奇异值分解滤波器在超声造影成像的应用及性能分析*

信噪比和造影-组织残留比是衡量造影性能的主要指标,现有提升方法多从频域入手,存在两个缺陷:噪声、组织残留与造影剂信号存在频带重叠时(如二次谐波),频域滤波性能受限;虽然非线性基波、次谐波等方法可解决频带重叠问题,但面对前端电路非对称性及信号饱和导致的组织残留,仍难在频域进行滤除。奇异值分解滤波器是一种非频域方法,可将接收信号分解成若干空时域子空间,通过对子空间进行保留及重组,可提取微泡且滤除噪声与组织残留。该文研究了该类滤波器在造影成像的应用,并采用性能分析实验论证了其可行性与优越性。

Application and performance analysis of singular value decomposition filter in contrast-enhanced ultrasound

The performance of contrast-enhanced ultrasound (CEUS) is usually evaluated in terms of signal-to-noise ratio (SNR) and contrast-to-tissue ratio (CTR), which are commonly improved by filters in frequency domain. Drawbacks of these filters can be addressed as two aspects: (1) the filtering performance is of limitation as a result of the band overlap problem occurring between noise/residual tissue signal and microbubble (MB) signal (e.g., second harmonic); (2) although the band overlap can be tackled with methods such as nonlinear fundamental and sub-harmonic, the residual tissue caused by asymmetry of emitting hardware and signal saturation is still difficult to be eliminated. The singular value decomposition (SVD) filter is a non-frequency domain method. In particular, it firstly decomposes the receiving signals into several spatiotemporal subspaces via SVD, and then separates MB from noise and residual tissue by preserving targeted subspaces and recomposing them. This paper has applied the SVD filter in CEUS, and verified its feasibility and superiority based on performance analysis experiments.