Title: The COSMIC Project: Tackling Problems in Biomedical and Astrophysical Imaging
Abstract: Magnetic Resonance Imaging (MRI) is the most powerful non-invasive technique for studying the human brain. Obtaining accurate and High Resolution (HR) images is of paramount importance for the early diagnosis of neurological disorders, such as Alzheimers. HR-MRI, however, requires prohibitively long scan times (over an hour) during which the motion of subjects significantly degrades the quality of the scans. This is particularly evident for younger subjects, such as newborns. Elegant acquisition strategies combined with cutting edge mathematical techniques like Compressed Sensing (CS) can significantly reduce the amount of time required to achieve HR images. The challenges faced for MRI are also relevant in an astrophysical context where e.g. radio images can be reconstructed using CS principles.
The DRF funded COSMIC project brings together NeuroSpin, experts in biomedical imaging, and CosmoStat, experts in astrophysical imaging, to work on developing optimal image reconstruction tools using sparsity and deep learning. The culmination of the project being the development of the Python Sparse data Analysis Package (PySAP). During the talk I will present the objectives of the COSMIC project and the progress made in the development of PySAP.