Pattern Recognition Approaches for Breast Cancer DCE-MRI Classification: A Systematic Review

Area of Science:

• Radiology
• Medical Imaging
• Machine Learning in Medicine

Background:

• Accurate breast lesion classification is crucial for effective cancer diagnosis and treatment.
• Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) offers valuable insights into lesion characteristics.
• Pattern analysis using machine learning can enhance the interpretation of DCE-MRI data.

Purpose of the Study:

• To systematically review pattern analysis approaches for classifying breast lesions using DCE-MRI.
• To evaluate the performance of various machine learning classifiers and feature sets.
• To identify the most effective methods for improving diagnostic accuracy.

Main Methods:

• Systematic review of 26 studies on breast lesion classification using DCE-MRI.
• Analysis of machine learning approaches: Artificial Neural Networks (ANN), Support Vector Machines (SVM), Linear Discriminant Analysis (LDA), Tree-Based Classifiers (TC), and Bayesian Classifiers (BC).
• Evaluation of classification performance based on dynamic, morphological, and textural features, individually and in combination.

Main Results:

• LDA and TC demonstrated the highest performance among the classifiers evaluated.
• A combination of dynamic and morphological features yielded the best diagnostic results.
• Reported sensitivities and specificities varied by classifier and feature set, with LDA achieving 96% sensitivity and 85% specificity.

Conclusions:

• Machine learning, particularly LDA and TC, significantly enhances breast lesion classification accuracy in DCE-MRI.
• Combining dynamic and morphological features is the most effective strategy for improving diagnostic performance.
• These findings support the integration of advanced pattern analysis techniques into clinical DCE-MRI interpretation.