A computational classification method of breast cancer images using the VGGNet model

Area of Science:

• Medical Imaging
• Computational Biology
• Machine Learning

Background:

• Breast cancer is a leading global health concern, necessitating accurate detection and classification.
• Existing computational methods like KNN, SVM, MLP, DT, and genetic algorithms have limitations in breast cancer diagnosis.
• Current deep learning models, such as VGGNet-16, face overfitting challenges with breast cancer datasets.

Purpose of the Study:

• To address the overfitting issue in VGGNet-16 for breast cancer classification.
• To propose a new, optimized convolutional neural network model, VGGNet-12.
• To evaluate the performance of the VGGNet-12 model against other CNN architectures.

Main Methods:

• Developed a novel VGGNet-12 model by reducing layers from the VGGNet-16 architecture.
• Trained and tested the VGGNet-12 model using a breast cancer dataset.
• Compared the classification performance of VGGNet-12 with existing CNN and LeNet models.

Main Results:

• The proposed VGGNet-12 model demonstrated reduced overfitting compared to VGGNet-16.
• VGGNet-12 achieved enhanced simulation results in breast cancer classification.
• Experimental findings confirmed the effectiveness of VGGNet-12 in classifying breast cancer characteristics.

Conclusions:

• The VGGNet-12 model offers a promising solution for accurate breast cancer classification.
• Optimizing convolutional neural network architectures, like VGGNet-12, can overcome limitations of existing models.
• This research contributes to advancing computational techniques for improved breast cancer detection.