Session ThOE. There are 3 abstracts in this session.

Session: MRI Advances, time: 4:00-4:40

Sensitive Paramagnetic Fluorine-19 MRI Probes and Novel Methods for In Vivo Immune Cell Detection

Eric Ahrens
UCSD, La Jolla, CA
In this talk we describe the development of next-generation perfluorocarbon nanoemulsion 19F MRI probes for in vivo cell tracking and inflammation detection. These probes incorporate novel metal chelates that are fluorous soluble. Paramagnetic metal ions provide a dramatic reduction in the 19F T1 thereby enhancing SNR/time and cell detection sensitivity. Cell penetrating peptide complexes, as nanoemulsion co-surfactant, can further boost cell labeling and sensitivity. Overall, these are key elements to a multi-pronged strategy to advance 19F cell detection sensitivity by an order of magnitude over current technologies. An overview of 19F MRI for preclinical and clinical trials to non-invasive image cell therapy and inflammation is discussed.

Session: MRI Advances, time: 4:40-5:20

T1ρ MRI: Potential Biomedical Applications

Ravinder Regatte
NYU Langone Health, New York, NY
Spin-lattice relaxation in the rotating frame (T1ρ) plays an important role for imaging extracellular matrix (ECM) in musculoskeletal diseases. In my presentation, I will briefly describe the T1ρ relaxation time, contrast and elicit its differences from the conventional relaxation times. I will also discuss methods for T1ρ measurement, including types of imaging sequences used for different applications. In collagen rich tissues such as cartilage, ligaments and skeletal muscle, the primary mechanisms that contribute to T1ρ are from chemical exchange and dipole-dipole interactions, which in turn contribute to image contrast and give a handle to compute changes in the matrix macromolecular content. Finally, I will also demonstrate some specific biomedical applications where T1ρ shows promise for characterizing ECM in abdominal applications.

Session: MRI Advances, time: 5:20-6:00

Learning to See: perceptual learning and MRI image reconstruction

Matthew Rosen1, 2
1MGH/A.A. Martinos Center, Boston, MA; 2Harvard Medical School, Boston, MA
Inspired by the biological perceptual learning archetype, we have developed a noise-robust image reconstruction approach that is based on a data-driven learning of the low-dimensional manifold representations of real-world data, and is implemented with a deep neural network architecture. AUTOMAP, Automated Transform by Manifold Approximation, is an end-to-end automated k-space-to-image-space generalized reconstruction framework that learns a highly-parameterized image reconstruction function optimized for a corpus of training data, and is less sensitive to input corruptions such as channel noise. In this case, we generated training data by tak­ing a large set of images from natural scenes and reverse-encoding them into the sensor domain via a known encoding function, making paired data sets.