3-D H-scan ultrasound imaging of relative scatterer size using a matrix array transducer and sparse random aperture compounding

Abstract

H-scan ultrasound (US) is a high-resolution imaging technique for soft tissue characterization. By acquiring data in volume space, H-scan US can provide insight into subtle tissue changes or heterogenous patterns that might be missed using traditional cross-sectional US imaging approaches. In this study, we introduce a 3-dimensional (3-D) H-scan US imaging technology for voxel-level tissue characterization in simulation and experimentation. Using a matrix array transducer, H-scan US imaging was developed to evaluate the relative size of US scattering aggregates in volume space. Experimental data was acquired using a programmable US system (Vantage 256, Verasonics Inc, Kirkland, WA) equipped with a 1024-element (32 × 32) matrix array transducer (Vermon Inc, Tours, France). Imaging was performed using the full array in transmission. Radiofrequency (RF) data sequences were collected using a sparse random aperture compounding technique with 6 different data compounding approaches. Plane wave imaging at five angles was performed at a center frequency of 8 MHz. Scan conversion and attenuation correction were applied. To generate the 3-D H-scan US images, a convolution filter bank (N = 256) was then used to process the RF data sequences and measure the spectral content of the backscattered US signals before volume reconstruction. Preliminary experimental studies were conducted using homogeneous phantom materials embedded with spherical US scatterers of varying diameter, i.e., 27 to 45, 63 to 75, or 106 to 126 μm. Both simulated and experimental results revealed that 3-D H-scan US images have a low spatial variance when tested with homogeneous phantom materials. Furthermore, H-scan US is considerably more sensitive than traditional B-mode US imaging for differentiating US scatterers of varying size (p = 0.001 and p = 0.93, respectively). Overall, this study demonstrates the feasibility of 3-D H-scan US imaging using a matrix array transducer for tissue characterization in volume space.

1. Introduction

Cancer accounts for nearly 600,000 deaths every year [1]. Traditional chemotherapy is not effective for all patients, which implicates that detection of any early response to drug treatment (or lack thereof) is critically important. To have a better understanding of cancer treatment protocols and to improve development of more personalized medicines, monitoring the tissue microstructure over different distance scales is an evolving clinical procedure. Noninvasive medical imaging techniques offer promising solutions as they can provide information from the entire tumor burden rather than from a few needle biopsy samples that pose issues like inadequate tissue sampling. The exploration of such methods is motivated by the congruence between parameters from imaging measurements and tumor histology. Moreover, medical imaging can typically be safely repeated over time and facilitates longitudinal tumor monitoring.

It is widely understood that diseased tissues like cancerous lesions are heterogenous in composition. This disease heterogeneity has implications in pathogenesis, diagnosis, and therapeutic management [2]. This suggests that a 2-dimensional (2-D) cross-sectional tissue assessment may not sufficiently represent conditions throughout the entire diseased tissue burden. While magnetic resonance (MR) allows imaging in volume space [3], there are some inherent safety issues. These include the effects of high magnetic fields and radiofrequency (RF) pulses on the body and on implanted devices. Claustrophobia and hearing loss are additional risks of clinical MR examinations. With the development of 3-dimensional (3-D) transducer technology, ultrasound (US) imaging can now provide complementary tissue information in volume space with isotropic measures [4]. While the earlier mechanically-driven wobbler transducers are sufficient for US-based tissue measurements [5], device bulkiness and issues like position tracking introduce control errors that make it relatively difficult to record tilt angle of each imaging plane within the volume acquisition. Depending on the region-of-interest (ROI) and transducer sweep speed, US imaging using mechanically driven wobbler transducers is limited to a prescribed arc at volume rates on the low Hz scale [6]. Conversely, a matrix array transducer is more flexible allowing electronic beamsteering and interrogation of a volume of tissue at much higher rates on an order of 100s per sec.

This paper explores the development and testing of a novel 3-D US-based approach for performing tissue measurements. Termed H-scan US imaging, we envision this technology can be used for noninvasive characterization of tissue microstructure. The main contributions of this study are as follows.

1. We detail use of simulations to help validate H-scan US imaging theory and guide future clinical translation of our new technique.

2. We introduce characterization of tissue scattering in volume space using an H-scan US imaging system equipped with a 1024-element matrix array transducer technology.

3. We determined an optimal sparse random aperture compounding mode for use with ultrafast plane wave imaging.

4. We evaluated performance of 3-D H-scan US imaging using phantom materials embedded with US scatterers of known size and distribution.

2. Related work

Tissue characterization using US exploits knowledge about the physical interaction of US with biological tissues to differentiate states of health and disease. Several US-based techniques have been investigated for the purpose of soft tissue characterization and include attenuation and backscatter coefficient estimation [7], tissue elastography [8], shear wave speed and attenuation quantification [9], and speckle pattern analysis [10]. Each of these techniques provide different insight into the underlying target tissue structures. To simplify tissue characterization in real-time, a novel high-resolution US-based modality has emerged. Dubbed H-scan US (where the ‘H’ stands for Hermite or hue) imaging, this technique uses a matched filter methodology to depict the relative size of US scatterer aggregates [11]. In approach, matched filters are created for different types of tissue scatterers and maximum outputs are color coded to allow visualization of local US scatterer size. Previous studies have demonstrated that H-scan US imaging can distinguish subtle changes at the cellular level that are otherwise not visible in the traditional B-scan US images [12].

The recent development of programmable US systems equipped with matrix array transducers has created new opportunities for volumetric imaging with high spatiotemporal resolution [13]. For example, use of plane or diverging waves with these transducer technologies allow acquisition of volume data at rates up to 20 kHz [14]. When coupled with advanced shear wave elastography techniques, the prospect of 3-D tissue characterization is becoming a reality [15]. Motivation for the research presented here was to introduce voxel-level characterization of US scattering using an H-scan imaging system equipped with a 32 × 32 element matrix array transducer. Use of the full 2-D array requires each transducer element to be connected to an independent channel of an US system, which is only achievable using several synchronized scanners working in parallel [16]. Therefore, an alternative strategy was necessary. To that end, newer programmable US system designs allow researchers to reduce the number of channels needed to control a matrix array transducer while offering flexibility in sequence programming. When combined with subaperture multiplexing these sparse matrix arrays allow volumetric US imaging [17].

3. Methods

3.1. Simulation program

A custom US simulation program was developed in MATLAB (MathWorks Inc, Natick, MA) using the Field II open-source toolkit [18]. This time domain model allows the study of beam patterns and properties of US images from use of 2-D matrix array transducers. A 1024-element (32 × 32 element) matrix array transducer was implemented to match the spatial configuration and layout as depicted in Fig. 1. The center frequency f₀ was set to 8 MHz with an effective bandwidth σ of 60% sampling frequency of 32 MHz. The spatial impulse response was defined for the simulated transducer and elements were excited using a modified function during transmission to match experimental setting (detailed below). Briefly, the simulated backscattered US signal e(t) was generated by convolving the US pulse-echo (one-way) response h(t) of the transducer with a medium consisting of randomly distributed US scatterers as follows [19]:

$$e(t)=h(t)\ast b(\alpha )s(t)$$ (1)

where * denotes convolution and the b(α)s(t) product describes the local scattering function. To simulate the impact of different-sized US scatterers, the scattering function was further defined to have the following distribution [20]:

$$b(\alpha )=\frac{8R_S^2{k}^4{\alpha }^3{e}^{\frac{-k^2{\alpha }^2}{4}}}{\sqrt{\pi }}$$ (2)

where α denotes scatterer size (radius), $R_s=R_i\sqrt{\left(N{\alpha }^3+N^2{\alpha }^6\right)}$, R_i is the amplitude reflection coefficient, N is the scatterer density, k = 2πf₀/c, and c is the speed of sound. To match the transmitted US signal generated by our experimental platform, the simulated one-way response was defined as [21]:

$$h(t)=\frac{e^{-\frac{t^2}{2{\sigma }^2}}}{\sigma \sqrt{(2\pi )}}\left(4t^2-1\right).$$ (3)

A pulse-echo US system with a round trip (transmit-receive) impulse response is then approximated as:

$$h(t)\ast b(\alpha )s(t)=A_{0}b(\alpha )\frac{e^{-\frac{t^2}{2{\sigma }^2}}}{\sigma }{H}_n(t)\approx A_{0}b(\alpha )G{H}_n(t)$$ (4)

where A₀ is an amplitude scaling constant. Note the n^th-order Hermite function H_n(t) was defined by Pierre-Simon Laplace [22] and successive differentiation of this polynomial yields the following:

$$H_n(t)={(-1)}^n{e}^{t^2}\frac{d^n}{d{t}^n}G(t),n=0,1,2,\dots ,-\infty <t<\infty$$ (5)

where $\frac{d^n}{d{t}^n}G(t)$ is the n^th order derivative for a Gaussian pulse,$G(t)=e^{-t^2}$. A previous study has shown that a pulse-echo US system with f₀ = 8 MHz has a round trip impulse response that is very similar to a 4^th-order Gaussian-weighted Hermite polynomial function, GH₄(t) [23].

Fig. 1.

Layout of the 2-D sparse matrix array ultrasound (US) transducer with red and black circles showing the selected elements. Random aperture compounding is performed by activating a sequence of four 256-element mutually exclusive random apertures in reception. The pitch between consecutive elements in the x and y directions is 0.3 mm. The rows at 9, 18, and 27 were intentionally left blank during manufacturing.

In practice, different GH_n functions can be used to produce a bandpass filter bank to isolate frequency information of interest. In fact, if we assume that the GH_n(t) function resembles a general broadband US pulse and that a pulse-echo system has a round trip impulse response of 𝐴₀GH_n(t), then the backscattered US signal can be approximated as:

$$e_l(t)=A_0{Z}_{0}G{H}_n(t)\left(t-t_o\right)$$ (6)

$$e_m(t)=A_0{Z}_{0}G{H}_{n+1}(t)\left(t-t_o\right)$$ (7)

and

$$e_s(t)=A_0{Z}_{0}G{H}_{n+2}(t)\left(t-t_o\right)$$ (8)

for a relatively large, moderate, and small scatterer or incoherent aggregate of scatterers, respectively. Under the assumption of minimal spatial variation, the constant Z₀ is a term related to the derivative of the acoustic impedance in the direction of the propagating US pulse [23]. For all simulations, material dimensions were fixed at a volume of 30 × 30 × 30 mm. Within that volume, 2 × 10⁵ scatterers were pseudo-randomly positioned to allow simulation of US images with speckle patterns [24]. An example 3-D phantom material embedded with a distribution of spherical scatterers is illustrated in Fig. 2.

Fig. 2.

Pseudo-random distribution of US scatterers and phantom material used in simulation.

3.2. Sparse synthetic aperture beamformer

A 2-D matrix array transducer was implemented using the gridded layout as shown in Fig. 1. Each element was set to be the same size in both the lateral and elevational dimensions. Before beamforming the raw RF data, sparse synthetic apodization was performed by adjusting the receive apodization functions to account for the missing transmit elements while maintaining spatial resolution [25]. Next, 3-D delay-and-sum beamforming was applied to the simulated RF data to form the final US image reconstructions. In this approach, the expected delay is calculated from transducer element position to each pixel location on the reconstruction grid. As illustrated in Fig. 3, the value for each pixel in the reconstruction grid is obtained by summing the signal traces for all elements at the time delay associated with that pixel location [26]. Given the signal emitted by an arbitrary transmit element at location (X_tx, Y_tx, Z_tx) is h, and considering that the target function at a given point location is (X_s, Y_s, Z_s) in the spatial domain, the beamformed RF data e_RF is obtained by the following expression [27]:

$$e_{RF}(rx,t)=\sum _{s}e(t-\delta )$$ (9)

where $\delta =\frac{1}{c}\left(\sqrt{{Z_s}^2+{\left(X_s-X_{tx}\right)}^2+{\left(Y_s-Y_{tx}\right)}^2}+\sqrt{{Z_s}^2+{\left(X_s-X_{rx}\right)}^2+{\left(Y_s-Y_{rx}\right)}^2}\right)$ is the round-trip delay of the received backscattered US signal e(t). Sidelobe magnitude was suppressed by applying an amplitude-weighted Hanning window on the aperture data.

Fig. 3.

Array geometry used for processing the backscattered US signals. By calculating the distance between the US scatterer and receive elements, weighted delay-and-sum beamforming was performed across at depth to provide the beamformed radiofrequency (RF) data.

3.3. H-scan US imaging

Experimental studies were performed using a programmable US scanner (Vantage 256, Verasonics Inc, Kirkland, WA) equipped with a 1024-element (32 × 32) matrix array transducer (Vermon, Tours, France). Plane wave imaging was performed at a center frequency of 8 MHz. The 2-D matrix array was divided into 4 aperture segments (see Fig. 1). This configuration allows for direct volume scanning by electronically interrogating a ROI and acquisition of a pyramidal volume of US data. A 2-D array (termed complementary) was designed by activating all elements for transmission (1 Tx) and use of 4 sets of complementary sparse random apertures for reception (4 Rx) to reduce the collected data size and side lobes [28]. Thus, a total number of transmit and receive events (i.e., 1 Tx × 4 Rx) were generated to interrogate the entire volume. A map of the transmitting and receiving elements was stored as a matrix for simulation validation. Coherent spatial angular compounding was implemented by successively steering and summing overlapping plane wave transmissions using five equally spaced angles in the ± 20° range. In addition, different data acquisition methods were designed to study the robustness of 3-D H-scan US imaging and potential improvement in image quality: (1) synthetic, (2) sparse, (3) random, (4) sparse complimentary (sparseComp), and (5) sparse transmit (sparseTx) apodization. By switching sub-banks, synthetic apodization was performed using a full reception with 1024 active elements for the full array transmissions, whereas sparse apodization was performed by using the same optimally selected sparse random element map in both transmission and reception, respectively. Two different sets of sparse random element maps were used for transmitting and receiving during random apodization. Using sparse elements for transmitting and sparse random aperture compounding for receiving, the sparseComp imaging scheme was designed. Finally, sparseTx was performed using 256-element sparse random aperture for transmit while using all 1024 elements for receiving by rapidly switching between four 256-element sub-banks after a single transit event.

H-scan US image processing was applied to each backscattered US signal [29]. Gaussian bandpass functions with 256 distinct center frequencies and bandwidth of 0.02 MHz were implemented in the 5.6 to 10.4 MHz range to measure the frequency content of the received signals [30]. The signal envelope for each of the filtered and compounded data sequences were then calculated using a Hilbert transformation. Thereafter, the best matched filter index at each voxel location was selected by finding the maximum value among the 256 different convolution results to generate the final H-scan US image. A colormap scheme is used to enhance visualization, whereby the relative strength of these filter outputs was color coded where the lower frequency signals were assigned redder values and the higher frequency components were bluer. In general, lower frequency content is generated from larger scattering structures while higher frequency backscattered US signal content is generated by an US wave interacting with small scatterers at a scale below the wavelength of the US transmit pulse. A schematic diagram highlighting the data processing strategy used for 3-D H-scan US imaging is illustrated in Fig. 4.

Fig. 4.

Schematic diagram highlighting the data processing strategy used for 3-D H-scan US imaging. After local attenuation correction, a bandpass filter bank is applied to the RF data for measuring the relative strength of the received signals before volume reconstruction.

3.4. Phantom material fabrication

A series of tissue-mimicking phantom materials were prepared to contain spherical US scatterers of varying size and distribution. Each phantom contained a base mixture of 75 g of gelatin (300 Bloom, Sigma Aldrich, St. Louis, MO), spherical US scatterers with diameters in the range of (1) 27 to 45 μm, (2) 63 to 75 μm, or (3) 106 to 125 μm (CoSpheric LLC, Santa Barbara, CA) at a 0.5% concentration, and 1 L of degassed water. After heating to 65 °C to promote gelatin crosslinking, the solution was poured into a solid mold and allowed to cool in a refrigerator overnight. The final material size was 30 × 30 × 30 mm. All phantom experiments were performed at room temperature after solidification. A reference calculation of the true scatterer size was performed from optical microscopy images of diluted raw particle samples.

3.5. Acoustic output measurements

The acoustic output of the US system was measured using a calibrated hydrophone setup (AIMS III, Onda Corp, Sunnyvale, CA). The system consists of a large degassed water tank and stepper motors to precisely control hydrophone movement as it spatially records the peak negative pressure parameter from a fixed transducer. Orthogonal planes of data were collected to form the final 3-D US beampattern.

3.6. Statistical analysis

For each experimental group, US image intensity was summarized as mean ± standard deviation. Spatial variance was measured throughout the entire volume space to evaluate the different US data acquisition approaches. Data deviation between each measurement was used to reflect robustness. To evaluate the impact of scatterer size on US image intensity, a one-way analysis of variance (ANOVA) test was performed. A p-value less than 0.05 was considered statistically significant. All analyses were performed using Prism 9.0 (GraphPad Software, San Diego, CA).

4. Results

The US beampattern obtained from the synthetic aperture (full flash) and 256-element sparse array transmission as measured by a calibrated hydrophone system with corresponding simulation results are shown in Fig. 5. A customized system impulse with a matched center frequency was used in simulation to have a comparable resolution as the 2-D matrix array transducer from the experimental studies. The comparison of simulated and experimental acoustic field also demonstrates that plane waves were successfully generated.

Fig. 5.

Simulated (top) and experimental (bottom) hydrophone measurements of the US beampattern from use of a sparse (left) and synthetic (right) aperture configuration. All images are normalized to the respective maximum.

Digital microscopy images of the source US scatterers used for phantom construction were obtained and a set of image processing algorithms calculated actual size and distribution. These US scatterer sizes were then used in the phantom simulation for validation. Note that the scatterers used in simulation and experimentation were similarly distributed over equivalent ranges. The segmentation approach for quantitative measurement of US scatterer size was done using digitized microscope images. After thresholding the image to identify pixels associated with scatterer location, an active contour was used for complete segmentation of each scatterer. Next, morphological operations were used to improve boundary definition. The scatterer boundaries obtained by the proposed segmentation technique were computed throughout the image. A histogram of US scatterer diameter measurements is presented in Fig. 6. To simplify measurements from the three different scatterer populations, scatterer size was assumed to be Rayleigh-distributed and used accordingly during the US material simulation (curve-fitting correlation value, R² > 0.95).

Fig. 6.

Summary of the segmentation approach for quantitative measurement of US scatterer size from digitized microscope images of different-sized spherical microparticles (top left). After thresholding the image to identify pixels associated with scatterer location, an active contour was used for object segmentation (top middle) before use of morphological operations to improve boundary definition (top right). The estimated size of the US scatterers was measured to be 42.6 ± 2.6, 70.8 ± 2.2, and 114.8 ± 4.1 μm (bottom).

3-D H-scan US imaging was performed to assess impact of the various apodization methods. Inspection of the reconstructed H-scan US images (orthogonal views) in Fig. 7 reveals a subtle intergroup color variation with use of different apodization approaches (p > 0.29). However, the proposed complimentary apodization technique demonstrated a marked lower spatial variation (± 3.7 × 10⁻³) versus the synthetic (± 1.1 × 10⁻¹), sparse (± 1.2 × 10⁻¹), random (± 1.3 × 10⁻¹), sparseComp (± 3.3 × 10⁻²), and sparseTx (± 3.5 × 10⁻²) approaches. Given improved performance and using the complimentary apodization technique, simulated B-scan and H-scan US images from phantom materials with scatterer sizes of 42.6 ± 2.6, 70.8 ± 2.2, and 114.8 ± 4.1 μm are presented in Fig. 8. These results highlight the progressive change in the backscattered US signal as the cross-sectional diameter of the individual scatterers in aggregate are increased. Analysis of this data using the convolutional filter bank reveals a progressive red color shift (with a diminished strength of blue) as the size of the US scatterers increases. These findings help validate H-scan US theory whereby the US signal from larger scatterers dominates the red channel and that from smaller scattering objects dominates the blue channel. B-scan and H-scan US images were reconstructed using voxel-based image reconstruction and the spatial distribution of the US scatterers could be displayed and visualized throughout the entire volume space.

Fig. 7.

Comparison of different apodization methods with various transmission and reception strategies for performing 3-D H-scan US imaging, namely, synthetic, sparse, random, complimentary, sparse complimentary (sparseComp), and sparse transmission (sparseTx) apodization. Note H-scan US image intensity is consistent for all the different groups.

Fig. 8.

Simulated and experimental B-scan and H-scan US images in volume space from a series of phantom materials containing different sized US scatterers. Note the progressive red color shift as US scatterer size increases.

All simulated and experimental B-scan and H-scan US results were summarized as the mean image intensity and plotted as a function of US scatterer size, Fig. 9. Note that changes in scatterer size produce relative changes in both B-scan and H-scan US images, which was consistent in both simulated and experimental data. Further inspection of this data reveals that the variation of scatterer size is better reflected in the H-scan US images compared to the matched B-scan US image (p < 0.01). These findings highlight the sensitivity of H-scan US imaging to changes in scatterer size and feasibility for voxel-level tissue classification.

Fig. 9.

Mean image intensity from registered B-scan and H-scan US images collected during both simulation and experimentation. Note that changes in US scatterer size have a greater impact on the H-scan US images compared to B-scan US results. A * indicates p < 0.05.

5. Discussion

In this study we investigated H-scan US image quality in volume space using a matrix array transducer. Due to the complexity of implementing 2-D array systems, it was particularly important to develop an accurate and efficient simulation model to investigate how image quality varies with transducer design and implementation. To that end, simulated H-scan US results were used to evaluate different transmit and receive strategies, which were then down selected and validated by experimental findings. Both demonstrate feasibility of using US as an imaging modality for volumetric tissue characterization. Analyzed data exhibited a high resemblance between simulated and experimental data for both B-scan and H-scan US imaging. Compared to previous H-scan US studies that collected cross-sectional images [31–34], data acquired throughout the entire volume (3-D space) enables multi-planar visualization. Although US systems with matrix array transducers and use of random apertures have been well studied to reduce the number of receive data channels, this research demonstrates the first use for H-scan US imaging. The proposed apodization method produced quality images while maintaining high volume rates. Moreover, the proposed method provides spatial information that could highlight US scattering behavior among different tissue microstructures. Furthermore, the 3-D H-scan imaging technology was accomplished by using a 2-D matrix array transducer and without mechanically sweeping a linear array transducer. This improvement could potentially expand the use of the 3-D H-scan US technique when rapid processing is needed to ensure real-time imaging. Additionally, volumetric H-scan US exhibits less variance compared to planar measurements due to an increased sample size and statistical averaging. Future work will compare H-scan US measures to those obtained using the more established quantitative US approaches for tissue characterization.

The 2-D sparse array H-scan US imaging technique was theoretically investigated and validated using experimental measures. A limited number of receive pairs allows for a reliable visualization in volume space, which has made 3-D H-scan US imaging with high spatiotemporal resolution possible [28]. A comparison of different acquisition techniques demonstrated that spatial variance of two complimentary random apertures affords a slight advantage (lower values) over the full array when imaging uniform phantom materials. This may be due to the presence of noise during data acquisition that is reduced by averaging the multiple random apertures. Additionally, random apertures were optimally selected from the full array using an optimization algorithm, which can potentially obtain comparable or even better image quality than with use of a larger number of transducer elements. Conversely, 3-D H-scan US images had higher spatial variance when the acquisition is performed using a sparse aperture. This is in agreement with a previous study that used the sparse-random-aperture compounding technique and found improved main-lobe-to-side-lobe ratios of 2.9 ± 0.5 and 1.5 ± 0.7 dB when compared to sparse and full aperture methods, respectively [35]. Compared to a previous H-scan US study [5], the complimentary apodization technique introduced herein can reduce the data acquisition time and achieve real-time H-scan US imaging. Random selection of transducer elements during apodization can significantly reduce the impact of grating lobes and can provide higher quality US images [36]. In another study, a fast 3-D US imaging approach with improved temporal resolution was developed using a novel image acquisition sequence [37]. Multiple datasets were sequentially acquired and synchronized for structural visualization of an ex vivo porcine eye at high resolution. This is one technique that can potentially improve US imaging system performance and functionality to achieve sufficiently high temporal resolution. A potential challenge is that simultaneous control of too many transducer elements using a high number of transmit-receive events would dramatically increase power consumption in addition to system and computational costs [38]. By contrast, our study was conducted by optimally selecting four sets of mutually exclusive 256-element apertures during receive events that can considerably reduce the computational burden while maintaining a sufficiently high data processing speed.

Analysis of our US data revealed that image resolution in the axial dimension was high whereas resolution in the lateral and elevational dimensions the spatial resolution was relatively lower. This is in part due to the missing element rows in the transducer assembly, which increases the size of the side lobes and impacts US image quality in those directions. A more advanced beamformer solution could help improve the existing H-scan US image quality. Sparse array design approaches (i.e., random approach, linear programming, or sparse periodic layout, etc.) can also be used to reduce grating lobes by optimally selecting the active transducer elements [39]. Alternatively, high-volume-rate 3-D US imaging can be achieved using synthetic aperture sequential beamforming with chirp-coded excitation [40]. Another limitation of our study is the number of transmit and receive events is higher than use of sparse apodization with one aperture. This led to improved image quality but at the expense of a reduced volume acquisition rate. If higher H-scan US volume acquisition rates are preferred, sparse apodization with one aperture is a logical solution. Impending work will explore use of advanced element apodization and coded excitations techniques in addition to a more complex phantom study to improve the robustness of 3-D H-scan US imaging.

6. Conclusions

The purpose of this paper was to introduce and evaluate a 3-D H-scan US imaging system and method for voxel-level tissue characterization. Experiments were conducted using tissue-mimicking phantom materials embedded with micrometer-sized US scatterers of varying size. Both simulated and experimental results revealed that 3-D H-scan US images have low spatial variance when tested with homogeneous materials. Furthermore, H-scan US was found to be considerably more sensitive than traditional B-mode US imaging for differentiating US scatterers of varying size. Research presented herein provides additional validation of 3-D H-scan US imaging using a matrix array transducer and will help guide future development of this promising modality and translation to preclinical and clinical studies like characterization of cancer tissue and monitoring early response to drug treatment.

1. A simulation study was performed to demonstrate feasibility of volumetric H-scan ultrasound (US) imaging using a 32 × 32 element matrix array

2. Different apodization methods were implemented and evaluated for maximal H-scan US image quality

3. Sparse volumetric H-scan US imaging technique was shown to differentiate acoustic scatterers of varying size