Finite Element Analysis of Identifying Breast Cancer Tumor Grades Through Frequency Spectral Variation of High-Frequency Ultrasound

Abstract

Ultrasound analysis is an instantaneous characterization tool to evaluate microstructural inhomogeneity. In this study, computational high-frequency ultrasound analysis was conducted to characterize histological features of malignant breast tissue. A high-frequency ultrasound signal was sent through the soft tissue model in a through-transmission manner. Histological features of the soft tissue were categorized as cell shape, nuclear pleomorphism, and malignant cell density. The design of experiment was created by combining various levels of histological features of tumor tissue. Transmitted ultrasound frequency spectrums from all combinations of histological features were analyzed in terms of peak density and mean peak to valley distance (MPVD) parameters. For the circular-shaped cell model, peak density and MPVD responded with increasing and decreasing trends respectively while the malignant histological features became gradually dominant. For the elliptical-shaped cell model, only peak density was effective to establish a relationship with the histological features. It was observed that added malignant cells had more contribution to the response parameters than nuclear pleomorphism. Furthermore, the frequency spectrum patterns from all histological combinations were evaluated to find further information about malignant features.

Keywords

• Histology-based modeling
• acoustic scattering
• peak density
• high-frequency ultrasound
• tumor grading
• breast cancer