Offre en lien avec l’Action/le Réseau : – — –/– — –

Laboratoire/Entreprise : Laboratoire ImViA, Dijon, France & CHU Dijon, Fran
Durée : 6 mois
Contact : franck.marzani@u-bourgogne.fr
Date limite de publication : 2025-03-31

Contexte :
Wide-band tympanometry (WBT) is a routine and non-invasive clinical tool to estimate the micromechanical properties of the middle and inner ears. The system delivers a pure-tone signal into the ear at several frequencies and records the reflecting signal in the ear canal. Il also analyses the absorbance at different pressures in the ear canal (-300 to +300 daPa). It estimates the reflectance and the absorbance of the recorded sound and thus provides interesting indicators of tympanic membrane and ossicular chain mobility, and middle ear pressure. The ear is connected to the intracranial space through the inner ear structures. An increase in the intracranial pressure (ICP) changes the micromechanical properties of the middle ear and the WBT signal. The relation between intracranial pressure and WBT signal alterations is complex and poorly studied. The WBT signal varies between individuals and these 2 obstacles hamper the estimation of intracranial pressure by WBT. Monitoring the ICP is crucial in patients undergoing intracranial surgical procedures or in those with head trauma. The gold standard for measuring ICP is invasive pressure sensors placed inside the head. Developing non-invasive ICP measurements would be a significant breakthrough in the neurosurgical field.
We have undertaken a clinical study in the intensive care unit (Dijon hospital) for the measurement of ICP with the intracranial invasive system and multiple WBT recordings on both ears in 20 patients after head trauma or surgery. The database includes 1200 recordings with corresponding invasive ICP values (ground truth). Each WBT recording is composed of absorbance values for probe sound frequencies ranging from 120 to 8000 Hz and for ear canal pressures ranging from -300 to +300 daPa).

Sujet :
The aim of the study is to develop a neural network to analyze the data and to estimate the ICP in 3 categories of pressure (low, medium, high). As a first step, the person recruited will have to understand the physical meaning of the signals handled. He/she will then have to propose and implement a deep learning architecture capable of modeling the relation between ICP and WBT signals in order to meet the objective of estimating.
This study opens insights to the understanding of the mechanisms connecting the middle ear micromechanics and ICP.

Profil du candidat :
Master or engineering student in imaging, computer vision.
Strong skills in image analysis, AI, programming such as Python.
Interest in working as part of a team in an interdisciplinarity environment. Interest in healthcare applications.
Enthusiasm for research. Autonomy.

Formation et compétences requises :
Applicants are invited to submit their application to: Franck MARZANI (franck.marzani@u-bourgogne.fr) and Alexis BOZORG GRAYELI (alexis.bozorggrayeli@chu-dijon.fr).
Application must contain the following documents: cover letter, CV, transcript of the last 2 years, at least 2 reference letters or referral contacts.

Adresse d’emploi :
Laboratoire ImViA, Dijon, Université de Bourgogne