Impact of Whole Slide Image Integrity on Image Analysis

Digital image analysis is a major value added application of digital pathology. Whole slide image (WSI) scanners provide a mechanism to digitize glass slides, allowing image analysis algorithms to be applied to selected regions of interest or to the entire digital slide. Several factors could influence image analysis results, including the consistency of image acquisition. The aim of this study was to evaluate the scanning performance of a WSI scanner and its impact on quantitative image analysis. The same tissue microarray (TMA) glass slide of prostate cancer cores stained with FOX M1 antibody (source AbCam, Cambridge, MA) was scanned weekly using an Aperio ScanScope XT WSI scanner (Aperio, Vista, CA), for a period of 12 consecutive weeks. A log of all WSI scanner maintenance was kept. Each digital WSI was analyzed employing Aperio Positive Pixel Count Image Analysis Algorithm (version 9.1) to evaluate four individual prostate cores selected from this TMA. The total pixel average intensity of each core was quantified and the coefficient of variation (CV) calculated. Subjective human evaluation of cytoplasmic immunoreactivity using a digital display revealed no differences between images. However, image analysis results for all digitized slides revealed minor score variations of each core with a CV between 0.0038 and 0.0059. The largest variations were noted when the WSI scanner light bulb was replaced (CV between 0.0071 and 0.0127). Minor variations in immunohistochemical image analysis results were observed when the same digital image was repeatedly scanned over a 3-month period. The largest variation occurred immediately after major WSI scanner maintenance. Additional studies are required to determine the clinical impact of these variations on image analysis.