Classification of breast cancer histopathological images using interleaved DenseNet with SENet (IDSNet)

Area of Science:

• Medical Imaging
• Computational Pathology
• Artificial Intelligence in Oncology

Background:

• Accurate breast cancer (BC) classification from histological images is crucial for effective treatment.
• Existing deep learning models face challenges in effectively utilizing complex feature information for BC diagnosis.

Purpose of the Study:

• To develop a novel convolutional neural network (CNN) architecture for improved classification of benign and malignant breast cancer (BC).
• To enhance feature information delivery and utilization within the CNN framework for more accurate BC detection.

Main Methods:

• Proposed a novel CNN architecture by integrating DenseNet with squeeze-and-excitation (SENet) modules.
• Utilized the public domain BreakHis dataset for comprehensive experimental validation.
• Compared the performance against existing state-of-the-art CNN methods.

Main Results:

• The proposed DenseNet-SENet framework demonstrated significantly improved accuracy in classifying benign and malignant breast cancer.
• The integration of SENet modules enhanced the feature representation capabilities of the DenseNet architecture.
• Achieved superior performance compared to other leading CNN approaches on the BreakHis dataset.

Conclusions:

• The novel CNN architecture offers a promising advancement for automated breast cancer classification in histopathology.
• The DenseNet-SENet hybrid model effectively captures and utilizes intricate features for enhanced diagnostic accuracy.
• This approach holds potential for improving the efficiency and reliability of breast cancer diagnosis.