Session ThOC. There are 6 abstracts in this session.



Session: in(ex)-vivo 2, time: 11:00am-11:20am

AUTOmated pulse SEQuence generation (AUTOSEQ) for MR spatial encoding in inhomogeneous B0 fields


Bo Zhu1, 2; Neha Koonjoo1, 2; Jeremiah Liu3; Bruce Rosen1, 2; Matthew Rosen1, 2
1MGH / Martinos Center for Biomedical Imaging, Cambridge, MA; 2Harvard Medical School, Boston, MA; 3Department of Biostatistics, Harvard University, Cambridge, MA
We describe a model-free reinforcement learning approach to pulse sequence generation, with an AI agent that explores an unknown MR imaging environment with pulse sequence “actions,” and constructs a model through corresponding RF receive-signal “rewards.” Unlike conventional optimal control methods which require an accurate description of the complete system to perform optimization, our reinforcement learning system is able to simultaneously learn the environmental dynamics and optimize the control (e.g. pulse sequence). In this work, we demonstrate our AI agent learning to generate optimal RF waveforms to perform slice selection in unknown inhomogeneous B0 settings.

Session: in(ex)-vivo 2, time: 11:20am-11:40am

Resolving fiber-specific relaxation and diffusion properties in heterogeneous brain voxels


Joao Pedro de Almeida Martins1; Chantal Tax2; Filip Szczepankiewicz3; Maxime Chamberland2; Carl-Fredrik Westin3; Derek Jones2; Daniel Topgaard1
1Lund University, Lund, Sweden; 2Cardiff University, Cardiff, UK; 3Harvard Medical School, Boston, MA
The brain connectome is comprised of multiple nerve tracts connecting brain areas dedicated to processes such as memory, cognition, language, or consciousness. Fiber tracking by conventional diffusion MRI faces problems in in heterogeneous voxels containing both axons and other tissue components. Combining approaches from solid-state and Laplace NMR, we quantify sub-voxel composition as nonparametric relaxation-diffusion distributions wherein contributions from different tissues are resolved without prior assumptions on the number or properties of the individual components. We achieve clean 3D mapping of the signal fractions from anisotropic tissues that can be used as an input in fiber-tracking algorithms. Relaxation and diffusion parameters are independently estimated for each distinct fiber bundle, potentially giving tract-specific information on chemical composition and microstructure.

Session: in(ex)-vivo 2, time: 11:40am-12:00

Inside-Out MRI for Diagnosing Rechargeable Batteries


Mohaddese Mohammadi
PhD Student, New York, New York
We have developed a Magnetic Resonance Imaging methodology that enables us to image commercial batteries non invasively and in situ. The technique works based on an inside-out approach and tracks the chemical and physical changes inside the battery. Moreover, it enables us to monitor the current distribution inside the battery by capturing the magnetic filed that it generates outside of the battery.

Session: in(ex)-vivo 2, time: 12:00-12:20

Inverse or Direct Detect Experiments and Probes: Which are "best" for In-vivo Research of 13C Enriched Organisms?


Monica Bastawrous1; Maryam Tabatabaei-Anaraki1; Ronald Soong1; Wolfgang Bermel2; Hermann Heumann3; Andre J Simpson1
1University of Toronto, Toronto, ON; 2Bruker BioSpin, Rheinstetten, Germany; 3Silantes GmbH, München, Germany
Recent in-vivo metabolic studies have emphasized the importance of 2D data collection and isotopic enrichment of organisms for metabolite identification. However, it has not been determined which multidimensional NMR experiments, inverse proton detected or 13C detected, offers more metabolite information or if they are complementary to each other and should be run simultaneously. Due to the need of uninterrupted experiments during in-vivo analysis, only one probe can be used during a study. Assuming both experiments need to be run, is it best to run them on an inverse or observe probe? To answer these questions and determine an optimal protocol for in-vivo studies HSQC and HETCOR experiments are compared on TCI and BBO cryoprobes using the keystone species D. magna.

Session: in(ex)-vivo 2, time: 12:20-12:40

Examining the Effects of Superparamagnetic Iron Oxide Nanoparticles (SPIONs) on Daphnia magna


Amy Jenne
University of Toronto, Scarborough, Canada
SPIONs are becoming an environmental contaminant of considerable interest. With their increased introduction to aquatic systems their impacts to organisms, specifically Daphnia magna, is of critical importance. SPIONs are able to be studied through traditional NMR, and Nuclear Magnetic Imaging as negative (T2) contrast agents. When Daphnia are exposed to these particles the changes to the T2 relaxation can be measured. Daphnia were exposed to nine different SPIONs of various sizes, and imaged to determine the uptake. 8 nm and 5 KD had the greatest uptake, and will be utilized in further 2D NMR metabolomic studies in-vivo to gain insight into the exact SPION uptake pathways. This will provide insight into the toxic impacts of these contaminants in the environment.

Session: in(ex)-vivo 2, time: 12:40-1:00pm

Diffusion-weighted (DW) MR Image Reconstruction using Automated Transform by Manifold Approximation (AUTOMAP) on Human Brains using different training sets


Neha Koonjoo1, 2; Bo Zhu1, 2; Matthew Christensen1; John E Kirsch1; Matthew Rosen1, 2
1MGH/Martinos Center/HMS, Boston, MA; 2Harvard University, Cambridge, MA
Low intrinsic Signal-to-Noise Ratio (SNR) in diffusion-weighted (DW) images are recurrent issues especially at high b-values. Here, we apply the deep neural network image reconstruction technique, AUTOMAP (Automated Transform by Manifold Approximation) to in-vivo diffusion-weighted MR data acquired at 1.5 T with varying b-values. We also compared the reconstruction of the images using two different training corpura. The results for AUTOMAP reconstruction showed a significant increase in SNR.