Boosted EfficientNet: Detection of Lymph Node Metastases in Breast Cancer Using Convolutional Neural Networks

Area of Science:

• Medical Imaging
• Computational Biology
• Artificial Intelligence

Background:

• Accurate detection of lymph node metastasis in breast cancer is crucial for effective treatment.
• Deep learning models, particularly Convolutional Neural Networks (CNNs), show promise in image-based cancer diagnosis.
• Existing models may face challenges with small resolution datasets and retaining critical image features.

Purpose of the Study:

• To enhance the performance of the EfficientNet model for classifying metastatic breast cancer (MBC).
• To introduce a novel data augmentation technique, Random Center Cropping (RCC), for small resolution images.
• To integrate attention and Feature Fusion (FF) mechanisms to improve feature extraction in CNNs.

Main Methods:

• Utilized CNNs for the detection and classification of lymph node metastasis in breast cancer.
• Developed and applied Random Center Cropping (RCC) to preserve image resolution and central features.
• Integrated attention and Feature Fusion (FF) mechanisms into the EfficientNet architecture.

Main Results:

• The proposed methods significantly improved the performance of basic CNN architectures.
• The best-performing model achieved an accuracy of 97.96% ± 0.03% and an Area Under the Curve (AUC) of 99.68% ± 0.01% on the RPCam dataset.
• All four technological improvements contributed to boosting the original EfficientNet's performance.

Conclusions:

• This study is the first to explore EfficientNet for MBC classification, comparing it with state-of-the-art CNN models.
• The novel RCC data augmentation method enhances data enrichment for small resolution datasets.
• The integrated enhancements offer a promising direction for deep learning-based image diagnosis in oncology.