Context : In general, there are different types of data available to plan a therapy. These data can be from blood tests, imaging, radiographic or nuclear images, electrocardiograms, etc. In this thesis, we deal with the diagnosis of parathyroid pathologies for which there is no automatic process of localization, visualization and detection of this diseases. Practitioners have to use Nuclear Imaging with the use of two X-ray tracers with different operations of 3D reconstruction then subtraction with highlighting of the parathyroids after a grey level shift. The treatment is still requiring manual adjustment of the parameters used, with the result depending on the operator before the images are interpreted. It is then necessary to use machine learning techniques to identify the right parameters, different for each acquisition to automate the process. Thus, thanks to the knowledge of the many known clinical cases (several hundred) at the Besancon University Hospital in the nuclear medicine department of the heart-lung unit, it will be possible to train a model derived from artificial intelligence (AI) to do this diagnostic assistance work, under the control of the department's practitioners. [...]
Supervision and research team : The PhD will be supervised by Moncef Soualhi (UFC Besancon), Jean-Marc Nicod (ENSMM Besançon), Noureddine Zerhouni (ENSMM Besançon), Laurent Comas (CHU Besançon) and Hatem Boulahdour (CHU Besançon). Thus, the supervision team is composed by automation and data scientist researchers in PHM field in multidisciplinary projects. Moreover, the thesis will be developed within hospital professor and informatic experts give significance to the image processing and artificial intelligence algorithms in this field.
Applicant Profile : The candidate with a Master's degree in computer science, data science, mathematical application, automation should have a broad knowledge of image processing, including deep learning techniques and their implementation in software and hardware. Fundamental notions in data explore are also required. Fluency in English is essential. Autonomy, scientific rigor and a strong motivation for the proposed subject will be undeniable assets to successfully complete the thesis. The programming langage can be with Matlab, Python, C++
Candidate to be sent on 23th May 2022