Spiculation Analysis of Breast Tumors on 3D Ultrasound

Architectural distortion is a common appearance of breast cancer, its stellate feature is similar to spiculation with subtle difference. Since these stellate features are nonpalpable but potential for malignancy, it is valuable to be able to characterize it on ultrasound imaging. Although screening identified architectural distortion or spiculation is commonly recognized as a manifestation of aggressive cancer, this type of radiological appearance has also frequently found to be a good prognostic factor in invasive breast cancer on mammographic screening. With the hypothesis that the corresponding correlation is also applicable for breast ultrasound, this study examines the correlation between histological grade and stellate feature on coronal section of three-dimensional (3D) ultrasound imaging from 150 malignant tumors. A novel method for image enhancement is proposed to stabilize the feature extraction, and the feature of stellate distortion can then be detected by Laplacian of Gaussian (LoG) filter. Tumor grades were analyzed within the index space formed from extracted features. The results indicate that the proposed features can be utilized as potential predictors for tumor grades.