years on medical issues related to retina diseases. ISEP and Telecom ParisTech have proposed innovative methods for the automatic or semi-automatic analysis of retina images, acquired with different imaging modalities. Software tools are exploited by the doctors of the CIC for clinical studies.
In the TRT_cSVD project, ISEP and Telecom ParisTech are involved in point 2), through the analysis of retinal vessels in adaptive optics (AO) images. AO is a recent technology that enables clinicians to observe retinal structures, such as vessels, with a micrometric resolution. As the features exhibited by retinal vessels are very close to the ones of small brain vessels, we assume that the analysis of retinal vessel alterations observable in AO images will enable us to define consistent biomarkers for clinical studies of SVD diseases.
So, the purpose of the PhD thesis consists in the segmentation of retinal vessels in AO images, for assessing morphological alterations due to SVDs, and in the analysis of sequences of images, acquired at about 100Hz, for the characterization of the vessel pulsatility and vasomotor activity. Research work will be conducted in the continuity of our previous work during the ANR Reveal project [1,2,3,4]. Possible approaches will rely on active contour models, dense registration, optical flow estimation, particle filtering, minimal path estimation. New biomarkers will be then extracted. Quantitative evaluation will be conducted in order to evaluate the robustness and the accuracy of the proposed algorithms. Research will be conducted in close collaboration with the CIC and also with Imagine Eyes [5], the industrial partners developing and improving the OA technology.

[1] N. Lermé, F. Rossant, I. Bloch, M. Paques, E. Koch, J. Benesty, A Fully Automatic Method For Segmenting Retinal Artery Walls in Adaptive Optics Images, Pattern Recognition Letters, 2015, Vol. 72, pp 72-81.
[2] N. Lermé, F. Rossant, I. Bloch, M. Pâques, E. Koch, Coupled Parallel Snakes For Segmenting Healthy and Pathological Retinal Arteries in Adaptive Optics Images,  International Conference on Image Analysis and Recognition (ICIAR), Vilamoura, Portugal, November 2014, vol. LNCS 8815, pp. 311-320.
[3] N. Lermé, F. Rossant, I. Bloch, M. Pâques, E. Koch, Segmentation of Retinal Arteries in Adaptive Optics Images, International Conference on Pattern Recognition (ICPR 2014), pp 574-578, Stockholm, Sweden, August 2014. Best Scientific Paper Award, Track Biomedical Image Analysis.
[4] ANR Reveal : ANR-12-TECS-0015-03