Section: Overall Objectives

Overall objectives

Medical Imaging, Neuroinformatics, Neuroimaging, Data Fusion, Image Processing, Neurological Pathologies, Modelling of normal and pathological behavior of the human brain, e-health, HealthGrids

In the ’70s, the major research advances in biomedical signal and image processing came from the introduction of tomography (literally from (TOMOS) – SECTION and (GRAPHO) – WRITE) which was in some way a revolution compared to the current image standard consisting of projected images (X-ray, scintigraphy ...) Some of these major discoveries include CT scans and, later on, MRI, SPECT or PET scans, complemented by the increasing influence of electrophysiological data (MEG, EEG, ECG ...). This period was the early stage of the in vivo image slices of the human body. Major research advances in the ’80s included 3D imaging in the clinical context. The ’90s, with the advent of data fusion algorithms (mostly registration) and new sequences of Magnetic Resonance Images – MRI (especially fast and functional sequences), added one or two more dimensions to clinical imaging. Most of our current research challenges follow this evolution by:

• adding new spatio-temporal dimensions to the anatomical and functional data at the acquisition and the analysis level,

• adding new scales of analysis (nano or micro biological and molecular images to macro medical images,

• adding network interconnection between clinical data centers which will extend the corpus of accessible information,

Then, it becomes possible today to acquire anatomical (i.e structural) and physiological information (e.g. functional), which are complementary in the same subject. However, this increase of available information for diagnosis and treatment must be balanced by an equivalent improvement in the quantity of data the user can integrate and interpret. So, the traditional way physicians use these data is often sub optimal implying that today important number of valuable information is still neglected during the medical decision process.

In order to advance, the physician has to consider not only the original biomedical signals but also the processing tools to get an optimal interpretation. Processing biomedical images and signals is still very complex and is not only based on matching these data but also on the knowledge about the observed structures and their interactions, either anatomical, structural or functional in nature. It is clear today that, except the emergence of new acquisition modalities, the advent of new algorithms and new systems, capable to jointly use all these information, will improve patient’s care, diagnosis and therapy.

Based on this perspective, the classical way to use medical imaging data, mostly based on subjective human interpretation, becomes increasingly suboptimal. In addition, the pressure from the society for a cost effective use of the equipments on the one hand, and a better traceability and quality assurance of the decision making process on the other hand, strengthened the need of advanced computer assisted biomedical imaging systems.

Besides this evolution regarding acquisition and application of medical images, in parallel, there is an evolution of high band electronic communication systems (e.g. through the Internet). This argues even more strongly to share different resources (medical images, processing algorithms...) between users of the same community sharing a common interest on pathologies, research domains or even education. This explains why recent advances coming from information technologies through GRID infrastructures are becoming an acute issue in clinical medical imaging for assisting medical imaging actors to share their heterogeneous and distributed resources for the purpose of improving their clinical practice, or being more acute (i.e. specific in doing research on pathologies).

Research activities of the VisAGeS U746 team follow the global evolution of our domain of activity for the design and development of computational models of living systems to be confronted to biological images/signals/measurements. This translates to better understand the behavior of normal and pathological CNS organs and systems, at different scales, which includes the purpose of imaging the brain pathologies in order to better understand the pathological behavior from the organ level to the cell and the molecule, and the modeling of normal and pathological group of individuals (cohorts) from image descriptors. This includes the challenge of discovery of unlikely facts (e.g. rare events in image series), data mining and knowledge discovery from image descriptors, validation and certification of new drugs from imaging features and more generally the integration of neuroimaging in Neuroinformatics.

The medical application objectives are focused on pathologies of the central nervous system, with a particular effort on extraction of new imaging biomarkers for brain pathologies (e.g. Multiple sclerosis, epilepsy, dementia, neuro-degenerative brain diseases, brain vascular diseases, strokes, neuropaediatrics, psychiatry, …).