Joshua Mennecke

Thesis Topic: Accelerating B1-Mapping in Magnetic Resonance Imaging

Supervisors: Robin Heidemann (Siemens Healthineers), Patrick Liebig (Siemens Healthineers), Florian Knoll

Description

In MRI the acquisition time of images is a crucial property, influencing the usability and cost of this medical imaging technique. In spatial optimizations the reduce of phase encoding steps causes an undersampling of the acquired data while gaining speed. Different approaches like SENSE and GRAPPA were developed for reconstructing the missing data during the postprocessing step. However, those techniques introduce additional aliasing artifacts caused by the way of undersampling. This problem was solved in semi-random undersampled methods like Compressed Sensing (CS) at the cost of longer reconstruction times.

In this thesis a different approach was considered, first presented by Hess et al. [1]. They used joint transmit and receive coil information to iteratively reconstruct missing data points in k-space. Hereby, the B1 mapping benefits essentially by the new dimension with transmit coil information. The main goal of this thesis will be the optimization of their developed approach by several hyperparameters to improve both its runtime and quality simultaneously.

Furthermore, an Autoencoder architecture will be constructed for solving the reconstruction with the use of Deep Learning. This could also yield a much faster reconstruction at the cost of possibly long training times.

[1] Hess AT, Dragonu I, Chiew M. “Accelerated calibrationless parallel transmit mapping using joint transmit and receive low-rank tensor completion”. Magn Reson Med. 2021 Nov; pp. 2454-2467.