Image Compression Impact On Quantitative Angiogenesis Analysis Of Ovarian Epithelial Neoplasms.

Abstract

This study aims to investigate the impact of digital image compression on manual and semiautomatic quantification of angiogenesis in ovarian epithelial neoplasms (including benign, borderline, and malignant specimens). We examined 405 digital images (obtained from a previously validated computer-assisted analysis system), which were equally divided into 5 groups: images captured in Tagged Image File Format (TIFF), low and high compression Joint Photographic Experts Group (JPEG) formats, and low and high compression JPEG images converted from the TIFF files. Microvessel density counts and CD34 endothelial areas manually and semiautomatically determined from TIFF images were compared with those from the other 4 groups. Mostly, the correlations between TIFF and JPEG images were very high (intraclass correlation coefficients >0.8), especially for low compression JPEG images obtained by capture, regardless of the variable considered. The only exception consisted in the use of high compression JPEG files for semiautomatic microvessel density counts, which resulted in intraclass correlation coefficients of <0.7. Nonetheless, even then, interconversion between TIFF and JPEG values could be successfully achieved using prediction models established by linear regression. Image compression does not seem to significantly compromise the accuracy of angiogenesis quantitation in the ovarian epithelial tumors, although low compression JPEG images should always be preferred over high compression ones.