Meeting Banner
Abstract #2829

Application of a Clinically Viable Deep-Learning-Based QSM Workflow on Stroke Cases

Juan Liu1 and Kevin Koch1,2

1Joint Department of Biomedical Engineering, Marquette University and Medical College of Wisconsin, Milwaukee, WI, United States, 2Department of Radiology, Medical College of Wisconsin, Milwaukee, WI, United States

Quantitative Susceptibility Mapping (QSM) can quantitatively estimate tissue magnetic susceptibility, which enables differentiation of diamagnetic calcifications and paramagnetic hemorrhages. The translation of QSM into clinical practice faces technical implementation challenges, particularly the QSM inversion process. In the clinical practice current QSM post-processing techniques are constrained due to large slick thicknesses, which result in compromised background field removal and streaking artifacts in QSM images. To address these limitations, here we a apply a deep-learning-based QSM pipeline, including: (1) a 2D neural network to construct brain masks, (2) a background field removal deep neural network reveal local tissue fields, and (3) a QSM inversion deep neural network. Nine patients with stroke were scanned using a clinical susceptibility-weighted MR protocol were used to demonstrate that the proposed clinically viable QSM workflow can effectively detect microbleeds and differentiate calcifications from hemorrhages.

This abstract and the presentation materials are available to members only; a login is required.

Join Here