Early detection of breast cancer, one of the important causes of cancer morbidity worldwide , can significantly improve the survival probability of the patients [2, 3]. Identifying molecules which regulate the growth of normal and transformed breast epithelium by means of immunohistochemistry, has been used as adjuncts for diagnostic, prognostic, and predictive decision-making . The most established validated molecular markers predicting responsiveness of patients for molecular targeted therapy are estrogen (ER) and progesterone (PR) receptors  and the type-2 human epidermal growth factor receptor (HER2).
Large scale randomized clinical trials proved that patients with double positive (ER/PR) status are likely to benefit from endocrine/hormonal therapy. According to the recent ASCO-CAP guidelines, ER and PR status should be determined in all invasive breast cancers and recurrences . However, the reliability of assay results depends on both the reproducibility of assay performance and its interpretation [7, 8].
Standardized protocols used by automated immunostainers set up high standards in test reproducibility. Whole slide digitalization, supported by dedicated software tools allows image object quantization based on colour and intensity segmentation for unbiased analysis of immunostaining results .
Several validation studies of image analysis applications for the detection of ER and PR in breast cancer specimens have been published [10–16] and results provided strong concordance between the pathologist's manual assessment of slides and scoring performed by the different software applications.
Our survey was performed to test the effectiveness of two connected software application: Pannoramic™ Viewer v. 1.14 (hereinafter PV) and NuclearQuant application for PV v. 1.13 (hereinafter NQ) both manufactured by 3DHISTECH Ltd. (Budapest, Hungary). PV enables the visualization of digital slides, users being able to inspect and annotate (i.e. select certain regions of an image) digital slides. These editable (i.e. possibility for deletion, rename) annotations can be rectangular and/or free hand. NQ is an image-analysis software connected to PV, suitable for unbiased automated analysis of digital image objects based on colour, intensity and size. It detects and separates nuclei-shaped connected pixel sets (e.g. immunolabelled cell nuclei) on microscopic digital slides (*.mrxs file format) created using one of the digital slide scanners manufactured by 3DHISTECH Ltd. (i.e. Pannoramic DESK, Pannoramic
MIDI, Pannoramic SCAN 150). For image standardization  Wallis image filters are used, to compensate rarely perceptible but possibly occurring local intensity deviations arising from unsuitable luminance and optical aberrations. NQ uses a "colour deconvolution" algorithm for the scoring process, which decomposes the RGB image into greyscale intensity images that represent the staining or dye concentration maps individually for each stain. Nuclei scoring is based on the intensity of connected pixel sets measured on the 3,3'-Diaminobenzidine (DAB) intensity image only. The nuclei detection is performed on the intensity normalized greyscale image, based on the morphometrical parameters, such as optimal roundness, average density and proper contrast of the intensity at the boundary. Scores are calculated for each detected object, based on the average intensity of the corresponding pixels of the intensity image. Users may filter object by size or shape, and can separate joint objects. The information necessary for the algorithm to be used is not saved on the image itself but are stored in Microscopic Image Segmentation Profile files (*.misp). The application can be run on whole slides or annotations, and the detected objects can be viewed, reclassified, relocated and visualized. Both research use applications are intended to support in vitro diagnostic decision-making, aiding pathologists in the detection, scoring, classification and counting of cells of interests.
The aims of our validation study were: 1) calibration of the algorithm for ER and PR detection built-in NQ on immunostained slides after whole-slide digitalization; and 2) assessment of equivalence of the semi-automated detection using the software application with the manual scoring of digital slides.