Abstract:Fluorescence in situ hybridization (FISH) is widely regarded as the gold standard for evaluating human epidermal growth factor receptor 2 (HER2) status in breast cancer; however, it poses challenges such as the need for specialized training and issues related to signal degradation from dye quenching. Silver-enhanced in situ hybridization (SISH) serves as an automated alternative, employing permanent staining suitable for bright-field microscopy. Determining HER2 status involves distinguishing between "Amplified" and "Non-Amplified" regions by assessing HER2 and centromere 17 (CEN17) signals in SISH-stained slides. This study is the first to leverage deep learning for classifying Normal, Amplified, and Non-Amplified regions within HER2-SISH whole slide images (WSIs), which are notably more complex to analyze compared to hematoxylin and eosin (H&E)-stained slides. Our proposed approach consists of a two-stage process: first, we evaluate deep-learning models on annotated image regions, and then we apply the most effective model to WSIs for regional identification and localization. Subsequently, pseudo-color maps representing each class are overlaid, and the WSIs are reconstructed with these mapped regions. Using a private dataset of HER2-SISH breast cancer slides digitized at 40× magnification, we achieved a patch-level classification accuracy of 99.9% and a generalization accuracy of 78.8% by applying transfer learning with a Vision Transformer (ViT) model. The robustness of the model was further evaluated through k-fold cross-validation, yielding an average performance accuracy of 98%, with metrics reported alongside 95% confidence intervals to ensure statistical reliability. This method shows significant promise for clinical applications, particularly in assessing HER2 expression status in HER2-SISH histopathology images. It provides an automated solution that can aid pathologists in efficiently identifying HER2-amplified regions, thus enhancing diagnostic outcomes for breast cancer treatment.

Accurate and reproducible invasive breast cancer detection in whole-slide images: A Deep Learning approach for quantifying tumor extent

Computer-aided Detection of Squamous Carcinoma of the Cervix in Whole Slide Images

Multicenter automatic detection of invasive carcinoma on breast whole slide images

Deep Learning for Identifying Metastatic Breast Cancer

Deep-Learning-Based Approach in Cancer-Region Assessment from HER2-SISH Breast Histopathology Whole Slide Images

Deep Learning for Survival Analysis in Breast Cancer with Whole Slide Image Data

Overcoming the limitations of patch-based learning to detect cancer in whole slide images

DeepSDCS: Dissecting cancer proliferation heterogeneity in Ki67 digital whole slide images

Robust whole slide image analysis for cervical cancer screening using deep learning

AI-Powered Segmentation of Invasive Carcinoma Regions in Breast Cancer Immunohistochemical Whole-Slide Images

Large scale digital prostate pathology image analysis combining feature extraction and deep neural network

Sampling Based Tumor Recognition in Whole-Slide Histology Image with Deep Learning Approaches.

Ensemble Deep Learning-Based Image Classification for Breast Cancer Subtype and Invasiveness Diagnosis from Whole Slide Image Histopathology

Using deep convolutional neural networks to identify and classify tumor-associated stroma in diagnostic breast biopsies

Comparison of a machine and deep learning model for automated tumor annotation on digitized whole slide prostate cancer histology

Deep learning-based breast cancer grading and survival analysis on whole-slide histopathology images

Breast Cancer Classification: A Deep Learning Approach for Digital Pathology

An Artificial Intelligent System for Prostate Cancer Diagnosis in Whole Slide Images

Determining breast cancer biomarker status and associated morphological features using deep learning

Deep Interactive Learning-based ovarian cancer segmentation of H&E-stained whole slide images to study morphological patterns of BRCA mutation

Automated Diagnosis of Lymphoma with Digital Pathology Images Using Deep Learning