Andreas Austeng

Adaptive Beamforming Applied to Medical Ultrasound Imaging

We have applied the minimum variance (MV) adaptive beamformer to medical ultrasound imaging and shown significant improvement in image quality compared to delay-and-sum (DAS). We demonstrate reduced main-lobe width and suppression of sidelobes on both simulated and experimental RF data of closely spaced wire targets, which gives potential contrast and resolution enhancement in medical images. The method is applied to experimental RF data from a heart phantom, in which we show increased resolution and improved definition of the ventricular walls. A potential weakness of adaptive beamformers is sensitivity to errors in the assumed wavefield parameters. We look at two ways to increase robustness of the proposed method; spatial smoothing and diagonal loading. We show that both are controlled by a single parameter that can move the performance from that of a MV beamformer to that of a DAS beamformer. We evaluate the sensitivity to velocity errors and show that reliable amplitude estimates are achieved while the mainlobe width and sidelobe levels are still significantly lower than for the conventional beam-former.

Benefits of minimum-variance beamforming in medical ultrasound imaging

Recently, significant improvement in image resolution has been demonstrated by applying adaptive beamforming to medical ultrasound imaging. In this paper, we have used the minimum-variance beamformer to show how the low sidelobe levels and narrow beamwidth of adaptive methods can be used, not only to increase resolution, but also to enhance imaging in several ways. By using a minimum-variance beamformer instead of delay-and-sum on reception, reduced aperture, higher frame rates, or increased depth of penetration can be achieved without sacrificing image quality. We demonstrate comparable resolution on images of wire targets and a cyst phantom obtained with a 96-element, 18.5-mm transducer using delay-and-sum, and a 48-element, 9.25-mm transducer using minimum variance. To increase frame rate, fewer and wider transmit beams in combination with several parallel receive beams may be used. We show comparable resolution to delay-and-sum using minimum variance, 1/4th of the number of transmit beams and 4 parallel receive beams, potentially increasing the frame rate by 4. Finally, we show that by lowering the frequency of the transmitted beam and beamforming the received data with the minimum variance beamformer, increased depth of penetration is achieved without sacrificing lateral resolution.

Sparse 2-D arrays for 3-D phased array imaging - design methods

One of the most promising techniques for limiting complexity for real-time 3-D ultrasound systems is to use sparse 2-D layouts. For a given number of channels, optimization of performance is desirable to ensure high quality volume images. To find optimal layouts, several approaches have been followed with varying success. The most promising designs proposed are Vernier arrays, but also these suffer from high peaks in the sidelobe region compared with a dense array. In this work, we propose new methods based on the principles of suppression of grating lobes to form symmetric and non-symmetric regular sparse periodic and radially periodic designs. The proposed methods extend the concept of sparse periodic layouts by exploiting either an increased number of symmetry axes or radial symmetry. We also introduce two new strategies to form designs with nonoverlapping elements. The performance of the new layouts range from the performance of Vernier arrays to almost that of dense arrays. Our designs have simplicity in construction, flexibility in the number of active elements, and the possibility of trade off sidelobe peaks against sidelobe energy.

A low complexity data-dependent beamformer

The classical problem of choosing apodization functions for a beamformer involves a trade-off between main lobe width and side lobe level, i.e., a trade-off between resolution and contrast. To avoid this trade-off, the application of adaptive beamforming, such as minimum variance beamforming, to medical ultrasound imaging has been suggested. This has been an active topic of research in medical ultrasound imaging in the recent years, and several authors have demonstrated significant improvements in image resolution. However, the improvement comes at a considerable cost. Where the complexity of a conventional beamformer is linear with the number of elements [O(M)], the complexity of a minimum variance beamformer is as high as O(M3). In this paper, we have applied a method based on an idea by Vignon and Burcher which is data-adaptive, but selects the apodization function between several predefined windows, giving linear complexity. In the proposed method, we select an apodization function for each depth along a scan line based on the optimality criterion of the minimum variance beamformer. However, unlike the minimum variance beamformer, which has an infinite solution space, we limit the number of possible outcomes to a set of predefined windows. The complexity of the method is then only P times that of the conventional method, where P is the number of predefined windows. The suggested method gives significant improvement in image resolution at a low cost. The method is robust, can handle coherent targets, and is easy to implement. It may also be used as a classifier because the selected window gives information about the object being imaged. We have applied the method to simulated data of wire targets and a cyst phantom, and to experimental RF data from a heart phantom using P = 4 and P = 12. The results show significant improvement in image resolution compared with delay-and-sum.

Minimum variance adaptive beamforming applied to medical ultrasound imaging

We have applied the minimum variance beam- former to medical ultrasound imaging and shown significant improvement in image quality compared to delay-and-sum. Reduced mainlobe width and suppression of sidelobes is demonstrated on both simulated and experimental RF data of closely spaced wire targets, resulting in increased resolution and contrast. The method has been applied to experimental RF data from a heart-phantom, demonstrating improved definition of the ventricular walls. We have evaluated the beamformers sensitivity to velocity errors and shown that reliable amplitude estimates are achieved if proper regular- ization is applied. I. INTRODUCTION Delay-and-sum (DAS) beamforming is the standard technique in medical ultrasound imaging. An image is formed by transmitting a narrow beam in a number of angles and dynamically delaying and summing the received signals from all channels. The large sidelobes of the DAS beamformer can be suppressed using aperture shading, resulting in increased contrast at the expense of resolution. In contrast to the predetermined shading in DAS, adaptive beamformers use the recorded wavefield to compute the aperture weights. By suppressing inter- fering signals from off-axis directions and allowing large sidelobes in directions where there is no received energy, the adaptive beamformers can increase resolution. The minimum variance (MV) adaptive beamformer (1) and subspace-based methods have mostly been studied in narrowband applications. Extensions to broadband imaging include preprocessing with focusing- and spa- tial resampling filters, allowing narrowband methods to be used on broadband data (2), (3). We have applied the MV beamformer to medical ultrasound imaging by prefocusing in the direction of the transmitted beam - as the delay-step in DAS - and replaced the summing with the MV method. Similar methods have been used by Mann and Walker (4), and Sasso and Cohen-Bacrie (5) in medical ultrasound imaging. The former use a con- strained adaptive beamformer on experimental data of a single point target and a cyst phantom demonstrating improved contrast and resolution, whereas the latter use an MV beamfomer on a simulated data-set, showing improved contrast in the final image. We demonstrate resolution improvement and sidelobe suppression on both simulated and experimental RF data of closely spaced wire targets, and show improvement in the image of a heart-phantom obtained from experimental RF data. We also evaluate robustness of the beamformer to errors in acoustic velocity, and show that reliable amplitude estimates are achieved by regularization.

Eigenspace Based Minimum Variance Beamforming Applied to Ultrasound Imaging of Acoustically Hard Tissues

Minimum variance (MV) based beamforming techniques have been successfully applied to medical ultrasound imaging. These adaptive methods offer higher lateral resolution, lower sidelobes, and better definition of edges compared to delay and sum beamforming (DAS). In standard medical ultrasound, the bone surface is often visualized poorly, and the boundaries region appears unclear. This may happen due to fundamental limitations of the DAS beamformer, and different artifacts due to, e.g., specular reflection, and shadowing. The latter can degrade the robustness of the MV beamformers as the statistics across the imaging aperture is violated because of the obstruction of the imaging beams. In this study, we employ forward/backward averaging to improve the robustness of the MV beamforming techniques. Further, we use an eigen-spaced minimum variance technique (ESMV) to enhance the edge detection of hard tissues. In simulation, in vitro, and in vivo studies, we show that performance of the ESMV beamformer depends on estimation of the signal subspace rank. The lower ranks of the signal subspace can enhance edges and reduce noise in ultrasound images but the speckle pattern can be distorted.

1D and 2D algorithmically optimized sparse arrays

A new criterion, minimization of maximal weighted sidelobe, is applied together with a genetic search algorithm to the problem of element placement in a discrete linear lattice. By processing RF-data the resulting 1D sparse arrays are compared experimentally with sparse periodic arrays and arrays found with a least square optimization criterion The images show that algorithmically optimized layouts can be found with lateral resolution and contrast comparable to sparse periodic arrays. The approach has been extended to 2D arrays. An example is given without overlap between elements on transmit and receive. The layout is optimized for both the visible and invisible region to assure a controlled behavior when steering is applied. Advantages of this approach are that it gives flexibility in the choice of beamwidth, the potential to trade off beamwidth, maximum sidelobe level and sidelobe energy, no need for apodization, and possibility to require e.g. no overlap between receive and transmit elements.

Implementing capon beamforming on a GPU for real-time cardiac ultrasound imaging

Capon beamforming is associated with a high computational complexity, which limits its use as a real-time method in many applications. In this paper, we present an implementation of the Capon beamformer that exhibits realtime performance when applied in a typical cardiac ultrasound imaging setting. To achieve this performance, we make use of the parallel processing power found in modern graphics processing units (GPUs), combined with beamspace processing to reduce the computational complexity as the number of array elements increases. For a three-dimensional beamspace, we show that processing rates supporting real-time cardiac ultrasound imaging are possible, meaning that images can be processed faster than the image acquisition rate for a wide range of parameters. Image quality is investigated in an in vivo cardiac data set. These results show that Capon beamforming is feasible for cardiac ultrasound imaging, providing images with improved lateral resolution both in element-space and beamspace.

Sparse Sampling in Array Processing

Sparsely sampled irregular arrays and random arrays have been used or proposed in several fields such as radar, sonar, ultrasound imaging, and seismics. We start with an introduction to array processing and then consider the combinatorial problem of finding the best layout of elements in sparse 1-D and 2-D arrays. The optimization criteria are then reviewed: creation of beampatterns with low mainlobe width and low sidelobes, or as uniform as possible coarray. The latter case is shown here to be nearly equivalent to finding a beampattern with minimal peak sidelobes.

Sparse 2-D arrays for 3-D phased array imaging - experimental validation

To be able to describe more precisely the behavior of a real-time 3-D ultrasound system with either a dense array or various sparse designs, experimental data from a 2-D fully connected array prototype with 50/spl times/50 elements have been collected. The data have been processed off line to form synthetic aperture 3-D volume images. Simulated and experimental results are compared and show good correlation. The performance of the best sparse designs, all thinned to more than 50%, offer performances comparable to a dense array.