Predicting Breast Cancer Using Hybrid Deep Learning Technique

Main Article Content

G. Vijaya, Y. Ramadevi

Abstract

Breast cancer remains one of the most pressing health challenges worldwide, necessitating advancements in diagnostic techniques to improve early detection and treatment outcomes. Despite the effectiveness of deep learning in medical imaging, there exists a gap in comprehensive evaluations of specific methodologies for breast cancer diagnosis. This study aims to develop a hybrid deep learning technique that combines Convolutional Neural Networks (CNN) and Double Deep Q-Networks (DDQN) to enhance the accuracy of breast cancer detection. Utilizing a dataset of 277,524 image patches from 162 whole-slide images categorized into IDC positive and IDC negative samples, this research addresses significant class imbalance through data augmentation and attention mechanisms. The hybrid model's performance is compared against traditional algorithms such as CNNs, VGGNet, and ResNet. Experimental results indicate that the proposed hybrid technique outperforms conventional models, achieving higher accuracy, precision, recall, and F1 score. The model effectively manages class imbalance and adapts to varying image characteristics, reinforcing its reliability in clinical applications. This study highlights the potential of integrating Q-learning with deep neural networks in medical image analysis. The findings suggest that this hybrid approach can significantly improve early breast cancer detection, paving the way for enhanced diagnostic tools and patient management strategies.

Article Details

Section
Articles