[Paper Review] On the Use of Singular Value Decomposition as a Clutter Filter for Ultrasound Flow Imaging

Kai Riemer, Marcelo Lerendegui+5 more|arXiv (Cornell University)|Apr 25, 2023

Ultrasound Imaging and ElastographyMedicine3 citations

TL;DR

This paper investigates artifacts introduced by singular value decomposition (SVD)-based clutter filtering in high-frame-rate ultrasound flow imaging, demonstrating that tissue motion, improper threshold selection, and flow dynamics cause intensity fluctuations, ghosting, and signal splitting. The study reveals that SVD filtering, while enhancing signal-to-noise ratio, can distort hemodynamic signals and recommends motion correction and adaptive filtering to mitigate these artifacts.

ABSTRACT

Filtering based on Singular Value Decomposition (SVD) provides substantial separation of clutter, flow and noise in high frame rate ultrasound flow imaging. The use of SVD as a clutter filter has greatly improved techniques such as vector flow imaging, functional ultrasound and super-resolution ultrasound localization microscopy. The removal of clutter and noise relies on the assumption that tissue, flow and noise are each represented by different subsets of singular values, so that their signals are uncorrelated and lay on orthogonal sub-spaces. This assumption fails in the presence of tissue motion, for near-wall or microvascular flow, and can be influenced by an incorrect choice of singular value thresholds. Consequently, separation of flow, clutter and noise is imperfect, which can lead to image artefacts not present in the original data. Temporal and spatial fluctuation in intensity are the commonest artefacts, which vary in appearance and strengths. Ghosting and splitting artefacts are observed in the microvasculature where the flow signal is sparsely distributed. Singular value threshold selection, tissue motion, frame rate, flow signal amplitude and acquisition length affect the prevalence of these artefacts. Understanding what causes artefacts due to SVD clutter and noise removal is necessary for their interpretation.

Motivation & Objective

• To identify and characterize image artifacts introduced by SVD-based clutter filtering in 2D ultrasound flow imaging.
• To investigate the impact of tissue motion, flow velocity, signal amplitude, and singular value threshold selection on artifact prevalence.
• To evaluate the limitations of SVD as a linear filtering method in handling non-linear clutter and time-varying tissue dynamics.
• To assess the effectiveness of motion correction (MoCo) and sliding window SVD in reducing artifacts.
• To provide practical guidance for clinicians and researchers on optimal SVD parameter selection to minimize misinterpretation of flow signals.

Proposed method

• Application of SVD to ultrasound flow imaging data to decompose signals into singular values representing tissue, flow, and noise components.
• Use of a threshold on singular values to retain only the flow-related components, assuming orthogonal subspaces for clutter, flow, and noise.
• Implementation of motion correction (MoCo) to estimate and compensate for tissue motion, reducing time-varying artifacts.
• Employment of a sliding window SVD filter to adaptively process data in temporal segments, improving stability under motion.
• Comparison of filtered and unfiltered images across simulations and in vivo experiments to isolate SVD-induced artifacts.
• Systematic variation of singular value thresholds and acquisition parameters to assess their impact on artifact generation and signal fidelity.

Experimental results

Research questions

• RQ1 What types of artifacts are introduced by SVD-based clutter filtering in ultrasound flow imaging, and how do they vary with imaging parameters?
• RQ2 How does tissue motion influence the appearance and persistence of SVD-induced artifacts such as flashing and intensity fluctuations?
• RQ3 To what extent does the choice of singular value threshold affect artifact prevalence and signal integrity?
• RQ4 Can motion correction and sliding window filtering reduce or eliminate SVD-induced artifacts in dynamic imaging scenarios?
• RQ5 How do flow velocity, microbubble distribution, and acquisition length modulate the severity of SVD filtering artifacts?

Key findings

• SVD-based clutter filtering induces temporal intensity fluctuations (flashing) that correlate with cardiac cycle and peak vessel motion, even with high singular value thresholds.
• Ghosting and signal splitting artifacts are prevalent in microvascular flow imaging due to sparse distribution of microbubble signals and improper thresholding.
• A low singular value threshold reduces ghosting and splitting but increases flashing, while a high threshold reduces flashing at the cost of reduced contrast-to-tissue ratio (CTR) and potential signal loss.
• Motion correction (MoCo) reduces ghosting and splitting by estimating tissue displacement, but does not fully eliminate flashing caused by dynamic clutter changes.
• The sliding window SVD filter improves artifact mitigation by adapting to local spatiotemporal variations, particularly in regions with motion or pulsatile flow.
• Artifacts are not inherent to SVD but arise from incorrect assumptions—such as signal orthogonality and static clutter—under conditions of tissue motion or non-stationary flow.