Stereotactic endovascular aortic navigation with a novel ultrasonic-based three-dimensional localization system.

BACKGROUND: Endovascular aortic procedures have been developed to treat many aortic diseases effectively. However, these procedures are also becoming increasingly complex given the development of branched or fenestrated endografts. Part of the difficulty lies in the limitations of current imaging paradigms. A more intuitive, three-dimensional (3D) mode of intraoperative imaging is desirable to accommodate the future progression of endovascular techniques. This article describes a novel endovascular catheter tracking device that uses ultrasonic signals, not ultrasound imaging. The tracking device displays real-time in vivo location on previously acquired 3D computed tomography (CT) images in an intuitive, endoluminal view. This system was tested in two swine and validated against fluoroscopy and by delivering stent grafts. METHODS: The ultrasonic-based localization system (ULS) provides real-time location information of a modified endovascular catheter and displays this location on preoperative 3D CT images. The 9F endovascular catheter has a small ultrasonic transmitter attached to its tip to signal its location to the ULS. Subsequent endovascular deployment of an aortic stent was carried out using only the ULS to target the stent placement position in the aorta of Yorkshire swine. System accuracy was measured against concurrent angiography as well as to deployed stents in situ. RESULTS: We successfully displayed the endovascular catheter tip location in real time along the registered CT aortic images, providing virtual endoluminal tracking. The relative accuracy of the ULS as compared with angiography for catheter movements in the abdominal aorta was found to have a mean error less than 1 mm. The ULS coordinates tracked within the lumen of the aortic image 98% of the time, as defined by the proportion of points within one radius distance of the aortic image centerline. Finally, three aortic stents were deployed using the ULS virtual image display to locate the target position in the aorta for stent deployment. Errors between target position and actual stent position ranged from -5.0 to +7.9 mm. CONCLUSIONS: This study demonstrates the feasibility of virtual image-guided endovascular aortic navigation using a ULS. This provides a 3D platform for virtual navigation on preoperative CT scan images during endovascular procedures that could assist in stent deployment as well as minimize or eliminate the need for procedural ionizing radiation and iodinated contrast. Future work will focus on miniaturization and refinements in accuracy that will be required to translate this technology into clinical application in endovascular procedures.

Beamforming of sound from two-dimensional arrays using spatial matched filters.

Fully-sampled two-dimensional (2D) arrays can have two-way focusing of the ultrasound beam in both lateral directions leading to high quality, real-time three-dimensional (3D) imaging. However, fully-sampled 2D arrays with very large element counts (>16,000) are difficult to manufacture due to interconnect density and large element electrical impedance. As an alternative, row-column or crossed electrode arrays have been proposed to simplify transducer fabrication and system integration. These types of arrays consist of two one-dimensional arrays oriented perpendicular to each other. Using conventional delay-and-sum beamforming, each array performs one-way focusing in perpendicular lateral directions which yield higher sidelobe and acoustic clutter levels compared to fully-sampled 2D arrays with two-way focusing. In this paper, the use of spatial matched filters to improve focusing of row-column arrays is investigated. On receive, data from each element are first spatial match filtered in the elevation direction. After summation, the data are filtered again in the azimuth direction. Beam widths comparable to one-way focusing are seen in azimuth and beam widths comparable to two-way focusing are achieved in elevation. 3D beam patterns from computer simulation results using a 7.5 MHz 128 × 128 row-column array are shown with comparison to a fully sampled 2D array.

Ultrasound Imaging Using the Coherence of Estimated Channel Data.

This article introduces a novel method to estimate the coherence of ultrasound channel data from beamformed radio frequency (RF) data. Estimates of ultrasound channel data are obtained by spatially filtering acquired RF data in the frequency domain. The frequency response of the spatial filters yields outputs similar to frequency domain representations of individual channel signals. This technique performs multiple normalized cross-correlations from the outputs of multiple spatial filters. The coefficients are summed together for each pixel in the coherence-based image. Simulation results using a 64 element 2.5-MHz phased array showed an improvement in contrast-to-noise ratio (CNR) of 67%-93% and a 125%-183% improvement in speckle signal-to-noise ratio (SNR) compared with standard beamformed data. Experimental CNR using a tissue-mimicking phantom showed improvement of 43%-58%, and experimentalSNR improvement was 23%-154%. Comparisons to a previously coherence method, short-lag spatial coherence, are also presented. Preliminary in vivo images of the heart and gall bladder are also shown. This method improves CNR enabling improved visualization of anechoic regions such as cyst and blood vessels.

Mitral Valve Segmentation Using Robust Nonnegative Matrix Factorization.

Analyzing and understanding the movement of the mitral valve is of vital importance in cardiology, as the treatment and prevention of several serious heart diseases depend on it. Unfortunately, large amounts of noise as well as a highly varying image quality make the automatic tracking and segmentation of the mitral valve in two-dimensional echocardiographic videos challenging. In this paper, we present a fully automatic and unsupervised method for segmentation of the mitral valve in two-dimensional echocardiographic videos, independently of the echocardiographic view. We propose a bias-free variant of the robust non-negative matrix factorization (RNMF) along with a window-based localization approach, that is able to identify the mitral valve in several challenging situations. We improve the average f1-score on our dataset of 10 echocardiographic videos by 0.18 to a f1-score of 0.56.

Gated Transmit and Fresnel-Based Receive Beamforming With a Phased Array for Low-Cost Ultrasound Imaging.

Low-cost ultrasound imaging systems are desired for many applications outside of radiology and cardiology departments. By making ultrasound systems smaller and lower cost, the use of ultrasound has spread from these mainstays to other areas of the hospital such as emergency departments and critical care. To further miniaturize and reduce the cost of ultrasound systems, we have investigated novel Fresnel-based beamforming methods to reduce front-end hardware requirements. Previous studies with linear and curvilinear arrays demonstrated comparable imaging performance using Fresnel-based beamforming versus delay-and-sum (DAS) beamforming. In this work, we extend Fresnel-based beamforming to phased arrays with beam steering. To accomplish this in transmit mode, we introduce a technique called a gated transmit beamformer where multicycle bursts are gated using multiplexers. In receive mode, a 64-element 2.5-MHz phased array is broken up into four 16-element subapertures, and each subaperture performs Fresnel beamforming before a final beamforming step is done. Timing errors are inevitable with Fresnel-based beamforming leading to higher sidelobe and clutter levels. To suppress sidelobe and clutter contributions, we also combine this with our previous technique, dual apodization with cross correlation (DAX) to improve contrast. Field II simulations are performed to evaluate spatial resolution and contrast-to-noise ratio and compared to standard DAS beamforming. Fresnel-based and gated transmit beamforming is also implemented using synthetic aperture data from tissue-mimicking phantoms. Lastly, a hardware proof-of-concept (PoC) Fresnel beamformer was designed, assembled, and evaluated with images from tissue-mimicking phantoms and initial in vivo images.

High-frequency ultrasound Doppler system for biomedical applications with a 30-MHz linear array.

In this paper, we report the development of the first high-frequency (HF) pulsed-wave Doppler system using a 30-MHz linear array transducer to assess the cardiovascular functions in small animals. This array-based pulsed-wave Doppler system included a 16-channel HF analog beamformer, a HF pulsed-wave Doppler module, timing circuits, HF bipolar pulsers and analog front ends. The beamformed echoes acquired by the 16-channel analog beamformer were fed directly to the HF pulsed-wave Doppler module. Then the in-phase and quadrature-phase (IQ) audio Doppler signals were digitized by either a sound card or a Gage digitizer and stored in a personal computer. The Doppler spectrogram was displayed on a personal computer in real time. The two-way beamwidths were determined to be 160 microm to 320 microm when the array was electronically focused at different focal points at depths from 5 to 10 mm. A micro-flow phantom, consisting of a polyimide tube with an inner diameter of 127 microm and the wire phantom were used to evaluate and calibrate the system. The results show that the system is capable of detecting motion velocity of the wire phantom as low as 0.1 mm/s, and detecting blood-mimicking flow velocity in the 127-microm tube lower than 7 mm/s. The system was subsequently used to measure the blood flow in vivo in two mouse abdominal superficial vessels, with diameters of approximately 200 microm, and a mouse aorta close to the heart. These results demonstrated that this system may become an indispensable part of the current HF array-based imaging systems for small animal studies.

Spatial Prediction Filtering for Medical Ultrasound in Aberration and Random Noise.

While medical ultrasound imaging has become one of the most widely used imaging modalities in clinics, it often suffers from suboptimal image quality, especially in technically difficult patients with a large amount of fat content that induces severe phase aberration effects and decreases the signal-to-noise ratio. Several researchers have proposed various techniques, which can be broadly categorized as either a phase aberration correction (PAC) technique or a coherence-based imaging technique, to address the challenges in imaging technically difficult patients. Although both families of techniques have shown some success in improving the image quality in the presence of a mild level of phase aberration and/or random noise, they often fail to achieve meaningful improvements in the image quality and, in some cases, even create severe image artifacts. In this paper, we employ an adaptive filtering technique called frequency-space prediction filtering (FXPF), which we recently introduced in ultrasound imaging, to overcome the weaknesses of existing techniques and achieve image quality improvements more effectively under varying levels of phase aberration and random noise. Using simulated and experimental phantom data with varying levels of phase aberration and random noise, we evaluate and compare the performance of FXPF with the most representative technique for each category: nearest-neighbor cross correlation (NNCC)-based PAC and the generalized coherence factor (GCF). Our simulation, experimental phantom, and in vivo results demonstrate that FXPF is highly robust in varying levels of phase aberration and noise, and always outperforms both NNCC-based PAC and GCF in terms of the contrast-to-noise ratio (CNR) and the contrast when both random noise and phase aberration are present.

A novel envelope detector for high-frame rate, high-frequency ultrasound imaging.

This paper proposes a novel design of envelope detectors capable of supporting a small animal cardiac imaging system requiring a temporal resolution of more than 150 frames per second. The proposed envelope detector adopts the quadrature demodulation and the lookup table (LUT) method to compute the magnitude of the complex baseband components of received echo signals. Because the direct use of the LUT method for a square root function is not feasible due to a large memory size, this paper presents a new LUT strategy dramatically reducing its size by using binary logarithmic number system (BLNS). Due to the nature of BLNS, the proposed design does not require an individual LOG-compression functional block. In the implementation using a field programmable gate array (FPGA), a total of 166.56 Kbytes memories were used for computing the magnitude of 16-bit in-phase and quadrature components instead of 4 Gbytes in the case of the direct use of the LUT method. The experimental results show that the proposed envelope detector is capable of generating LOG-compressed envelope data at every clock cycle after 32 clock cycle latency, and its maximum error is less than 0.5 (i.e., within the rounding error), compared with the arithmetic results of square root function and LOG compression.

A low-cost bipolar pulse generator for high-frequency ultrasound applications.

A design of a low-cost bipolar pulse generator for high-frequency (HF) ultrasound applications is presented. The pulse generator can produce N cycle (1-255 cycles) bipolar pulses with center frequency over 60 MHz. The measured pulse amplitude was over 160 Vpp, and the pulse ringing was less than 0.3 Vpp (i.e., signal-to-ring ratio is 55 dB). The pulser can be used in high-frequency ultrasound Doppler and B-mode imaging applications with arrays.

A high-frame rate high-frequency ultrasonic system for cardiac imaging in mice.

We report the development of a high-frequency (30-50 MHz), real-time ultrasonic imaging system for cardiac imaging in mice. This system is capable of producing images at 130 frames per second (fps) with a spatial resolution of less than 50 microm. A novel mechanical sector probe was developed that utilizes a magnetic drive mechanism and custom-built servo controller for high speed and accuracy. Additionally, a very light-weight (< 0.28 g), single-element transducer was constructed and used to reduce the mass load on the motor. The imaging electronics were triggered according to the angular position of the transducer in order to compensate for the varying speed of the sector motor. This strategy ensured the production of equally spaced scan lines with minimal jitter. Wire phantom testing showed that the system axial and lateral resolutions were 48 microm and 72 microm, respectively. In vivo experiments showed that high-frequency ultrasonic imaging at 130 fps is capable of showing a detailed depiction of a beating mouse heart.

Improved contrast for high frame rate imaging using coherent compounding combined with spatial matched filtering.

The concept of high frame rate ultrasound imaging (typically greater than 1000 frames per second) has inspired new fields of clinical applications for ultrasound imaging such as fast cardiovascular imaging, fast Doppler imaging and real-time 3D imaging. Coherent plane-wave compounding is a promising beamforming technique to achieve high frame rate imaging. By combining echoes from plane waves with different angles, dynamic transmit focusing is efficiently accomplished at all points in the image field. Meanwhile, the image frame rate can still be kept at a high level. Spatial matched filtering (SMF) with plane-wave insonification is a novel ultrafast beamforming method. An analytical study shows that SMF is equivalent to synthetic aperture methods that can provide dynamic transmit-receive focusing throughout the field of view. Experimental results show that plane-wave SMF has better performance than dynamic-receive focusing. In this paper, we propose integrating coherent plane-wave compounding with SMF to obtain greater image contrast. By using a combination of SMF beamformed images, image contrast is improved without degrading its high frame rate capabilities. The performance of compounded SMF (CSMF) is evaluated and compared with that of synthetic aperture focusing technique (SAFT) beamforming and compounded dynamic-receive-focus (CDRF) beamforming. The image quality of different beamforming methods was quantified in terms of contrast-to-noise ratio (CNR). Our results show that the new SMF based plane-wave compounding method provides better contrast than DAS based compounding method. Also CSMF can obtain a similar contrast level to dynamic transmit-receive focusing with only 21 transmit events.

2-D array for 3-D ultrasound imaging using synthetic aperture techniques.

A two-dimensional (2-D) array of 256 X 256 = 65,536 elements, with total area 4 X 4 = 16 cm2, serves as a flexible platform for developing acquisition schemes for 3-D rectilinear ultrasound imaging at 10 MHz using synthetic aperture techniques. This innovative system combines a simplified interconnect scheme and synthetic aperture techniques with a 2-D array for 3-D imaging. A row-column addressing scheme is used to access different elements for different transmit events. This addressing scheme is achieved through a simple interconnect, consisting of one top, one bottom single-layer, flex circuits that, compared to multilayer flex circuits, are simpler to design, cheaper to manufacture, and thinner so their effect on the acoustic response is minimized. We present three designs that prioritize different design objectives: volume acquisiton time, resolution, and sensitivity, while maintaining acceptable figures for the other design objectives. For example, one design overlooks time-acquisition requirements, assumes good noise conditions, and optimizes for resolution, achieving -6 dB and -20 dB beamwidths of less than 0.2 and 0.5 mm, respectively, for an F/2 aperture. Another design can acquire an entire volume in 256 transmit events, with -6 dB and -20 dB beamwidths in the order of 0.4 and 0.8 mm, respectively.

Development of a real-time, high-frequency ultrasound digital beamformer for high-frequency linear array transducers.

A real-time digital beamformer for high-frequency (>20 MHz) linear ultrasonic arrays has been developed. The system can handle up to 64-element linear array transducers and excite 16 channels and receive simultaneously at 100 MHz sampling frequency with 8-bit precision. Radio frequency (RF) signals are digitized, delayed, and summed through a real-time digital beamformer, which is implemented using a field programmable gate array (FPGA). Using fractional delay filters, fine delays as small as 2 ns can be implemented. A frame rate of 30 frames per second is achieved. Wire phantom (20 microm tungsten) images were obtained and -6 dB axial and lateral widths were measured. The results showed that, using a 30 MHz, 48-element array with a pitch of 100 microm produced a -6 dB width of 68 microm in the axial and 370 microm in the lateral direction at 6.4 mm range. Images from an excised rabbit eye sample also were acquired, and fine anatomical structures, such as the cornea and lens, were resolved.

Ultrasonic Reverberation Clutter Suppression Using Multiphase Apodization With Cross Correlation.

Despite numerous recent advances in medical ultrasound imaging, reverberation clutter from near-field anatomical structures, such as the abdominal wall, ribs, and tissue layers, is one of the major sources of ultrasound image quality degradation. Reverberation clutter signals are undesirable echoes, which arise as a result of multiple reflections of acoustic waves between the boundaries of these structures, and cause fill-in to lower image contrast. In order to mitigate the undesirable reverberation clutter effects, we present, in this paper, a new beamforming technique called multiphase apodization with cross correlation (MPAX), which is an improved version of our previous technique, dual apodization with cross correlation (DAX). While DAX uses a single pair of complementary amplitude apodizations, MPAX utilizes multiple pairs of complementary sinusoidal phase apodizations to intentionally introduce grating lobes from which an improved weighting matrix can be produced to effectively suppress reverberation clutter. Our experimental sponge phantom and preliminary in vivo results from human subjects presented in this paper suggest that MPAX is a highly effective technique in suppressing reverberation clutter and has great potential for producing high contrast ultrasound images for more accurate diagnosis in clinics.

Real-time rectilinear volumetric imaging using a periodic array.

Current real-time phased array volumetric scanners use a 2-D array to scan a pyramidal volume comprised of many sector scans stacked in the elevation direction. This scan format is primarily useful for cardiac imaging to avoid interference from the ribs. However, a real-time rectilinear volumetric scan with a wider field-of-view (FOV) close to the transducer could prove more useful for abdominal, breast or vascular imaging. In our previous work, a 94 x 94 Mills cross array operating at 5 MHz was fabricated, and the first real-time rectilinear volumetric images were made using a 2-D array and the Duke real-time 3-D scanner. The FOV for the Mills cross was 30 x 8 x 60 mm. Despite reasonable success with the Mills cross array, the array had limitations of poor off-axis sensitivity and a smaller FOV in one direction. To overcome these limitations, a new rectilinear array containing over 65,500 elements was developed with a periodic geometry to increase the FOV to 30 mm x 30 mm x 60 mm and improve the off-axis sensitivity. Images of tissue-mimicking phantoms and the carotid artery in vivo were obtained. In addition, spectral and color flow Doppler results from a pulsatile flow phantom were obtained.

Real-time rectilinear 3-D ultrasound using receive mode multiplexing.

In previous work, we developed two generations of a real-time rectilinear volumetric scanner operating at 5 MHz for abdominal, breast, or vascular imaging using a Mills cross two-dimensional (2-D) array and a rectilinear periodic 2-D array. To improve spatial resolution performance and sensitivity, we developed a new design using 4:1 receive mode multiplexing. With 4:1 multiplexing, the new 65,000 element 2-D array has 4 x 256 = 1024 receivers so that 256 receivers can be used on any image line. The two major benefits of using receive mode multiplexing are an increase in receive sensitivity due to a greater number of receive elements, and a decrease in grating lobe and clutter levels due to increased receive element density. Theoretical simulations and analysis show an increase of about 13 dB in sensitivity compared to our previous work. With these encouraging results, a new 65,000 element 5-MHz, 2-D array having 1024 receivers and 169 transmitters was prototyped. In addition, the multiplexer and control circuitry were designed, built, and interfaced with both the transducer and volumetric scanner. Images of tissue-mimicking phantoms and in vivo targets were obtained. Using a spherical cyst phantom, experimental results showed a +12 dB improvement in signal-to-noise ratio and a +6 dB improvement in contrast compared to our previous work.

Real-time rectilinear volumetric imaging.

Current real-time volumetric scanners use a 2-D array to scan a pyramidal volume consisting of many sector scans stacked in the elevation direction. This scan format is primarily useful for cardiac imaging to avoid interference from the ribs. However, a real-time rectilinear volumetric scan with a wider field of view close to the transducer could prove more useful for abdominal, breast, or vascular imaging. In previous work, computer simulations of very sparse array transducer designs in a rectilinear volumetric scanner demonstrated that a Mills cross array showed the best overall performance given current system constraints. Consequently, a 94 x 94 Mills cross array including 372 active channels operating at 5 MHz has been developed on a flexible circuit interconnect. In addition, the beam former delay software and scan converter display software of the Duke volumetric scanner were modified to achieve real-time rectilinear volumetric scanning consisting of a 30-mm x 8-mm x 60-mm scan at a rate of 47 volumes/s. Real-time rectilinear volumetric images were obtained of tissue-mimicking phantoms, showing a spatial resolution of 1 to 2 mm. Images of carotid arteries in normal subjects demonstrated tissue penetration to 6 cm.

Effects of dual apodization with cross-correlation on tissue harmonic and pulse inversion harmonic imaging in the presence of phase aberration.

Dual apodization with cross-correlation (DAX) is a relatively new beamforming technique which can suppress side lobes and clutter to enhance ultrasound image contrast. However, previous studies have shown that with increasing aberrator strength, contrast enhancements with DAX diminish and DAX becomes more prone to image artifacts. In this paper, we propose integrating DAX with tissue harmonic imaging (THI) or pulse inversion harmonic imaging (PIHI) to overcome their shortcomings and achieve higher image contrast. Compared with conventional imaging, our experimental results showed that DAX with THI allows for synergistic enhancements of image contrast with improvements of more than 231% for a 5-mm pork aberrator and 703% for a 12-mm pork aberrator. With PIHI, improvements of 238% and 890% were observed for the two pork tissue samples. Our results suggest that the complementary contrast enhancement mechanism employed by the proposed method may be useful in improving imaging of technically difficult patients in clinics.

Performance improvement of Fresnel beamforming using dual apodization with cross-correlation.

Fresnel beamforming is a beamforming method that has a delay profile with a shape similar to a physical Fresnel lens. With 4 to 8 transmit channels, 2 receive channels, and a network of single-pole/single-throw switches, Fresnel beamforming can reduce the size, cost, and complexity of a beamformer. The performance of Fresnel beamforming is highly dependent on focal errors resulting from phase wraparound and quantization of its delay profile. Previously, we demonstrated that the performance of Fresnel beamforming relative to delayand- sum (DAS) beamforming is comparable for linear arrays at f-number = 2 and 50% bandwidth. However, focal errors for Fresnel beamforming are larger because of larger path length differences between elements, as in the case of curvilinear arrays compared with linear arrays. In this paper, we present the concept and performance evaluation of Fresnel beamforming combined with a novel clutter suppression method called dual apodization with cross-correlation (DAX) for curvilinear arrays. The contrast-to-noise ratios (CNRs) of Fresnel beamforming followed by DAX are highest at f-number = 3. At f-number = 3, the experimental results show that using DAX, the CNR for Fresnel beamforming improves from 3.7 to 10.6, compared with a CNR of 5.2 for DAS beamforming. Spatial resolution is shown to be unaffected by DAX. At f-number = 3, the lateral beamwidth and axial pulse length for Fresnel beamforming with DAX are 1.44 and 1.00 mm larger than those for DAS beamforming (about 14% and 21% larger), respectively. These experimental results are in good agreement with simulation results.

Synergistic enhancements of ultrasound image contrast with a combination of phase aberration correction and dual apodization with cross-correlation.

Dual apodization with cross-correlation (DAX) is a novel adaptive beamforming technique which utilizes two distinct apodization functions in suppressing side lobes and clutter. Previous studies have shown that the performance of DAX in minimizing the effects of phase aberration diminishes with increasing aberrator strength. To achieve greater improvement in image contrast, we propose, in this paper, to combine DAX with a phase aberration correction algorithm based on nearest-neighbor cross-correlation (NNCC). Our simulation and experimental results presented in this work showed that the proposed method allows for synergistic enhancements of image contrast and achieves greater improvement in image quality than using DAX alone or phase aberration correction alone in the presence of weak and strong aberrators. Compared with standard delay-and-sum (DAS) beamforming, using the proposed method on simulated data with weak and strong aberrations increased the contrast-to-noise ratio (CNR) values from 4.10 to 10.96 and from 1.69 to 9.80, respectively. Experimental results were obtained using pork tissues of 4 and 10 mm thickness and a tissue-mimicking phantom. The CNR values increased from 3.74 to 9.72 for the 4-mm pork aberrator and from 1.27 to 8.17 for the 10-mm pork aberrator.