Adaptive Independent Component Analysis-Based Clutter Filtering for Improved Non-Contrast Perfusion Ultrasound Imaging

Tissue motion makes perfusion ultrasound imaging without contrast difficult. We previously developed an adaptive demodulation (AD) method to correct for this motion prior to tissue filtering. Additionally, 2D space and time eigen-based clutter filtering has been shown to be superior to conventional frequency domain filtering performed only in the slow-time dimension, especially when long ensembles are used. We have shown that combining AD with singular value decomposition (SVD) filtering can improve slow blood flow detection. Here, we aim to develop and evaluate an adaptive independent component analysis (ICA) approach to clutter filtering and compare it to adaptive SVD filtering with and without AD using simulations and a single vessel phantom. We show that AD+ICA and ICA produce the highest blood-to-background signal-to-noise ratios in simulations and in the phantom, respectively.