Surgery – Diagnostic testing – Detecting nuclear – electromagnetic – or ultrasonic radiation
Reexamination Certificate
2001-03-16
2002-12-17
Lateef, Marvin M. (Department: 3737)
Surgery
Diagnostic testing
Detecting nuclear, electromagnetic, or ultrasonic radiation
C600S449000
Reexamination Certificate
active
06494834
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method of measuring strain in a target body, using the transmission, reception, processing and normalization of ultrasound signals.
2. Description of the Prior Art
Imaging of elastic parameters of soft tissue has developed into a new tool for diagnosis of disease. Current estimators of tissue motion, include a time-domain cross-correlation based speckle tracking algorithm, and a Fourier based speckle phase-tracking technique. These techniques are coherent estimation techniques, i.e., these methods are sensitive to phase variations. The coherent estimation techniques generally have the advantage of being highly precise. Strain Filter (SF) analysis has shown, however, that they are not very robust in the presence of even a small amount of decorrelation between the pre- and post-compression signals. The term ‘robustness’ has been used in statistical analysis to denote the good performance of statistical tests, i.e., the homogeneity of the variance calculation, even if the data deviates from theoretical requirements. By equivalence in elastography, ″robustness′ denotes the consistently good performance of the estimator even at high decorrelation noise, i.e., keeping the variance of estimation at a relatively constant and low level at a large range of noise levels.
The term “decorrelation” as used herein, is defined as the loss of full correlation between the pre- and postcompressed windowed signal segments. Therefore, decorrelation may be encountered due to many sources, such as intrawindow axial motion undesired lateral or elevational motion, jitter (i.e., any cause of misregistration between the pre- and post-compressed A-line segments), unstable mechanical setup, etc. The main idea in this study is to introduce a new estimator that is more immune to decorrelation compared to other estimators.
SUMMARY OF THE INVENTION
The tissue strain estimator is a spectral estimator that estimates strain directly. Since the proposed estimator uses the power spectrum, it is incoherent, i.e., it does not use the phase of the signal. Previously reported incoherent methods include optical flow speckle tracking, envelope cross-correlation, and spectral chirp z-transform techniques. Generally, incoherent methods may be less precise but significantly more robust. For example, we have demonstrated this property for the case of time-delay estimation using the envelope of echo-signals. This may be a significant advantage where elastography is to be practiced in situations involving (1) undesired scanning motion, such as the case of using an unstable handheld transducer and/or (2) undesired tissue motion, such as abdominal or intravascular elastography. This property of the estimator is demonstrated later in this paper in the simulation results section through testing of its immunity to noise caused by jitter.
The main idea behind a spectral strain approach is based on the Fourier scaling property, which implies that a compression or expansion of the time-domain signal should lead to an expansion or compression of its power spectrum, respectively. One of the most well known and thoroughly studied spectral motion effects is the Doppler shift, which typically links the frequency shift to the scatterer velocity between emissions. Velocities towards the transducer result in a positive frequency shift, while the opposite is true for scatterers that move away from the transducer. However, since the scatterers within a given resolution length do not move at the same velocities, a spectrum of Doppler frequencies is observed. Therefore, initially in ultrasound, the methods of velocity estimation for the measurement of blood flow mainly operated in the frequency domain, otherwise known as spectrum analysis techniques and measured the mean velocity of scatterers across the vessel lumen (indicative of the volumetric flow rate) by estimating the mean frequency of the power spectrum. Despite the success of these techniques even in in vivo vessels, detection of the Doppler frequency shift, which is typically on the order of 1 kHz, is not possible for pulsed instruments, since the downshift in frequency due to attenuation (on the order of 10-100 kHz) is expected to dominate over the Doppler shift. Since in elastography the pre- and postcompressed segments are approximately identical depths, the attenuation effect on the two spectra is assumed to be identical and cancelled out when the two spectra are compared.
Strain estimation using spectral methods depends on the subsequent change in the scatterer statistics. Spectral methods typically link one or more signal parameters to the change in mean scatterer spacing. One prior art relates the relative change in the mean scatterer spacing to the strain incurred during a cardiac cycle. This method assumes the presence of underlying scatterer periodicities. Despite the fact that this has also been demonstrated to work in in vivo intravascular applications, the main assumptions of regular spacing or periodicities may not hold for most tissues. In contrast, as shown in the theory section, the spectral methods mentioned in this paper make no assumptions regarding the composition of the tissue scatterers.
Typically in elastography time-domain techniques are used that involve the computation of the time-delay to estimate the displacement following an applied compression, and the estimation of strain by applying gradient operations on the previously obtained time-delay estimates. As mentioned earlier, an important advantage associated with these spectral methods as well as other estimators, such as the adaptive stretching estimator, is that they can be used to estimate strain directly; i.e., without involving the use of noisy gradient operators. In the latter case, the gradient operation introduces additional amplification of the noise into the strain estimation process, thus degrading the strain estimates. Furthermore, similar to the adaptive stretching estimator, only one estimation window is needed, for both the magnitude and the sign of the strain to be estimated.
As shown later in the theory section, spectral estimators can be divided into two main groups: a) the spectral shift methods and b) the spectral bandwidth methods. Despite the fact that we develop expressions that show direct strain estimation in both cases, in this paper we focus primarily on a spectral shift method; we estimate the relative shift in the spectral centroid caused by compressive or tensile tissue strain. Therefore, throughout this paper this new estimator is referred to as the “centroid strain estimator”, “centroid estimator” or “centroid method”. Current investigations that deal with the development of alternative shift estimators as well as bandwidth estimators will not be reported in this study.
The spectral centroid has been widely used in estimating the Doppler shift, attenuation and backscattering. The theory underlying the use of centroid strain estimators is presented in the next section. One-dimensional (1-D), instead of two-dimensional (2-D), motion simulations are used in order to more accurately study the performance of the estimator, i.e., independent of the effect of signal decorrelation in two dimensions that complicates the measurements. Simulation results in 1-D illustrate the insensitivity of the centroid strain estimator to signal decorrelation effects. It is important to note that, as mentioned earlier, decorrelation can be due to several sources. For the purpose of this paper, we consider solely the axial decorrelation effect in this 1-D model. We, thus, assume that the robustness demonstrated by the spectral estimator vis-à-vis this effect is a more general property that can be further applied at other decorrelation scenarios. For example, it is shown in the results section how the spectral method is indeed more immune to jitter, another source of decorrelation. The elastograms obtained using these simulations as well as phantom experiments illustrate the robustness of the spectral
Konofagou Elisa
Ophir Jonathan
Duane Morris LLP
Imam Ali M.
Lateef Marvin M.
The Board of Regents of the University of Texas System
LandOfFree
Power spectral strain estimators in elastography does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Power spectral strain estimators in elastography, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Power spectral strain estimators in elastography will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2937325