Randomized Spatial Downsampling based Robust PCA Clutter Filtering for Ultrafast Ultrasound Imaging

In this study, a modified robust principal component analysis (RPCA) clutter filtering method is applied for efficient microvessel imaging by decomposing the acquired ultrasound signal into low-rank and sparse components. A combination strategy based on alternating direction method of multipliers (ADMM) and randomized spatial downsampling is applied to accelerate the computation. A parallel processing is further developed. The proposed method is applied to in vivo data of brain and spinal cord in rats. The results illustrated that comparing with the classical SVD method, it enables to enhance the clutter removal and noise filtering, which demonstrates the potentials of the proposed method in ultrafast ultrasound microvasculature imaging.