Immunohistochemical HER2 Recognition and Analysis of Breast Cancer Based on Deep Learning

Abstract

Breast cancer is one of the common malignant tumors in women. It seriously endangers women’s life and health. The human epidermal growth factor receptor 2 (HER2) protein is responsible for the division and growth of healthy breast cells. The overexpression of the HER2 protein is generally evaluated by immunohistochemistry (IHC). The IHC evaluation criteria mainly includes three indexes: staining intensity, circumferential membrane staining pattern, and proportion of positive cells. Manually scoring HER2 IHC images is an error-prone, variable, and time-consuming work. To solve these problems, this study proposes an automated predictive method for scoring whole-slide images (WSI) of HER2 slides based on a deep learning network. A total of 95 HER2 pathological slides from September 2021 to December 2021 were included. The average patch level precision and f1 score were 95.77% and 83.09%, respectively. The overall accuracy of automated scoring for slide-level classification was 97.9%. The proposed method showed excellent specificity for all IHC 0 and 3+ slides and most 1+ and 2+ slides. The evaluation effect of the integrated method is better than the effect of using the staining result only.

Deep Neural Networks for HER2 Grading of Whole Slide Images with Subclasses Levels

“…These results may be attributed to the network's lack of access to supplementary texture or tumour microenvironment data, instead relying on the same information presented at a varying scale. Finally, when we apply the ResNet34 network used in Che et al [20] to our dataset, the results are comparable to those previously reported, with an average accuracy of 94%. Thus, the proposed method outperforms other existing methods, as proven by the obtained results with an average accuracy of 97%.…”

“…In this specific case, a 9.76% of the tissue is reported as 3+, so the system evaluates the slide as a 3+ positive, providing a warning to check for equivocal tissue regions for further analysis by the pathologist. Comparing the outcomes achieved using the proposed method to those cited in the current literature [13][14][15][16][17][18][19][20], a comparison can be drawn with the outcomes of five of the mentioned methods, all of which used the AIDPATH database [14][15][16][17]19]. Among these, Cordeiro et al [14] reported the highest accuracy rate of 94%.…”

“…The remaining networks, ResNet and DenseNet were selected based on their ability to offer new attributes, achieving good results while maintaining a trade-off between computational time and workload. In addition, ResNet has been frequently used for HER2 scoring in breast cancer [18,20,22] and more recently for predicting the HER2 status in oesophageal cancer [32]. Moreover, DenseNet has also been employed recently for HER2 scoring in breast cancer [5].…”

“…Finally, the work of Che et al [20] use deep learning to automatically assess breast cancer Her2 WSI scores on 95 IHC section images with labelled tumour areas. The proposed method has three steps.…”