Intelligent Breast Mass Classification Approach Using Archimedes Optimization Algorithm with Deep Learning on Digital...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast cancer (BC) detection from mammograms is complex and time-consuming for human radiologists.
• Computer-aided diagnosis (CAD) systems enhance BC analysis, but require further performance improvements.
• Deep learning (DL) shows promise for early BC detection using convolutional neural networks (CNNs).

Purpose of the Study:

• To present an Intelligent Breast Mass Classification Approach using the Archimedes Optimization Algorithm with Deep Learning (BMCA-AOADL) for digital mammograms.
• To enhance breast mass classification by integrating DL models with a bio-inspired algorithm.
• To improve the accuracy and efficiency of breast cancer detection and classification.

Main Methods:

• The BMCA-AOADL technique employs median filtering (MF) for noise removal and U-Net for segmentation as pre-processing steps.
• Feature extraction is performed using the SqueezeNet model, with the Archimedes Optimization Algorithm (AOA) for hyperparameter tuning.
• Breast mass detection and classification are achieved using a deep belief network (DBN) approach.

Main Results:

• The BMCA-AOADL technique was evaluated on the MIAS dataset.
• The system demonstrated significant outcomes compared to other DL algorithms.
• A maximum accuracy of 96.48% was achieved, showcasing the effectiveness of the proposed method.

Conclusions:

• The BMCA-AOADL technique effectively classifies breast masses in digital mammograms.
• Integrating DL with bio-inspired optimization algorithms like AOA enhances CAD system performance.
• The study highlights the potential of advanced AI techniques for improving early breast cancer diagnosis.