2018/11/14 by Mahdi S. Hosseini, Hosseini, Mahdi S., Yueyang Zhang +9 · 1 citation
Engineering · Computer Science · Biochemistry, Genetics and Molecular Biology · #Image Processing Techniques and Applications #Advanced Image Processing Techniques #Cell Image Analysis Techniques
paper · pdf · doi:10.48550/arxiv.1811.06038
One of the challenges facing the adoption of digital pathology workflows for\nclinical use is the need for automated quality control. As the scanners\nsometimes determine focus inaccurately, the resultant image blur deteriorates\nthe scanned slide to the point of being unusable. Also, the scanned slide\nimages tend to be extremely large when scanned at greater or equal 20X image\nresolution. Hence, for digital pathology to be clinically useful, it is\nnecessary to use computational tools to quickly and accurately quantify the\nimage focus quality and determine whether an image needs to be re-scanned. We\npropose a no-reference focus quality assessment metric specifically for digital\npathology images, that operates by using a sum of even-derivative filter bases\nto synthesize a human visual system-like kernel, which is modeled as the\ninverse of the lens' point spread function. This kernel is then applied to a\ndigital pathology image to modify high-frequency image information deteriorated\nby the scanner's optics and quantify the focus quality at the patch level. We\nshow in several experiments that our method correlates better with ground-truth\nz-level data than other methods, and is more computationally efficient. We\nalso extend our method to generate a local slide-level focus quality heatmap,\nwhich can be used for automated slide quality control, and demonstrate the\nutility of our method for clinical scan quality control by comparison with\nsubjective slide quality scores.\n