ARFI displacement imaging provides a qualitative assessment of tissue stiffness by tracking micrometer displacements “on-axis” to the ARF beam and is diagnostically relevant for various diseases. However, displacement estimation is challenged by physiological motion, reverberation clutter, and noise, all of which degrade the image quality. We present an adaptive Blind-Source-Separation method via principal component analysis (BSS-PCA) that is applied to the three-dimensional ARFI data to isolate the ARF-induced displacements while suppressing noise. Displacement profiles were filtered by finding the optimal eigenvectors that maximize the Contrast-to-noise ratio (CNR) in the peak displacement image. BSS-PCA was evaluated on simulated six heterogeneous materials (inclusion: elasticity (6 and 12 kPa), viscosity (0.3 and 1.5 Pa·s), and radius (1.5 and 3 mm), and background:elasticity 3 kPa), with echo SNR from 0 to 35 dB using LSDYNA3D and Field II simulation. CNR was significantly higher when BSS-PCA was applied compared to the case without BSS-PCA, irrespective of material properties, echo SNR, and ROI size, with CNR gains of up to 38% at 0 SNR. The average optimal eigenvectors varied from 2.0 to 24.5 when the SNR varied from 0 to 35 dB. Future studies will validate this method in experimental settings and in vivo imaging.
Chen et al. (Wed,) studied this question.