超快超声心动图中的快速稳健杂波滤波
本研究将随机奇异值分解(rSVD)用于超快超声心动图功率多普勒分析前的杂波滤波,并与全SVD比较。在9例正常人体内超快超声心动图中,rSVD加速滤波12倍,局部对比度(CNR)显著更优(p < 0.001),整体对比度(CR)效果相当(p = 0.20),且在最低信噪比下CR与CNR的标准差更小,稳健性更优。
要点
- rSVD较全SVD将杂波滤波加速12倍,适合超快超声心动图大数据量处理。
- rSVD在功率多普勒图像上提供显著更优的局部对比度(CNR,p < 0.001)。
- rSVD在低信噪比条件下对比度时间变异更小,稳健性优于全SVD。
核心结论
- rSVD可作为超快超声心动图功率多普勒分析前快速且稳健的杂波滤波器,建议用于血液动力学分析。
研究设计与样本
- 研究类型为方法学对比研究,纳入9例正常人体内超快超声心动图,采用级联合成孔径成像方法采集。
- 其中1例受试者在4种不同超声信噪比水平下额外扫描,比较rSVD与全SVD滤波后的功率多普勒图像。
- 参考标准为全SVD滤波,评估指标为对比度(CR)和对比噪声比(CNR)的均值与标准差。
诊断标准与关键数值
- rSVD较全SVD加速杂波滤波12倍。
- rSVD的局部对比度(平均CNR)显著优于全SVD(p < 0.001),整体对比度(平均CR)效果相当(p = 0.20)。
- 最低信噪比场景下,rSVD的CR标准差为1.32 dB(全SVD为5.49 dB),CNR标准差为1.27 dB(全SVD为5.49 dB)。
临床应用与阅片要点
- 在超快超声心动图功率多普勒血流分析中,可优先考虑rSVD滤波以提升计算效率并维持对比度。
- 低信噪比条件下rSVD对比度时间变异更小,阅片时需注意全SVD可能因组织运动导致对比度不稳定。
局限性
- 研究仅纳入9例正常人体内数据,样本量小且未涉及病理状态。
- 仅在1例受试者中评估不同信噪比水平下的稳健性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
Singular value decomposition (SVD)-based filters have become the norm for clutter filtering in ultrasound blood flow applications but are computationally expensive and susceptible to large and fast tissue motion. Randomized SVD (rSVD) has later been shown to successfully accelerate filtering of in vivo stationary tissues. However, little is known about its performance on ultrafast echocardiography, which produces thousands of frames to assess complex myocardial deformation and blood dynamics. Neither has its inherently robust randomized scheme been proven in any ultrasound blood flow imaging methods. This study thus proposed to employ rSVD as a fast and robust clutter filter for ultrafast echocardiograms prior to power Doppler analysis. Ultrafast echocardiograms of nine normal human hearts were acquired in vivo by our cascaded synthetic aperture imaging method. One subject was additionally scanned under four different sonographic signal-to-noise ratio (SNR) levels. Contrast ratio (CR) and contrast-to-noise ratio (CNR) of in vivo power Doppler images obtained from filtered ultrafast echocardiograms were calculated, and their mean and standard deviation within a cardiac cycle represented temporal average and variation of contrast resolution, respectively. Our in vivo results showed that rSVD accelerated clutter filtering by 12-fold and provided significantly better local contrast (mean CNR values: p < 0.001) while being equally effective (mean CR values: p = 0.20) compared with full-SVD. rSVD yielded smaller standard deviations of CR (1.32 dB vs. 5.49 dB) and CNR (1.27 dB vs. 5.49 dB) than full-SVD in the lowest SNR scenario, thus substantiating its superior robustness. Our findings suggest using rSVD in ultrafast echocardiographic blood dynamics analysis.
引用
- Xu, Yiu, Lee, et al. Fast and Robust Clutter Filtering in Ultrafast Echocardiography. Ultrasound Med Biol. 2022. PMID: 36372594. doi:10.1016/j.ultrasmedbio.2022.09.013
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。