Metrics for quality control of results from super-resolution machine-learning algorithms – Data extracted from publications in the period 2017- May 2021

The quality enhancement and restoration of poor-quality low-resolution magnetic resonance (MR) data are paramount for improving patient care, accuracy in diagnosis and quality in clinical research. Since 2017, with the popularity of deep-learning machine-learning algorithms, there have been an increase in interest, and consequently funding, in applying these algorithms to enhance the spatial resolution of MR data. Deep-learning schemes have demonstrated superiority over the more conventional machine-learning algorithms, as they have produced very accurate results in different medical image processing tasks. The increase in the attempts of applying these algorithms to increase the spatial resolution of MRI data parallels an increase in the number of metrics considered for evaluating their performance. This dataset summarises the metrics and strategies to evaluate the performance of super-resolution machine-learning algorithms applied to MRI, from the articles published up to May 2021 in this field.
The aims are two-fold: 1) to inform on the metrics used to evaluate results of super-resolution algorithms 2) to inform on publications that have applied the state-of-the-art deep-learning algorithms to increase the spatial resolution of magnetic resonance images

Castorina, Leonardo V.; Li, Bryan M.; Storkey, Amos; Valdés Hernández, Maria. (2021). Metrics for quality control of results from super-resolution machine-learning algorithms – Data extracted from publications in the period 2017- May 2021, 2017-2021 [dataset]. University of Edinburgh. Centre for Clinical Brain Sciences and School of Informatics. https://doi.org/10.7488/ds/3062.