Abstract
Magnetomotive ultrasound (MMUS) is an emerging technique to image super-paramagnetic iron oxide nanoparticles (SPIOs) in tissues. An oscillating external magnetic field is applied to induce magnetomotion of the SPIOs, and then sub-wavelength motion tracking is performed to track motion sources for the mapping of the SPIO distribution because generally such magnetomotion is too small to be visible directly in conventional B-mode imaging. Our previous work attempted to use frequency-based filters to perform motion magnification and visualize motion patterns from the SPIOs in B-mode imaging. However, in many MMUS applications, the frequencies of the magnetomotion signals from the SPIOs and motion noises are overlapped, and thus the two cannot be separated simply by frequency-based filters, resulting in poor performance of motion magnification. To solve this problem, we propose a novel principle-component-analysis (PCA) based motion magnification method for direct B-mode visualization of SPIO magnetomotion in MMUS. The proposed method can reveal originally invisible sub-wavelength magnetomotion in B-mode images without the need of sub-wavelength motion tracking. PCA-based filtering with different order selection and eigen-space weighting enables the extraction and magnification of the target motions which are previously non-separable from motion noises by using frequency-based filters. MMUS simulation and experimental results (not shown here) show that it is easy to localize the magnetomotion sources, i.e., the SPIOs, and visualize the shear wave propagation induced by magnetic excitation in B-mode images even when the magnetomotion is corrupted by the frequency non-separable tissue motion. Overall, the PCA-based motion magnification technique offers a more generalized B-mode representation of the originally invisible magnetomotion and magnetic-excitation-induced shear wave propagation, waiving the need of sub-wavelength motion tracking.