Thesis (M/F) in multimodality spectro-imaging and machine learning methods for the bio-optical characterization of skin cancers
- Vandoeuvre-lès-Nancy, Meurthe-et-Moselle
- CDD
- Temps-plein
This project combines the complementary skills and areas of expertise of researchers from 2 Joint Research Units common to the University of Lorraine (UL) and the CNRS: CRAN UMR 7039 UL-CNRS (Institute of Computer Sciences) and Institut Jean Lamour IJL UMR 7198 UL-CNRS (Institute of Chemistry).
The recruited person will join the Photodiagnostic project of the BioSIS department of CRAN and the Nanomaterials for optoelectronics team of IJL, specialized in the optical characterization of living and inert materials respectively, using different multi-spectral and multi-scale imaging methods.
This project is part of the current local and regional dynamics of structuring interdisciplinary networks (Interdisciplinary Programs of I-SITE Lorraine University of Excellence and Institut de Photonique Grand Est) in which CRAN and IJL are engaged. The recruited student (M/F) will work in collaboration with the CRAN and IJL teams, as well as in interaction with their partners i.e. CHR Metz-Thionville, GeorgiaTech Institut and CHU Nancy. The project draws on the resources (in equipment and personnel) of the CRAN PhotoVivo platform and the IJL Optics-Lasers competence center.Skin cancers are the most common form of cancer in humans. Among them, melanomas and carcinomas respectively present the worst prognosis (high mortality) and the highest morbidity. The main factors that can improve the prognosis are early detection and complete removal of cancerous tissue. The standard clinical diagnostic procedure involves visually examining the surface of the skin (sometimes using a dermoscope) and, if a cancerous lesion is suspected, performing a surgical biopsy for subsequent analysis and analysis. histopathological classification validating or not the suspicion of cancer. But this procedure has multiple disadvantages: it is invasive (painful and traumatic), long and expensive (it involves at least 4 doctors) and has low diagnostic accuracy: 60% of skin biopsies turn out to be benign and 20% of skin cancers, including 2/3 carcinomas, are not diagnosed. Surgical treatment of the tumor site, and in particular the intraoperative delineation of healthy resection margins, must also be optimized because in 1/4 of cases, the cancer is not completely removed implying surgical revision or a recurrence of the cancer if the patient is not treated.
Faced with these medico-economic challenges, optical spectro-imaging methods are the subject of numerous research studies given their potential for early detection and non-invasive, in vivo and real-time characterization of skin cancers. The interest of these so-called “optical biopsy” methods lies in their sensitivity to modifications in the optical properties of the skin at the tissue, cellular and subcellular scales directly linked to pathological morphological and metabolic modifications during cancerous development. In this field, multimodal approaches provide the best diagnostic classification performance because they make it possible to combine the complementary strengths of different optical methods (imaging and/or spectroscopy).
The challenge of the proposed thesis subject is, through a multi-scale interdisciplinary approach and by coupling for the first time information from four optical spectroscopy modalities (diffuse reflectance DRS, autofluorescence AFS, Raman RS and infrared IRS), to make the link between the biophotonic characteristics (spectral signatures, optical parameters) of healthy and pathological skin samples (i) on the macroscopic scale in vivo (DRS at AFS) and (ii) on the microscopic scale ex vivo (AFS, RS mappings and IRS), in order to identify new combinations of discriminating bio-optical markers to more effectively solve the problem of in vivo diagnosis of skin cancers.
Within CRAN, the student (M/F) will be in charge of (i) the preparation of biological samples e.g. recovery of tissue blocks and preparation of slides (materials, thickness of sections, staining, etc.), (ii) of the experimental study of macroscopic optical properties (absorption, diffusion) on slides and (iii) data processing such as: analysis and unmixing of in vivo spectra using in vitro spectra, supervised classification by merging all modalities and modeling (numerical simulations, estimation of optical properties).
Within IJL, the student (M/F) will be in charge of (i) experimental studies of endogenous fluorescence and Raman mapping on slides without staining and IR mapping (THZ, Brillouin) on slides with and without staining and (ii) data processing and analysis such as: correction of Raman spectra and segmentation of fluorescence, Raman, THz and Brillouin maps.RequirementsResearch Field Engineering Education Level PhD or equivalentResearch Field Computer science Education Level PhD or equivalentResearch Field Mathematics Education Level PhD or equivalentLanguages FRENCH Level BasicResearch Field Engineering Years of Research Experience NoneResearch Field Computer science Years of Research Experience NoneResearch Field Mathematics Years of Research Experience NoneAdditional InformationAdditional commentsRequested profile :
- Graduated from MSc in biomedical engineering and/or signal/image processing and/or biophysics
Expected technical and scientific skills:
- Biological tissue optics, biophotonics spectroscopy and imaging, instrumentation
- Numerical signal/data processing and analysis (MATLAB)
- Machine learning
- Light-tissue interaction modelling
- Very good level in English (speaking and writing)
- Autonomous, go-ahead and proactive
- Motivated by the development of innovative solutions for the field of health engineering
- Able to synthesize and exploit data from different partners associated in the project Website for additional job detailsWork Location(s)Number of offers available 1 Company/Institute Centre de Recherche en Automatique de Nancy Country France City VANDOEUVRE LES NANCY GeofieldWhere to apply WebsiteContact CityVANDOEUVRE LES NANCYSTATUS: EXPIRED
EURAXESS