13:00 - Representation learning for accelerating multi-contrast MRI
Chinmay Rao, PhD-student - Leiden University Medical Center, The Netherland
MRI is an inherently multi-contrast modality, capable of measuring different aspects of a given anatomy. My PhD work has focused on accelerating MR imaging in its capacity as a multi-contrast technique. In this talk I will focus on one of the main projects of PhD, namely of accelerating multicontrast acquisitions by undersampling the kspace and using side contrast information in the reconstruction through a representation learning technique called content/style modeling.
Additionally, I will also give my account of more human aspects of my PhD experience such as collaboration with industry partners and radiologists.
13:30 - RIM for quantitative MRI
Dirk Poot, Principal Investigator - Erasmus MC, The Netherlands
We use RIMs to quantify MRI tissue properties. In this presentation I will focus on applications of the RIM for quantification of T1, T2, and diffusion, in particular focussing on the way in which we validated the methods.
14:00 - Generalizable, robust and fast: applications of deep-learning accelerated MRI
Matthan Caan, Principal Investigator - Amsterdam UMC, The Netherlands
The Recurrent Inference Machine has been developed as a learned inverse problem solver. We created RIMs for accelerating MRI. This presentation will focus on applications, in reconstructing multiple contrast MRI, generalizability to pathology, methods and evaluation for mid-field strength MRI, and dynamic imaging.
14:30 - AI-Based Interventions Along the Quantitative Molecular MRI Pipeline: The Quest for Speed, Specificity, and Histological Fidelity
Or Perlman. Assistant Professor - Sagol School of Neuroscience - Tel Aviv University, Israel
Changes in in vivo molecular properties, such as intracellular pH and protein/metabolite concentrations, often underlie neurological diseases and precede anatomical or structural changes. While non-invasive imaging of these processes is key for early diagnosis, current clinical in-vivo molecular imaging techniques are often slow, nonspecific, or require radioactive or metal-based contrast agents. This talk will present several rapid and quantitative molecular MRI strategies that integrate biophysical models with AI-based interventions. Potential implications for understanding molecular mechanisms, monitoring cancer treatment, and characterizing neurodegeneration will be discussed.