Image Quality Transfer Enhances Contrast and Resolution of Low-Field Brain MRI in African Paediatric Epilepsy Patients

1.5T or 3T scanners are the current standard for clinical MRI, but low-field\n(<1T) scanners are still common in many lower- and middle-income countries for\nreasons of cost and robustness to power failures. Compared to modern high-field\nscanners, low-field scanners provide images with lower signal-to-noise ratio at\nequivalent resolution, leaving practitioners to compensate by using large slice\nthickness and incomplete spatial coverage. Furthermore, the contrast between\ndifferent types of brain tissue may be substantially reduced even at equal\nsignal-to-noise ratio, which limits diagnostic value. Recently the paradigm of\nImage Quality Transfer has been applied to enhance 0.36T structural images\naiming to approximate the resolution, spatial coverage, and contrast of typical\n1.5T or 3T images. A variant of the neural network U-Net was trained using\nlow-field images simulated from the publicly available 3T Human Connectome\nProject dataset. Here we present qualitative results from real and simulated\nclinical low-field brain images showing the potential value of IQT to enhance\nthe clinical utility of readily accessible low-field MRIs in the management of\nepilepsy.\n