Development and evaluation of an automated, Web-based tool for quantitative assessment of fluorescence in situ hybridization

AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA 4412 Assessment of the HER2 status in breast cancer is routinely achieved by immunohistochemical analysis or in situ hybridization techniques. Quantification of fluorescence in situ hybridizations (FISH) is a laborious work, since it is most commonly done by visual counting of the hybridized markers. We have developed an automated method for quantification of HER2 fluorescence in situ hybridization. The algorithm, entitled FishJ, is based on a popular open source image manipulation tool, ImageJ, and is freely available for download. In addition to that, FishJ can be utilized through a web interface (www.webmicroscope.net/fishj) with an option of uploading FISH images for analysis. In order to evaluate the FishJ algorithm, we tested it against traditional manual counting in breast cancer specimens from 42 patients. The hybridized tumor specimens were digitized within a virtual microscopy framework. Markers for the HER2 gene and chromosome 17 centromere (CEP17) were counted both manually and automatically with our macro. The correlation coefficient between the automatic and manual HER2/CEP17 ratios was 0.985. The corresponding percent-agreement was 95% and the kappa value 0.90. The main findings of this study show that the automated determination of HER2 amplification using open-source software is possible and that the results are comparable to manual counting. In addition, the automatic counting saves sample processing time and provides possibilities to enhance inter- and intralaboratory reproducibility of results. A web-based, automated FISH quantification tool could potentially enable improved standardization of gene copy number evaluation.