Computer-aided diagnosis with textural features for breast lesions in sonograms

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Oncology

Background:

• Computer-aided diagnosis (CAD) systems enhance diagnostic accuracy in medical imaging.
• Accurate classification of breast tumors is crucial for effective treatment.
• Ultrasound (US) imaging is a common modality for breast lesion assessment.

Purpose of the Study:

• To develop and evaluate a CAD system utilizing texture analysis for classifying breast tumors in ultrasound images.
• To assess the system's ability to differentiate between benign and malignant breast lesions.
• To determine the clinical utility of the developed CAD system as a second opinion tool.

Main Methods:

• A dataset of 1020 ultrasound sonograms from 255 patients was analyzed.
• Six textural features were extracted from region of interest (ROI) subimages.
• Principal Component Analysis (PCA) was employed for feature dimension reduction, followed by image retrieval for classification.
• Performance was evaluated using k-fold cross-validation (k=10) and receiver operating characteristic (ROC) curves.

Main Results:

• The proposed CAD system demonstrated a satisfactory classification ability for breast tumors using textural information.
• The area under the ROC curve (A(Z)) for the system was 0.925±0.019.
• The system effectively differentiated benign from malignant breast tumors.

Conclusions:

• The developed CAD system shows significant potential for clinical application.
• It provides a reliable second opinion for differentiating benign from malignant breast tumors.
• The integration of texture analysis and PCA in CAD systems is effective for breast tumor classification.