A Feature Fusion Attention-Based Deep Learning Algorithm for Mammographic Architectural Distortion Classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Radiology

Background:

• Architectural Distortion (AD) is a critical mammographic abnormality, often challenging to detect in dense breast tissue due to subtle presentations and heterogeneous patterns.
• Existing detection methods face limitations in sensitivity and efficiency, particularly with complex textural features and background noise in digital mammograms.

Purpose of the Study:

• To develop and evaluate a novel feature fusion-based Vision Transformer (ViT) attention network integrated with VGG-16 for improved accuracy and efficiency in detecting architectural distortion (AD) in mammograms.
• To enhance the robustness of AD classification by addressing limitations in texture analysis, background boundary detection, and deep neural network performance.

Main Methods:

• Implementation of a hybrid deep learning model combining a Vision Transformer (ViT) attention network with VGG-16 architecture.
• Utilizing feature fusion techniques to integrate diverse image features for comprehensive analysis.
• Experimental validation on the PINUM and DDSM mammography datasets.

Main Results:

• The proposed model achieved state-of-the-art performance, outperforming eight existing deep learning models.
• Achieved high performance metrics on the PINUM dataset: 0.97 sensitivity, 0.92 F1-score, 0.93 precision, 0.94 specificity, and 0.96 accuracy.
• Demonstrated strong results on the DDSM dataset: 0.93 sensitivity, 0.91 F1-score, 0.94 precision, 0.92 specificity, and 0.95 accuracy.

Conclusions:

• The novel feature fusion-based ViT-VGG-16 model significantly enhances the accuracy and efficiency of architectural distortion detection in digital mammograms.
• This approach shows great potential for computer-aided diagnosis of breast cancer, particularly in resource-limited settings.
• The method offers a promising tool for improving breast cancer screening and early intervention globally.