From Machine Learning to Ensemble Approaches: A Systematic Review of Mammogram Classification Methods

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Breast cancer is a leading cause of mortality in women, emphasizing the need for improved diagnostic tools.
• Early detection and accurate classification are critical for enhancing treatment efficacy and patient survival rates.

Purpose of the Study:

• To review and compare machine learning (ML), deep learning (DL), and hybrid/ensemble models for breast cancer classification using mammograms.
• To evaluate the performance, strengths, and limitations of different AI approaches in computer-aided diagnosis.

Main Methods:

• Systematic literature search adhering to PRISMA guidelines, including 50 studies from 2018-2025.
• Analysis of models based on mammogram datasets, focusing on preprocessing, feature extraction, optimization, and classification performance.
• Comparative evaluation of ML, DL, and hybrid model architectures.

Main Results:

• Machine learning (ELM) and deep learning (Vision Transformers) achieved 100% accuracy in binary classification tasks.
• Hybrid models like IEUNet++ demonstrated high accuracy (99.87%) and robust multi-class classification capabilities.
• ML and DL models often require extensive preprocessing and feature engineering, unlike hybrid approaches.

Conclusions:

• Hybrid models offer a promising balance of high accuracy, robustness, and efficiency for multi-class breast cancer classification.
• Future research should focus on developing AI solutions that integrate accuracy, interpretability, and resource efficiency for clinical application.
• Advancements in AI-driven classification systems are vital for supporting early breast cancer detection and improving patient outcomes.