Ovarian tumor characterization using 3D ultrasound

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Ovarian cancer is a leading cause of gynecological cancer deaths.
• Distinguishing benign from malignant ovarian tumors preoperatively is challenging.
• Inconclusive diagnostic tests lead to unnecessary surgeries, increasing patient anxiety and healthcare costs.

Purpose of the Study:

• To develop an automated Computer-Aided Diagnostic (CAD) system for classifying ovarian tumors using ultrasound images.
• To improve the accuracy of preoperative tumor assessment and reduce unnecessary surgical interventions.

Main Methods:

• Extraction of texture features using Local Binary Patterns (LBP) and Laws Texture Energy (LTE).
• Training a Support Vector Machine (SVM) classifier with a Radial Basis Function (RBF) kernel.
• Validation using a dataset of 1000 benign and 1000 malignant ovarian ultrasound images.

Main Results:

• Achieved a high classification accuracy of 99.9%.
• The novel combination of 16 texture-based features effectively quantified subtle image differences.
• Developed the Ovarian Cancer Index (OCI) for transparent adjunct classification.

Conclusions:

• The proposed CAD technique demonstrates high accuracy and effectiveness in classifying ovarian tumors.
• The system is cost-effective, automated, easily deployable, and user-comprehensible.
• The OCI provides physicians with a transparent tool for improved ovarian tumor diagnosis.