Classification of breast cancer using hybridization version of walrus and particle swarm optimization algorithm

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Optimization Algorithms

Background:

• Breast cancer (BC) is a leading cause of cancer mortality in women globally.
• Early and accurate classification of mammographic lesions is critical for improving patient prognosis.
• Current computer-aided detection systems require enhanced classification accuracy.

Purpose of the Study:

• To introduce a novel hybrid optimization algorithm, Walrus Particle Swarm Optimisation (WPS).
• To apply WPS for tuning hyperparameters of a Convolutional Neural Network enhanced with Swapping of Proficiency (CNN-SP).
• To evaluate the performance of the WPS-CNN-SP model for breast cancer image classification.

Main Methods:

• Developed a hybrid Walrus Particle Swarm Optimisation (WPS) algorithm, integrating Particle Swarm Optimisation (PSO) and Walrus Optimiser (WO).
• Utilized WPS to optimize the hyperparameters of a CNN-SP model.
• Tested the WPS-CNN-SP model on benchmark functions and two mammographic datasets (CBIS-DDSM and MIAS).

Main Results:

• WPS demonstrated a strong exploration-exploitation balance, achieving near-optimal solutions on benchmark functions.
• The WPS-CNN-SP model significantly outperformed recent works on both CBIS-DDSM and MIAS datasets.
• Achieved high performance metrics, including Area Under the Curve (AUC) up to 99.76% and Accuracy up to 98.99%.

Conclusions:

• The proposed WPS algorithm is a fast and reliable optimizer for computer-aided breast cancer screening.
• WPS-CNN-SP shows significant potential for enhancing the accuracy and efficiency of mammographic lesion classification.
• This approach can contribute to improved patient prognosis through earlier and more accurate breast cancer detection.