Ph. D. Project
Title:
Characterization of the modifications of skin optical properties with carcinogenesis in order to define bio-optical markers of skin carcinoma differential diagnosis, using multimodal spectro-imaging methods.
Dates:
2024/10/01 - 2027/09/30
Supervisor(s):
Description:
The "SpectroLive" research project carried out by the CRAN team since 2013, and in collaboration with the CHR Metz-Thionville, was marked by the development (and patent) of an innovative medical device for optical tissue spectroscopy (2013- 2016), followed by the acquisition of transcriptomic data (dataset posted online and published article), histological (slides and diagnostics), and spectroscopic (dataset posted online) as part of a clinical trial carried out on 140 patients with skin carcinomas (2016-2021). The work currently in progress (2021-2024) concerns the exploitation of the original spectroscopic database thus generated and more particularly the development of machine learning algorithms for supervised multi-class classification (including fusion strategies decision) in order to provide the surgeon with real-time diagnostic assistance.
The proposed thesis subject is a continuation of this work by exploring a complementary strategy for exploiting and processing this multidimensional spectroscopic data set (spatial and spectral resolutions, diffuse reflectance RD and autofluorescence AF spectra at several excitations). The objective is to extract information concerning the optical properties of the different skin layers associated with different types of lesions e.g. actinic keratoses, basal and squamous cell carcinomas.
This objective will be achieved on the one hand, by exploiting the set of histological slides produced during the clinical trial (~200) in order to (i) quantify the thicknesses of the skin layers of interest (useful as an a priori knowledge base for modeling) and (ii) carry out an experimental measurement of the optical coefficients (absorption, diffusion or even fluorescence) of these layers thanks in particular to the optical bench with double integral spheres of the CRAN PhotoVivo platform.
On the other hand, all of these parameters will be used for the development and fine-tuning of a new experimental modeling algorithm, i.e. estimation of optical properties, from the set of spectra acquired in vivo. This axis includes the optimization of the existing photon propagation simulation model (by integrating fluorescence) and the resolution of the inverse problem aimed at robustly extracting the values of the optical coefficients of interest.
The proposed thesis subject is a continuation of this work by exploring a complementary strategy for exploiting and processing this multidimensional spectroscopic data set (spatial and spectral resolutions, diffuse reflectance RD and autofluorescence AF spectra at several excitations). The objective is to extract information concerning the optical properties of the different skin layers associated with different types of lesions e.g. actinic keratoses, basal and squamous cell carcinomas.
This objective will be achieved on the one hand, by exploiting the set of histological slides produced during the clinical trial (~200) in order to (i) quantify the thicknesses of the skin layers of interest (useful as an a priori knowledge base for modeling) and (ii) carry out an experimental measurement of the optical coefficients (absorption, diffusion or even fluorescence) of these layers thanks in particular to the optical bench with double integral spheres of the CRAN PhotoVivo platform.
On the other hand, all of these parameters will be used for the development and fine-tuning of a new experimental modeling algorithm, i.e. estimation of optical properties, from the set of spectra acquired in vivo. This axis includes the optimization of the existing photon propagation simulation model (by integrating fluorescence) and the resolution of the inverse problem aimed at robustly extracting the values of the optical coefficients of interest.
Keywords:
Cancer, skin, tissue optics, modeling, light-tissue interactions, instrumentation
Conditions:
Duration: 3 years
Employer: UL
Location: CRAN Campus Brabois Santé site - Faculty of Medicine
Remuneration: doctoral contract
Expected profile: experience in Matlab and/or Python programming, skills in optical experimentation, knowledge of light-biological tissue interactions
Employer: UL
Location: CRAN Campus Brabois Santé site - Faculty of Medicine
Remuneration: doctoral contract
Expected profile: experience in Matlab and/or Python programming, skills in optical experimentation, knowledge of light-biological tissue interactions
Department(s):
| Biology, Signals and Systems in Cancer and Neuroscience |
Publications: