Context of the work :
The TR‑NEWS ultrasonic platform developed at INSA Centre Val de Loire enables the acquisition of nonlinear ultrasonic signatures from soft biological tissues using advanced excitation schemes such as quaternion-coded signals[1]. These experiments, applied to skin imaging, are designed to quantify nonclassical nonlinearities linked to structural changes in biological tissues.

In the DICA#UP project, a major challenge is the lack of interoperability between heterogeneous ultrasonic datasets and metadata formats (proprietary TR‑NEWS descriptors, experimental context, acquisition parameters, etc.) as identified in the project description. This lack of standardization makes it difficult to compare experiments, automate the analysis pipeline, or feed machine learning algorithms.


The present internship focuses on the processing and modelling of nonlinear ultrasonic data, and on the construction of an interoperable annotation dictionary compatible with DICONDE-like structures. The goal is to formalize the signal-processing workflow and enable standardized annotation of advanced ultrasonic experiments on skin.
Experiments to be analysed include:
It is therefore necessary to develop tools allowing (i) the analysis of the experimental results obtained with the trained models as well as (ii) the annotation of the acousto-mechanical metadata and ultrasonic images.
Previous research in the team has shown that automatic classifiers using information-divergence based approaches [1-3] are more efficient for remote health monitoring and diagnosing suspects.
Objectives:
References :
[1] S. Dos Santos, M. Maslouhi, and K. A. Okoudjou, Recent Advances in Mathematics and Technology (Applied and Numerical Harmonic Analysis), Springer Nature, 2019. https://www.springer.com/gp/book/9783030352011
[2] C. Kozena, V. Kus and S. Dos Santos, "Hysteresis and memory effects in skin aging using PM space density identification," 2016 15th IEEE BEC, 2016, pp. 179-182, doi: 10.1109/BEC.2016.7743758 .
[3] Dos Santos, S., Farova, Z., Kus, V., & Prevorovsky, Z. (2012, May). Echodentography based on nonlinear time reversal tomography: Ultrasonic nonlinear signature identification. In AIP Conference Proceedings (Vol. 1433, No. 1, pp. 203-206). American Institute of Physics.
Profile :
The student should have a strong motivation on research domain such as bioengineering, computational simulation and modelling. The intern should be proficient in modelling and programming in different languages such as C, Java, Matlab, and Python. In addition to programming skills, knowledge of the basics of image processing is highly desirable. Intern should have good written and verbal communication skills and enjoy working in a multi thematic team. Good English language skills are required.
Starting period : Spring Semester 2025-2026
Supervisors :
INSA Centre Val de Loire, serge.dossantos@insa-cvl.fr
INSA Lyon, frederique.biennier@insa-lyon.fr
The monthly gratuity will be paid according to the French law :
(https://www.service-public.fr/particuliers/vosdroits/F32131 )