Multi-View Attention-Guided Multiple Instance Detection Network for Interpretable Breast Cancer Histopathological Image Diagnosis

Abstract

Deep learning approaches have demonstrated significant progress in breast cancer histopathological image diagnosis. Training an interpretable diagnosis model using high-resolution histopathological image is still challenging. To alleviate this problem, a novel multi-view attention-guided multiple instance detection network (MA-MIDN) is proposed. The traditional image classification problem is framed as a weakly supervised multiple instance learning (MIL) problem. We first divide each histopathology image into instances and form a corresponding bag to fully utilize high-resolution information through MIL. Then a new multiple-view attention (MVA) algorithm is proposed to learn attention on the instances from the image to localize the lesion regions in this image. A MVA-guided MIL pooling strategy is designed for aggregating instance-level features to obtain bag-level features for the final classification. The proposed MA-MIDN model performs lesion localization and image classification, simultaneously. Particularly, we train the MA-MIDN model under the deep mutual learning (DML) schema. This transfers DML to a weakly supervised learning problem. Three public breast cancer histopathological image datasets are chosen to evaluate classification and localization results. The experimental results demonstrate that the MA-MIDN model is superior to the latest baselines in terms of diagnosis accuracy, AUC, Precision, Recall, and F1. Notably, it achieves better localization results without compromising classification performance, thereby proving its higher practicality. The code for the MA-MIDN model is available at https://github.com/lcxlcx/MA-MIDN.INDEX TERMS breast cancer diagnosis; multiple instance learning; multi-view attention; diagnosis interpretability; deep mutual learning.

Artificial intelligence-assisted stimulated Raman molecular cytology enables accurate and rapid intraoperative assessment of the surgical margin in pancreatic ductal adenocarcinoma

“…Our scheme of MIL using cell vectors maintains the single-cell interpretability. Current artificial intelligence-assisted digital histology 41 and cytology 60 using MIL typically comprises image patch embedding, data aggregation and prediction, and the diagnosis result is then interpreted with the attention mechanism 42 or the saliency map 43 . These methods guide the attribution of model predictions to particular regions on the image patches.…”

A Transfer Learning-Based Deep Learning Model for Automated Breast Cancer Identification in Mammograms

“…Guangli Li et al [24] The proposed approach has the potential to enhance breast cancer analysis, as evidenced by its accuracy of 86% on a variety of datasets, which is better than both earlier deep learning techniques and regular methods.…”

“…Guangli Li et al [24] proposed a new model called Multi-View Attention-Guided Multiple Instance Detection Networks, which divides each histopathology image into instances to fully use high-resolution information. The multiple-view attention (MVA) approach was established by the algorithm to locate and localize lesion patches within the image by focusing on certain instances.…”

A Transfer Learning-Based Deep Learning Model for Automated Breast Cancer Identification in Mammograms

“…The model achieved better localization results without sacri cing classi cation performance, making it highly practical. Amin Ul Haq et al [25] proposed supervised and unsupervised techniques for related feature selection from a data set and determined by PCA algorithms and the Relief algorithms. These features are closely connected with accurate breast cancer diagnosis.…”