A novel embedded kernel CNN-PCFF algorithm for breast cancer pathological image classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Early breast cancer screening via image recognition significantly improves patient survival rates.
• Breast cancer pathological images are crucial for medical diagnosis and clinical research.
• Deep learning, particularly Convolutional Neural Networks (CNNs), is widely used for breast cancer image classification.

Purpose of the Study:

• To address the high dimensionality and computational complexity issues in CNN-based breast cancer image classification.
• To propose a novel algorithm for effective feature fusion and dimensionality reduction.
• To enhance the performance of CNN architectures and subsequent classifiers in breast cancer pathological image analysis.

Main Methods:

• A novel embedded kernel CNN principal component feature fusion (CNN-PCFF) algorithm was developed.
• Kernel functions were embedded into principal component analysis to create multi-kernel principal components.
• Multi-kernel principal component analysis fused high-dimensional CNN features into representative kernel principal components for dimensionality reduction.

Main Results:

• The proposed CNN-PCFF algorithm effectively reduced feature dimensionality while retaining important information.
• Experimental analysis on public datasets demonstrated improved performance of mainstream CNN architectures and classifiers.
• The algorithm proved to be an effective tool for breast cancer pathological image classification.

Conclusions:

• The CNN-PCFF algorithm offers an effective solution for dimensionality reduction in breast cancer image analysis.
• This method enhances the performance of deep learning models for improved diagnostic accuracy.
• The proposed approach contributes to advancing early breast cancer detection and clinical research.