Candidate Profile:
• Good level of programming (C++ desired, Visual Studio or Qt Creator, VTK library)
• Experience and skills in image processing field
• Knowledge in medical image analysis is expected
• Medical knowledge would be a plus
Subject: Cardiovascular diseases are the leading cause of death worldwide. It is essential to prevent, detect, identify changes and treat these diseases. Multimodal imaging is a promising method for the combined study of cardiac function and molecular targets. The thesis goal is coupling metabolic imaging of PET (Positron Emission Tomography) and the anatomical and functional imaging of MRI (Magnetic Resonance Imaging). Especially, the combined analysis of carbohydrate metabolism with 18F-fluorodeoxyglucose PET (FDG) and the extent of necrosis with MRI could help refine the assessment of post-infarction myocardial viability, and identify within the myocardium of "trigger points" that may cause ventricular rhythm disorders. This functional and metabolic characterization of the myocardium could allow to develop or to direct new therapeutic strategies.
The large volume of data makes manual analysis thereof, for each patient, complex and tedious. In addition, manual data analysis lacks objectivity and reproducibility. This due to a high variability between and within operators, compromising the validity of the results obtained both in research and clinical field. However, the multimodal imaging is underused due to the lack of tool for the joint analysis of PET and MRI in the cardiovascular field. It therefore appears necessary to develop analytical tools assisted by dedicated computer to extract relevant information from clinical and experimental observations.
The information fusion will enable practitioners to obtain a rich and meaningful representation for better viewing. It will facilitate the information reading to enlighten the returned diagnosis. The development of this multimodal medical image analysis method consists in:
• Performing the modalities registration (dynamic MRI/late MRI/PET). This registration will be automatic or semiautomatic to guarantee a high level of robustness. The study of the state of the art will determine which family of methods is the most suitable (rigid/non rigid, geometric / iconic).
• Analyzing various sources to determine the health of the heart muscle and classify the myocardial tissue, establishing the degree of viability of these tissues.
The aim of this thesis is to develop a new approach and a generic tool for analyzing multimodal image suitable for clinical imaging of patients with ischemic pathology.
This work will benefit from a privileged partnership with members of IMAC team Dijon University Hospital, led by Professor François BRUNOTTE in close cooperation with MCU-PH Alain LALANDE which has a large multidisciplinary experience in the medical world, with recognized expertise and an operative control of involved technologies, especially within the framework of the IMAPPI Equipex, and Professor Alexandre COCHET.