L.N. Bohs

A novel method for angle independent ultrasonic imaging of blood flow and tissue motion

A simple algorithm for angle independent motion imaging is described. This method requires only one absolute difference operation per pixel, compared to eight operations for normalized cross correlation. Quantitative studies using speckle-generating targets translated by fixed amounts both axially and laterally indicate that the technique tracks moving speckle as accurately as correlation. Color flow images generated from clinical blood and liver data highlight the success of the technique for tracking both large and small motions in two dimensions. The algorithms suitability for implementation in digital hardware makes possible the development of clinical instruments for angle independent ultrasonic imaging of blood flow and tissue motion in real time.<<ETX>>

Speckle tracking for multi-dimensional flow estimation.

Speckle tracking methods overcome the major limitations of current Doppler methods for flow imaging and quantification: angle dependence and aliasing. In this paper, we review the development of speckle tracking, with particular attention to the advantages and limitations of two-dimensional algorithms that use a single transducer aperture. Ensemble tracking, a recent speckle tracking method based upon parallel receive processing, is described. Experimental results with ensemble tracking indicate the ability to measure laminar flow in a phantom at a beam-vessel angle of 60 degrees, which had not been possible with previous 2D speckle tracking methods. Finally, important areas for future research in speckle tracking are briefly summarized.

A real time system for quantifying and displaying two-dimensional velocities using ultrasound

This paper describes a system that has been developed for measuring two-dimensional velocities in real time using ultrasound. The instrument tracks interframe speckle pattern motion using a Sum-Absolute-Difference (SAD) algorithm in order to produce a vector map of 2D velocities. The systems parallel architecture allows calculation of approximately 20,000 vectors per second using the current tracking geometry. A programmable graphics processor encodes individual velocity vectors with color and displays them superimposed on the B-mode image in real time. In vitro tests indicate that the system can track velocities well over the Doppler aliasing limit in any direction in the scan plane with greater than 94% accuracy. A color encoded image obtained from a flow phantom highlights the systems ability to display lateral motion with uniform coloration, in contrast to the two-color display of current ultrasonic Doppler instruments.

Experimental velocity profiles and volumetric flow via two-dimensional speckle tracking

The performance of a two-dimensional speckle tracking system in measuring in vitro laminar flow is evaluated. The system uses a pattern matching algorithm to track subresolution-sized speckle regions between successive ultrasonic 2D pulse-echo acquisitions in order to determine both the axial and the lateral components of velocity. In this study, multiple 2D vector velocity maps were acquired in real time using a calibrated laminar flow phantom, and then statistically analyzed off-line. At a 90 degrees transducer angle, volumetric flow rates computed from measured velocity profiles exhibited excellent linearity (R2 > 0.99), with a mean error of -6.1%, over the range 5-30 mL/s. At 105 degrees and 120 degrees, experimental volume flow rates also agreed well with actual rates, although measured velocity profiles appeared more irregular with decreasing Doppler angles. Velocity profiles estimated using sampled radio-frequency data rather than envelope-detected data were inconsistent due to an insufficient sampling rate and the quantization of the velocity grid. Results indicate that excellent flow velocity and volume rate estimates can be obtained from vector velocity measurements along a single line of sight, without a priori knowledge of the flow direction, at transducer angles near 90 degrees where Doppler instruments are prone to large errors.

Relative performance of two-dimensional speckle-tracking techniques: normalized correlation, non-normalized correlation and sum-absolute-difference

The authors present computer simulations that compare the performance of three two-dimensional block-matching algorithms as applied to ultrasonic speckle. The three algorithms, normalized correlation, non-normalized correlation and sum absolute difference (SAD), were applied to simulated speckle patterns with varying levels of noise and with different kernel sizes. The results show that SAD and normalized correlation have similar performance characteristics and are able to track speckle motion with a kernel size as small as one resolution cell. Non-normalized correlation performed poorly with small kernel sizes but improved as the kernel size increased.

Speckle decorrelation due to two-dimensional flow gradients

The performance of ultrasonic velocity estimation methods is degraded by speckle decorrelation, the change in received echoes over time. Because ultrasonic speckle is formed by the complex sum of echoes from subresolution scatterers, it is sensitive to the relative motion of those scatterers. Velocity gradients in flowing blood result in relative scatterer motion and can be a significant source of speckle decorrelation. Computer simulations were performed to evaluate speckle decorrelation due to two-dimensional flow gradients. Results indicate that decorrelation due to flow gradients is sensitive to the angle of flow and has a maximum at a beam-vessel angle of 0/spl deg/, i.e., purely axial flow. A quantitative summary of the major factors causing speckle decorrelation indicates that flow gradients are the most significant contributors under the conditions modeled.

A novel interpolation strategy for estimating subsample speckle motion

Multidimensional, high-resolution ultrasonic imaging of rapidly moving tissue is primarily limited by sparse sampling in the lateral dimension. In order to achieve acceptable spatial resolution and velocity quantization, interpolation of laterally sampled data is necessary. We present a novel method for estimating lateral subsample speckle motion and compare it with traditional interpolation methods. This method, called grid slopes, requires no a priori knowledge and can be applied to data with as few as two samples in the lateral dimension. Computer simulations were performed to compare grid slopes with two conventional interpolation schemes, parabolic fit and cubic spline. Results of computer simulations show that parabolic fit and cubic spline performed poorly at translations greater than 0.5 samples, and translations less than 0.5 samples were subject to an estimation bias. Grid slopes accurately estimated translations between 0 and 1 samples without estimation bias at high signal-to-noise ratios. Given that the grid slopes interpolation technique performs well at high signal-to-noise ratios, one pertinent clinical application might be tissue motion tracking.

Ensemble tracking for 2D vector velocity measurement: Experimental and initial clinical results

We describe a new method, called ensemble tracking, for estimating two-dimensional velocities with ultrasound. Compared to previous speckle tracking techniques, ensemble tracking measures motion over smaller times and distances, increasing maximum velocities and reducing errors due to echo decorrelation. Ensemble tracking uses parallel receive processing, 2D pattern matching, and interpolation of the resulting tracking grid to estimate sub-pixel speckle translations between successive ultrasonic acquisitions. In this study, small translations of a tissue mimicking phantom were quantified at transducer angles of 0/spl deg/, 45/spl deg/, and 90/spl deg/. Measurements over three parallel beam spacings and all transducer angles had mean errors from -4% to +11%, when parallel beam amplitudes were normalized. Such amplitude normalization substantially improved results at 45/spl deg/ and 90/spl deg/. The amplitude, spacing, and correlation between the parallel beams were quantified, and their effects on the accuracy and precision of estimates are discussed. Finally, initial clinical results demonstrate the ability to track and display blood flow in the carotid artery.

2-D motion estimation using two parallel receive beams

We describe a method for estimating 2-D target motion using ultrasound. The method is based on previous ensemble tracking techniques, which required at least four parallel receive beams and 2-D pattern matching. In contrast, the method described requires only two parallel receive beams and 1-D pattern matching. Two 1-D searches are performed, one in each lateral direction. The direction yielding the best match indicates the lateral direction of motion. Interpolation provides sub-pixel magnitude resolution. We compared the two beam method with the four beam method using a translating speckle target at three different parallel beam steering angles and transducer angles of 0, 45, and 90/spl deg/. The largest differences were found at 90 degrees, where the two beam method was generally more accurate and precise than the four beam method and also less prone to directional errors at small translations. We also examined the performance of both methods in a laminar flow phantom. Results indicated that the two beam method was more accurate in measuring the flow angle when the flow velocity was small. Computer simulations supported the experimental findings. The poorer performance of the four beam method was attributed to differences in correlation among the parallel beams. Specifically, center beams 2 and 3 correlated better with each other than with the outer beams. Because the four beam method used a comparison of a kernel region in beam pair 2-3 with two different beam pairs 1-2 and 3-4, the 2-to-1 and 3-to-4 components of this comparison increased the incidence of directional errors, especially at small translations. The two beam method used a comparison between only two beams and so was not subject to this source of error. Finally, the two beam method did not require amplitude normalization, as was necessary for the four beam method, when the two beams were chosen symmetric to the transmit axis. We conclude that two beam ensemble tracking can accurately estimate motion using only two parallel receive beams.

A comparison of algorithms for tracking sub-pixel speckle motion

We have previously described Ensemble Tracking, a speckle tracking method that uses parallel receive processing and 2D pattern matching to generate a vector velocity map with high spatial and temporal resolution. Due to the mismatch in axial and lateral spatial frequencies of speckle and limited lateral sampling, Ensemble Tracking requires lateral interpolation to improve lateral velocity resolution. In this paper, we compare three interpolation algorithms for measuring lateral sub-pixel speckle motion: Cubic Spline, Parabolic Fit, and Grid Slopes. Simulation results show that Grid Slopes performs better than Cubic Spline and Parabolic Fit, and is accurate over a wider range of speckle translations.