Acutance is a subjective parameter which indicates the quality of edges in an image. Objective metrics for measuring image acutance are helpful for designing new imaging protocols and sequences in magnetic resonance imaging (MRI) studies. In addition to this, image acutance metrics have a significant role in the design and optimisation of post-processing algorithms used for restoration and sharpening of MR imagery. Most of the existing blur/sharpness metrics are specifically designed for natural-scene (panoramic) images. A blur/sharpness metric suitable for MR imaging applications is absent in the literature. To fill this gap, a computationally fast metric, 'largest local gradient-based sharpness metric (LLGSM)', for measuring sharpness and blur in MR imagery, is proposed in this paper. The LLGSM is the root mean square (RMS) of exponentially weighted elements in an array of lexicographically ordered largest local gradient (LLG) values in the image, sorted in descending order. In terms of overall agreement with subjective scores, and computational speed, the LLGSM is observed to be more efficient than its alternatives available in the literature.
Keywords: Acutance; Blur metric; Edge quality index; Image sharpness; Unsharp masking.
© 2022. International Federation for Medical and Biological Engineering.