Breast Research Group, Leeds Institute of Cancer and Pathology, St. James University Hospital, , Leeds, UK.
J Clin Pathol. 2014 Jan;67(1):72-5. doi: 10.1136/jclinpath-2013-201680. Epub 2013 Aug 28.
The aim of this study was to validate ImmunoRatio, a web-based automated image analysis application, by comparing the manual and automated analysis scores for oestrogen receptor α (ERα) in breast carcinomas. Tissue microarrays comprising 200 breast cancer cases prestained for ERα were scanned and scored manually using ImageScope viewing software. Corresponding images were then uploaded and assessed according to the web-based ImmunoRatio programme. There was excellent correlation between manual and ImmunoRatio ERα scores (Spearman correlation=0.872; p≥0.001). The manual and ImmunoRatio ERα scores showed only a moderate agreement (κ=0.421; Weighted kappa=0.874 (CI 0.839 to 0.902)), most probably due to lack of specificity of the algorithm to differentiate between cancer and non-cancer nuclei. Further development to enable differentiation of cancer and non-cancer elements should improve the specificity of the application. Our results support the use of ImmunoRatio software for analysing ERα immunohistochemistry in breast cancer tissues for the purposes of research.
本研究旨在通过比较乳腺癌中雌激素受体 α (ERα) 的手动和自动分析评分来验证基于网络的自动化图像分析应用程序 ImmunoRatio。使用 ImageScope 查看软件对包含 200 例 ERα 预染乳腺癌病例的组织微阵列进行扫描和手动评分。然后将相应的图像上传并根据基于网络的 ImmunoRatio 程序进行评估。手动和 ImmunoRatio ERα 评分之间具有极好的相关性（Spearman 相关系数=0.872；p≥0.001）。手动和 ImmunoRatio ERα 评分仅具有中等一致性（κ=0.421；加权κ=0.874（CI 0.839 至 0.902）），这很可能是由于算法缺乏区分癌和非癌核的特异性所致。进一步开发以区分癌症和非癌症元素的功能应能提高该应用程序的特异性。我们的研究结果支持在乳腺癌组织中分析 ERα 免疫组化时使用 ImmunoRatio 软件进行研究。
