An Adaptive Multi-Scale Multi-Expert Hybrid Deep Learning Framework for Explainable Breast Cancer Histopathological Image Classification

Abstract

Breast cancer is one of the leading causes of cancer-related mortality among women worldwide, making accurate histopathological image classification essential for early diagnosis and effective treatment planning. However, most existing deep learning approaches rely on single-scale feature extraction or a single net-work architecture, limiting their ability to simultaneously capture fine-grained cellular details and the global tissue context. To address these challenges, this study proposes an Adaptive Multi-Scale Multi-Expert Hybrid Deep Learning Framework for Explainable breast cancer histopathological image classification. The proposed framework employs multi-scale image representations to preserve morphological information at different spatial resolutions, while EfficientNetB7 extracts discriminative local features, and Vision Transformer (ViT-B/16) learns long-range contextual dependencies. An adaptive component dynamically learns expert fusion weights using a multilayer perceptron (MLP), enabling input specific feature integration instead of conventional fixed weights feature fusion. An adaptive multilayer perceptron (MLP) based fusion mechanism dynamically integrates complementary features from both expert networks, followed by an attention refinement module to enhance the diagnostically significant representations. The framework was evaluated on the publicly available BreakHis dataset, which comprises eight breast cancer histopathological subtypes. The experimental results demonstrate an accuracy of 98.4%, precision of 98.3%, recall of 98.4%, F1-score of 98.3 % and area under the receiver operating characteristic curve (AUC) of 99.7%, outperforming several state-of-the-art methods. Furthermore, Grad-CAM provides interpretable visual explanations by highlighting diagnostically relevant tissue regions, enhancing the transparency and clinical applicability of the proposed computer-aided diagnostic system.