Hongxiang Lin1, Matteo Figini1, Felice D'Arco2, Godwin Ogbole3, David W Carmichael4,5, Ikeoluwa Lagunju6, Helen Cross4,7, Delmiro Fernandez-Reyes6,8, and Daniel C Alexander1
1Centre for Medical Image Computing, Department of Computer Science, University College London, London, United Kingdom, 2Department of Radiology, Great Ormond Street Hospital, London, United Kingdom, 3Department of Radiology, College of Medicine, University of Ibadan, Ibadan, Nigeria, 4Great Ormond Street Institute of Child Health, University College London, London, United Kingdom, 5Department of Biomedical Engineering, King’s College London, London, United Kingdom, 6Department of Paediatrics, College of Medicine, University of Ibadan, Ibadan, Nigeria, 7Great Ormond Street Hospital for Children, London, United Kingdom, 8Department of Computer Science, University College London, London, United Kingdom
1Centre for Medical Image Computing, Department of Computer Science, University College London, London, United Kingdom, 2Department of Radiology, Great Ormond Street Hospital, London, United Kingdom, 3Department of Radiology, College of Medicine, University of Ibadan, Ibadan, Nigeria, 4Great Ormond Street Institute of Child Health, University College London, London, United Kingdom, 5Department of Biomedical Engineering, King’s College London, London, United Kingdom, 6Department of Paediatrics, College of Medicine, University of Ibadan, Ibadan, Nigeria, 7Great Ormond Street Hospital for Children, London, United Kingdom, 8Department of Computer Science, University College London, London, United Kingdom
Machine-learning based image quality transfer estimates convincing 1.5T or 3T T1w, T2w, and FLAIR images from 0.36T images revealing fine detail and lesion conspicuity.
Figure 5: An example of IQT enhancement on real LMIC patient data for T1w, T2w and FLAIR images. Middle cerebral artery ischemia was detected involving the perirolanding regions. The lesion area highlighted by the red arrows shows hyperintense signal in T2w and FLAIR, and hypointense signal in T1w.
Figure 4: An example of IQT enhancement on real LMIC T1w, T2w and FLAIR images from a normal brain subject. The 1.5T reference has a different orientation with 0.36T input and IQT enhancement. The demonstrated three sequences were enhanced to 3T-like image quality with isotropic voxels. Low-field images show cross-talk and field inhomogeneity artifacts, but they were corrected during IQT Pre-processing.
