DF-dRVFL: A novel deep feature based classifier for breast mass classification.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Oncology

Background:

• Breast cancer is a leading cause of mortality in the UK, necessitating early detection for effective treatment.
• Mammography is a cost-effective screening tool, but image-based analysis faces challenges with accuracy and high false positive rates.
• Computer-aided detection (CAD) systems aim to assist radiologists in analyzing mammograms for breast masses, but often struggle with accuracy and computational demands.

Purpose of the Study:

• To develop a novel and accurate computer-aided system for breast mass classification on mammograms.
• To address the limitations of existing CAD systems, specifically low accuracy and high computational power requirements.
• To improve the early detection of breast cancer through enhanced image analysis.

Main Methods:

• Development of a novel breast mass classification system named DF-dRVFL, utilizing a deep random vector functional link network.
• Evaluation of the DF-dRVFL system on the public DDSM dataset, comprising over 3500 mammographic images.
• Implementation of a five-fold cross-validation strategy to assess model performance.

Main Results:

• The best DF-dRVFL model achieved a promising average Area Under the Curve (AUC) of 0.93 and high average accuracy.
• The developed system demonstrated a significant increase in average accuracy by 0.38 compared to sole deep learning methods.
• DF-dRVFL outperformed state-of-the-art methods in breast mass classification, considering evaluation metrics and overall accuracy.

Conclusions:

• The DF-dRVFL system offers a promising advancement in computer-aided detection for breast cancer.
• This novel approach enhances accuracy in breast mass classification, potentially leading to earlier and more reliable diagnoses.
• The system's improved performance suggests a valuable tool for radiologists in breast cancer screening and analysis.