Post-doctoral position in the field of image registration

Post‐doctoral position in the field of image registration 18 month contract (extensible by another 18 month). Start: between February and May 2016 Monthly remuneration: 2914 € gross (2360 € net approximatively) Host laboratory Centre de Recherche en Automatique de Nancy (CRAN) UMR 7039 CNRS/Université de Lorraine 2 avenue de la Forêt de Haye 54516 Vandœuvre‐Lès‐Nancy Department of CRAN Health Biology Signal (Santé Biologie Signal, SBS) Work context Project from the French Research Agency (ANR): Multi‐Modal Endoscopy for Inflammatory lesions of the Stomach (Endoscopie MultiModale pour les lésions Inflammatoires de l’Estomac, EMMIE) Project partners ‐ Laboratoire Electronique Informatique et Image (Le2i, Dijon, Université de Bourgogne). ‐ Laboratoire Pluridisciplinaire de Recherche en Ingénierie des Systèmes, Mécanique et Energétique (PRISME, Orléans, Université dʹOrléans). ‐ Hôpital Ambroise Paré (Assistance Publique ‐ Hôpitaux de Paris, Boulogne‐Billancourt) ‐ Institut Pasteur (IP, Paris) ‐ Institut National de la Santé et de la Recherche Médicale (INSERM UMR 1173, Montigny‐Le‐
Bretonneux) Candidate profile: ‐ Candidate with a PhD (doctorate) in image processing/computer vision OR ‐ Candidate with a PhD (doctorate) in applied mathematics with a significant experience in the field of image processing/computer vision. Required skills ‐ Strong scientific capacities, able to conduct innovative research work ‐ Good writing skills with regard to scientific publications ‐ The candidate must be willing and able to program in C++ language (significant experience in programming) ‐ Familiarity with Matlab Valued skills ‐ Experience in the field of image registration and/or optical flow. Work context The ANR project “EMMIE” aims at the development of a complete endoscope prototype (multimodal gastroscope, software, as well as signal and image processing algorithms) facilitating the diagnosis and follow‐up of stomach inflammatory lesions. This multi‐
disciplinary project gathers scientists with complementary skills. The project partners will deal with different scientific and technical topics which will contribute to the development of the multimodal endoscopic system: ‐
Light‐tissue interaction modelling to understand and to optimize the differentiation of healthy stomach mucosa tissue and tissue parts affected by various degrees of inflammations. ‐
Design and integration into a gastroscope (endoscope used during a gastroscopy) of a system for the “simultaneous” acquisition of a white light image sequence and of sequences acquired at different wavelength (multispectral image sequences). ‐
The multimodal real‐time registration of the white light and multispectral images. ‐
Real‐time detection of the inflammatory lesions using spectrum analysis and classification methods. ‐
Offline cartography of the white light images to increase the limited field of view of the endoscope. The work to be done in the frame of this post‐doctorate position relates to the last topic, namely the development of cartography algorithms based on the mosaicing of images acquired under the white light modality. The aim of mosaicing algorithms is to register the images of video‐sequences visualizing the pyloric antrum of the stomach. (see figure 1(b)). These mosaics with an extended field of view will allow for a second diagnosis after the endoscopic examination, will facilitate the follow‐up of lesion evolution (with the comparison of white light maps built for different examination and on which spectral information is superimposed) and will allow for data archiving which is currently not performed in gastroscopy. More generally, the mosaics will lead to the improvement of the examination and treatment traceability. Region with
few textures
Textured region
Figure 1 – Bi‐dimensional (2D) mosaics built with the algorithms of the CRAN laboratory. (a) Registration of a 200 image sequence of the bladder. (b) First registration tries of the stomach antrum region. Work description The recruited researcher will develop image registration and mosaicing algorithms for the pyloric antrum. The registration of such images is a real challenge since few homologous textures are visible in the images and both the illumination and the viewpoint changes can be important (see figure 1(b)). The recruited researcher will propose 2D registration methods which are notably based on optical flow techniques [1‐4]. He will also examine other registration approaches adapted to the data to be treated. The goal is to conceive algorithms for the registration of images pairs with the best compromise between computation speed, robustness and accuracy. Although the robustness and accuracy criteria are by far the most important, the real‐time mosaic computation would be a real asset of the endoscopic system. The dense correspondence between homologous points (for instance obtained with optical flow methods) will be used to determine the (non‐linear) geometrical transformations required for superimposing consecutive and non‐consecutive images of the video‐sequences on the one hand, and for building the mosaics on the other hand. Optimization methods for the correction of textures and colour discontinuities in the mosaics are also a part of this work. While 2D cartography represents the most important part of the work in the EMMIE project, the feasibility of 3D mosaicing will also be assessed by the end of the project. Approaches for instance based on SLAM (simultaneous localization and mapping) will notably be adapted to the endoscopic data. Work environment The scientific work and the related developments will be mentored by researchers from the image processing field belonging to the CRAN laboratory. The programs will be successively written with Matlab and in C++ under the supervision of the research engineer responsible for the mosaicing software at CRAN. The CRAN laboratory has a long experience in the field of cystoscopic image registration (see the mosaic given by figure 1(a), built with endoscopic images of the inner bladder wall). The recruited researcher can take the developments of the CRAN laboratory in the fields of endoscopic data registration/mosaicing [5‐7] and optical flow [1‐4] as starting point of his work. Publications du CRAN dans le domaine
[1] Sharib Ali, Christian Daul, Ernest Galbrun, François Guillemin and Walter Blondel, Anisotropic
motion estimation on edge preserving Riesz wavelets for robust video mosaicing, Pattern
Recognition, vol.51, pages 425-442, March 2016.
[2] Sharib Ali, Christian Daul, Ernest Galbrun and Walter Blondel, Illumination invariant optical flow
using neighborhood descriptors, Computer Vision and Image Understanding, Accepted - to be
published in 2016.
[3] Sharib Ali, Khuram Faraz, Christian Daul and Walter Blondel, Optical flow with structure
information for epithelial image mosaicing, 37th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society (EMBC), pages 1981-1984, Milan, Italy, August
25-29, 2015.
[4] Sharib Ali, Christian Daul, Thomas Weibel and Walter Blondel. Fast mosaicing of cystoscopic
images from dense correspondence: combined SURF and TV-L1 optical flow method, IEEE
International Conference on Image Processing (ICIP), pages 1291-1295, Melbourne, 2013
[5] Thomas Weibel, Christian Daul, Didier Wolf, Ronald Rösch, François Guillemin. Graph based
construction of textured large field of view mosaics for bladder cancer diagnosis, Pattern
Recognition, vol. 45, issue 12, pages 4138-4150, 2012.
[6] Yahir Hernandez-Mier, Walter Blondel, Christian Daul Didier Wolf, François Guillemin, Fast
construction of panoramic images for cystoscopic exploration, Computerized Medical Imaging
and Graphics, vol. 34, issue 7, pages 579-592, 2010.
[7] Rosebet Miranda, Christian Daul, Walter Blondel, Yahir Hernandez-Mier, Didier Wolf, François
Guillemin. Mosaicing of bladder endoscopic image sequences: distortion calibration and
registration algorithm, IEEE Transactions on Biomedical Engineering, vol. 55, issue 2, pages
541-553, 2008
Application file The application file must include:  A letter of motivation  A Curriculum Vitae  A synthetic description of the past research activities  A publication list  Any documents deemed to be useful for the (letter of recommendation, etc.) Postal and electronic addresses for application: Centre de Recherche en Automatique de Nancy (CRAN‐ENSEM‐UL) M. Christian Daul 2 avenue de la Forêt de Haye 54516 Vandœuvre‐Lès‐Nancy Tel : 03 83 59 57 15 Email : christian.daul@univ‐lorraine.fr