Menu Close

Thesis Prize in Biomedical Engineering

Congratulations to Florentin Kucharczack who won the thesis prize in biomedical engineering.

The French Society for Biological and Medical Engineering, the EMBS Chapter of the IEEE French Section and the Alliance for Biological and Medical Engineering, with the support of the Stic-Santé Community, awarded their 2019 thesis prizes during the SFGBM General Assembly on May 27, 2020.

Summary of Florentin’s thesis: Positron Emission Tomography (PET) is a nuclear imaging modality that has a prominent place in the diagnosis of neurodegenerative dementia. The most widely used radiopharmaceutical tracer, 18F-FDG, allows a volume mapping of cerebral metabolism. The dementia scintigraphy argument is based on the highlighting of a relative hypo-metabolism of a particular region of interest (ROI) with respect to another, usually its contralateral symmetry. However, some case studies are very much difficiles to be interpreted with the naked eye, mainly at an early stage of disease progression. Until now, the development of (semi-)automatic tools for direct comparison of ROIs has been limited by the lack of knowledge of the statistics that the reconstructed data follow; the main methods already developed
preferring then to use large databases to evaluate the reconstruction to be analyzed through a dissimilarity score with respect to a group of controls. In this thesis, we propose a new fully integrated methodology ranging from the reconstruction of activity images to help in the diagnosis of dementia. Based on the reconstruction of intervals from confiance, the proposed approach allows 1/ to directly access information on the statistical variability of the data, 2/ to reconstruct qualitatively and quantitatively conclusive images to facilitate the reading of the examination by the nuclear physician, 3/ to provide a risk score of the patient to be affected by neurodegenerative dementia. The results obtained with the latter are comparable with tools validated in routine clinical practice, without requiring any information other than the PET acquisition data alone.

Key words: PET, medical imaging, tomographic reconstruction, statistical variability, intervallist approaches, region of interest comparison, diagnostic aid, neurology.