Evaluating deep learning models for breast cancer classification: a comparative study

2024-08-29

Medical Imaging 2025: Digital and Computational Pathology

AI Summary

This study addresses fine-grained classification of breast cancer histopathological images. We systematically evaluate eight state-of-the-art deep learning architectures—including ResNet-50, DenseNet-121, and Vision Transformer (ViT)—on a large-scale dataset of 277,000 image patches. To our knowledge, this is the first empirical demonstration that ViT substantially outperforms CNN-based models in this domain, achieving 94.0% validation accuracy versus a maximum of 91.2% for CNNs—highlighting the superiority of self-attention mechanisms in modeling subtle morphological textures and spatial relationships among minute lesions. Methodologically, we integrate transfer learning with multi-strategy data augmentation to enhance generalization and robustness. Our work provides reproducible, deployable guidance for optimal model selection in AI-assisted early breast cancer diagnosis, thereby advancing the clinical translation of Transformer architectures in digital pathology.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Deep Neural Architectures and Foundation Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

This study evaluates the effectiveness of deep learning models in classifying histopathological images for early and accurate detection of breast cancer. Eight advanced models, including ResNet-50, DenseNet-121, ResNeXt-50, Vision Transformer (ViT), GoogLeNet (Inception v3), EfficientNet, MobileNet, and SqueezeNet, were compared using a dataset of 277,524 image patches. The Vision Transformer (ViT) model, with its attention-based mechanisms, achieved the highest validation accuracy of 94%, outperforming conventional CNNs. The study demonstrates the potential of advanced machine learning methods to enhance precision and efficiency in breast cancer diagnosis in clinical settings.

Problem

• Evaluating deep learning models for breast cancer classification
• Comparing eight advanced models on histopathological image datasets
• Vision Transformer achieves highest accuracy for clinical diagnosis

Innovation

• Compares eight deep learning models for cancer classification
• Vision Transformer achieves highest accuracy with attention mechanisms
• Demonstrates machine learning potential in clinical cancer diagnosis