Digital image analysis of membrane connectivity is a robust measure of HER2 immunostains

Breast Cancer Res Treat. 2012 Feb;132(1):41-9. doi: 10.1007/s10549-011-1514-2. Epub 2011 Apr 22.


The purpose of this study was to develop and validate a new software, HER2-CONNECT(TM), for digital image analysis of the human epidermal growth factor receptor 2 (HER2) in breast cancer specimens. The software assesses immunohistochemical (IHC) staining reactions of HER2 based on an algorithm evaluating the cell membrane connectivity. The HER2-CONNECT algorithm was aligned to match digital image scorings of HER2 performed by 5 experienced assessors in a training set and confirmed in a separate validation set. The training set consisted of 167 breast carcinoma tissue core images in which the assessors individually and blinded outlined regions of interest and gave their HER2 score 0/1+/2+/3+ to the specific tumor region. The validation set consisted of 86 core images where the result of the automated image analysis software was correlated to the scores provided by the 5 assessors. HER2 fluorescence in situ hybridization (FISH) was performed on all cores and used as a reference standard. The overall agreement between the image analysis software and the digital scorings of the 5 assessors was 92.1% (Cohen's Kappa: 0.859) in the training set and 92.3% (Cohen's Kappa: 0.864) in the validation set. The image analysis sensitivity was 99.2% and specificity 100% when correlated to FISH. In conclusion, the Visiopharm HER2 IHC algorithm HER2-CONNECT(TM) can discriminate between amplified and non-amplified cases with high accuracy and diminish the equivocal category and thereby provides a promising supplementary diagnostic tool to increase consistency in HER2 assessment.

Publication types

  • Validation Study

MeSH terms

  • Algorithms
  • Area Under Curve
  • Breast Neoplasms / diagnosis*
  • Breast Neoplasms / metabolism
  • Carcinoma, Ductal, Breast / diagnosis*
  • Carcinoma, Ductal, Breast / metabolism
  • Cell Membrane / metabolism*
  • Female
  • Humans
  • Image Processing, Computer-Assisted*
  • Immunohistochemistry
  • In Situ Hybridization, Fluorescence
  • Observer Variation
  • Receptor, ErbB-2 / metabolism*
  • Sensitivity and Specificity
  • Software*
  • Tissue Array Analysis


  • ERBB2 protein, human
  • Receptor, ErbB-2