Kieran O’Brien

This paper evaluates the accuracy and repeatability of quantitative susceptibility mapping using an anthropomorphic phantom, finding better performance at 3T than 7T.

This study evaluates four phantom doping materials for quantitative susceptibility mapping (QSM) by measuring molar susceptibility and test-retest repeatability at 3T and 7T.

The paper develops a multi-model consensus deep learning approach for automated gold fiducial marker segmentation in prostate MRI, achieving high sensitivity and precision.

QSMxT improves QSM processing by reducing artefacts and requiring minimal parameter tuning through a robust masking and two-pass reconstruction method.

DeepQSM uses deep learning to solve the dipole inversion in MRI for susceptibility mapping, improving the identification of brain substructures and their magnetic properties in diseases like multiple sclerosis.

SHARQnet improves background field removal in QSM by using a deep convolutional neural network, outperforming current methods without parameter tuning.