Anomaly Detection of Breast Cancer Using Deep Learning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast cancer remains a leading cause of death for women globally.
• Early detection and treatment are crucial for improving patient survival rates and reducing healthcare costs.
• Imbalanced datasets are a common challenge in medical AI research.

Purpose of the Study:

• To propose an efficient and accurate deep learning framework for detecting breast abnormalities (benign and malignant).
• To address the challenge of imbalanced data in medical image analysis.
• To develop a system capable of recognizing anomalies by learning from normal data.

Main Methods:

• A two-stage framework involving image pre-processing and feature extraction.
• Utilizing a pre-trained MobileNetV2 model for efficient feature extraction.
• Employing a single-layer perceptron for the final classification step.

Main Results:

• The framework demonstrated high efficiency and accuracy in anomaly detection, with Area Under the Curve (AUC) ranging from 81.40% to 97.36%.
• The proposed method outperformed existing relevant works in detecting breast abnormalities.
• The framework successfully addressed limitations of previous approaches in handling imbalanced medical data.

Conclusions:

• The developed deep learning framework offers a promising solution for accurate and efficient breast cancer anomaly detection.
• The approach effectively handles imbalanced datasets, a significant hurdle in medical AI.
• This work contributes to advancing early breast cancer diagnosis through artificial intelligence.