Digital image analysis in breast cancer: an example of an automated methodology and the effects of image compression

In the current practice of pathology, the evaluation of immunohistochemical (IHC) markers represents an essential tool. The manual quantification of these markers is still laborious and subjective, and the use of computerized systems for digital image (DI) analysis has not yet resolved the problems of nuclear aggregates (clusters). Furthermore, the volume of DI storage continues to be an important problem in computer-assisted pathology. In the present study we have developed an automated procedure to quantify IHC nuclear markers in DI with a high level of clusters. Furthermore the effects of JPEG compression in the image analysis were evaluated. The results indicated that there was an agreement with the results of both methods (automated vs. manual) in almost 90% of the analyzed images. On the other hand, automated count differences increase as the compression level increase, but only in images with a high number of stained nuclei (>nuclei/image) or with high area cluster (>25μm2). Some corrector factors were developed in order to correct this count differences. In conclusion, the proposed automated procedure is an objective, faster than manual counting and reproducible method that has more than 90% of similarity with manual count. Moreover, the results demonstrate that with correction factors, it is possible to carry out unbiased automated quantifications on IHC nuclear markers in compressed DIs.