Trainee Project
Title:
Classification by machine learning methods of optical spectra acquired in healthy human skin (patients) according to biological parameters characteristic of healthy skin
Dates:
2022/02/28 - 2022/08/26
Description:
ontext
As part of the SpectroLive clinical trial (carried out at the CHR Metz-Thionville), optical spectra were acquired on
140 patients. The experimental data were gathered in a database made up of more than 2000 optical spectra (at
least 140 x 15) each corresponding to a skin site (each corresponding to several types of healthy skin sites, several
histological classes of cancerous sites or precancerous). The objective of the proposed internship will be to
complete and then use the database (Python-SQL) to develop methods for classifying biological parameters
(apparent skin age, phototype).

Internship missions
The trainee will have to carry out a state of the art in order to define the classification methods (supervised or
not) most suited to the database and to the objectives set (SVM, k-NN, etc.). He / she will have to implement the
method (s) chosen to characterize the relationship between one (more) optical parameters (e.g. intensity on a
given spectral band) and biological parameters: apparent skin age, civil age and phototype, respectively.
The results obtained can be used in a scientific publication.
Department(s): 
Biology, Signals and Systems in Cancer and Neuroscience