Dynamic mode decomposition technique for doppler-based imaging of cavitation bubbles in pulsed high intensity focused ultrasound

Abstract

Dynamic mode decomposition technique for doppler-based imaging of cavitation bubbles in pulsed high intensity focused ultrasound is described. A method for imaging transient cavitation bubbles in a tissue of a patient using ultrasound imaging includes: generating a burst of therapy ultrasound waves by a focused therapy transducer; and generating cavitation bubbles by the burst of therapy ultrasound waves at a target region that is a focal region of the tissue. After generating a plurality of the gas bubbles at the target region and within a time period of the dissolving process, imaging ultrasound pulses are generated by an imaging probe. Generating ultrasound images as B-scan images based on returning ultrasound signals. The filtering is configured for isolating representations of dissolving gas bubbles from representations of surrounding tissue, and determining spatial and temporal attributes of the sub-mm scale gas bubbles.

Description

[0001] DYNAMIC MODE DECOMPOSITION TECHNIQUE FOR DOPPLER-BASED IMAGING OF CAVITATION BUBBLES IN PULSED HIGH INTENSITY FOCUSED ULTRASOUND CROSS-REFERENCE TO RELATED APPLICATION This application claims the benefit of U.S. Application No. 63 / 554,368, filed on February 16, 2024, the disclosure of which is hereby incorporated by reference in its entirety. STATEMENT OF GOVERNMENT LICENSE RIGHTS This invention is made with government support under Grant Nos.7R01CA154451 and R01EB023910, awarded by the National Institutes of Health. The government has certain rights in the invention. BACKGROUND Pulsed high-intensity focused ultrasound (pHIFU) is a therapeutic ultrasound approach in which sparse de novo inertial cavitation is induced in tissue without the introduction of exogenous contrast agents, (e.g. microbubbles), resulting in mild mechanical disruption of the targeted tissue. This regime is explored in conjunction with several clinical applications, including tumor permeabilization to enhance passive diffusion of systemically administered drugs, and inducing changes in tumor and benign tissue microenvironment. The pHIFU treatments are commonly performed under real-time ultrasound (US) imaging guidance, whereas conventional B-mode imaging with a probe integrated coaxially into the HIFU transducer is typically used for treatment planning and targeting. However, B-mode US has low sensitivity for detection of cavitation activity during treatment because of the transient nature of the bubbles and their rapid dissolution after HIFU pulse. Furthermore, even in cases where cavitation can be discerned in B-mode 1 3915-P1337WO.UW images as a hyperechoic region, conventional methods do not provide quantitative metric of cavitation or the resulting tissue disruption. Conversely, passive methods, including passive cavitation detection (PCD) and passive acoustic mapping (PAM), quantify broadband acoustic emissions resulting from cavitation collapses (i.e. inertial cavitation) during HIFU pulses. These passive methods provide a sensitive quantitative metric of the cavitation activity, but not of the resulting tissue damage, and the spatial information on cavitation occurrence is limited (e.g. axial resolution in PAM). Accordingly, systems and methods for improved ultrasound imaging are still needed. SUMMARY This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This summary is not intended to identify key features of the claimed subject matter. Bubble Doppler imaging is an active ultrasound method that is found to be more sensitive than B-mode and to provide better spatial resolution than PAM in mapping bubbles induced by pHIFU in a tissue of a patient. In Bubble Doppler each HIFU pulse is followed by an ensemble of planewave Doppler pulses. Those pulses encounter a distribution of residual cavitation bubbles that are rapidly (i.e. within milliseconds or the duration of the ensemble) changing due to dissolution. Doppler processing algorithm interprets those changes as motion, and thus the distribution of bubbles is displayed as an area of high Doppler power. The velocity associated with this area of high Doppler power is typically randomly fluctuating around zero, as the bubbles are not actually moving, only changing in size. This creates a problem with conventional Doppler signal processing to remove clutter from stationary or slowly moving tissue: signal from nearly stationary yet temporally transient cavitation bubbles may be inadvertently filtered out. Furthermore, although Doppler power is found to correlate with PCD-derived broadband emissions, it does not provide any quantitative information on the cavitation bubble parameters, such as size or dissolution time, that could be linked to the level of tissue disruption. 2 3915-P1337WO.UW One possible wall filter is the infinite impulse response (IIR) high pass filter, which eliminates low frequency components from signals in the Doppler ensemble domain (i.e. slow time). This filter is effective when tissue motion is considerably slower than the motions of interest (e.g. blood flows speed vs. surrounding tissue motion), and when prior knowledge about the appropriate cut-off frequency is available. However, the design of high pass filter and the cut-off frequency selection can be challenging with pHIFU-induced transient bubbles imaging due to the limited temporal sampling points during the bubbles’ lifespan. Moreover, the quality of IIR filter can be degraded by transient clutter signal from HIFU reverberation. Another possible wall filter in blood flow imaging is singular value decomposition (SVD) filter. The SVD filter decomposes a matrix of beamformed images, where each column is reshaped from an image corresponding to a Doppler ensemble pulse, into spatiotemporally coherent patterns (i.e. modes). These modes are orthogonal to one another and characterized by singular values that indicate the dominance of each mode. Filtering is achieved by reconstructing the image using a selective set of modes corresponding only to the tissues and motion of interest. The primary assumption underlying SVD filtering is that the amplitude of the clutter signal from background tissue consistently exceeds that from scatterers of interest, therefore, the modes corresponding to several largest singular values are removed. However, the amplitude of the backscattered signals from de novo cavitation bubbles can be comparable or even larger than those from tissue, making a priori mode selection challenging. Furthermore, because the residual bubbles typically dissolve within a few milliseconds after the HIFU pulse, a limited number of Doppler ensemble pulses can be acquired to capture the bubbles, resulting in degradation of the quality of SVD filter. In some embodiments, the inventive technology is directed to using of dynamic mode decomposition (DMD) filtering as a technique to address the limitations of Bubble Doppler signal processing, namely, by improving the reliability and sensitivity of detecting and mapping the cavitation bubbles and by providing quantitative information on their parameters. Therefore, DMD is a data-driven technique to analyze the dynamics of complex phenomena. This technique originated in fluid dynamics and is applied first toward understanding the coherent structures and patterns of high dimensional complex 3 3915-P1337WO.UW fluid flows, but it was then extended to various other fields including robotics control, imaging processing, neuroscience, plasma physics, and finance. DMD filtering decomposes time-series data into a reduced-order set of spatial and temporal modes (also referred to as spatiotemporal modes), each of which consists of spatially coherent patterns and having the same behavior in time. While both SVD and DMD extract spatiotemporal information from the time-series snapshots, significant distinctions exist between the two. The spatial modes of SVD are not affected by the temporal order or arrangement of the input snapshots. That is, SVD treats the input data more like a collection of images, focusing on its intrinsic structure and relationships between the individual images, rather than the temporal order in which the input data is collected. In contrast, DMD considers and captures both the intrinsic structure of the snapshots and their physically meaningful behavior over a span of time. In the case of pHIFU-induced gas bubbles, the DMD captures rapid decay of the bubble structures over a span of time. In some embodiments, the DMD modes can be assumed to represent quasi- harmonic motion with exponential growth or decay in amplitude. Thus, this type of DMD implementation generally matches the temporal behavior of ultrasound imaging signals backscattered from dissolving bubbles. In one example, a method for imaging transient cavitation bubbles in a tissue of a patient using ultrasound imaging includes: generating a burst of therapy ultrasound waves by a focused therapy transducer; and generating cavitation bubbles by the burst of therapy ultrasound waves at a target region that is a focal region of the tissue. The cavitation bubbles are sub-mm scale gas bubbles, where, in absence of additionally generated therapy ultrasound waves, the gas bubbles undergo a dissolving process. The method also includes, after generating a plurality of the gas bubbles at the target region and within a time period of the dissolving process, generating imaging ultrasound pulses by an imaging probe. The method also includes acquiring returning ultrasound signals as imaging planes, where the imaging planes at least partially correspond to a space occupied by the gas bubbles; composing ultrasound images as B-scan images based on the returning ultrasound signals; filtering the B-scan images, where the filtering is configured for isolating representations 4 3915-P1337WO.UW of dissolving gas bubbles from representations of surrounding tissue; and determining spatial and temporal attributes of the sub-mm scale gas bubbles. In one example, the filtering the ultrasound images comprises applying a dynamic mode decomposition (DMD) filter. In one example, the dynamic mode decomposition filter at least partially filters out motion of the surrounding tissue. In one example, the spatial and temporal attributes include an average time required for decay of the gas bubbles. In one example, the spatial and temporal attributes include a rate of decrease of sizes of the gas bubbles. In one example, the spatial and temporal attributes include rates of decay and angular frequency that corresponds to a Doppler phase shift of the ultrasound wave by the gas bubbles. In one example, the spatial and temporal attributes include at least one spatial distribution of the gas bubbles. In one example, the at least one spatial distribution of the gas bubbles includes a shape, and where brightness of all gas bubbles of the shape has a same behavior in time. In one example, the therapy transducer is a pulsed high-intensity focused ultrasound (pHIFU). In one example, the therapy transducer operates at a frequency of about 1 MHz. In one example, the therapy transducer operates based on a duty cycle of less than 5%. In one example, the therapy transducer generates bursts of the therapy ultrasound waves, and where adjacent bursts of the therapy ultrasound waves are separated by a quiet period of time during which the therapy transducer does not generate the therapy ultrasound waves. In one example, the gas bubbles are generated during the bursts of the therapy ultrasound waves. In one example, the gas bubbles decay during quiet periods of time during which the therapy transducer does not generate the therapy ultrasound waves. 5 3915-P1337WO.UW In one example, the imaging transducer operates at least partially during the quiet periods of time. In one example, the dissolving process of the gas bubbles includes collapsing the gas bubbles due to a pressure exerted by the surrounding tissue on the gas bubbles. In one example, growth of the gas bubbles damages the surrounding tissue. In one example, the method further includes reconstructing of a bubble mode,where the reconstructing is expressed as:( ) = ,index of selected modes that are interpreted as ithelement of initial mode amplitude vector that is related to an initial image . In one example, a system for imaging transient cavitation bubbles in a tissue of a patient using ultrasound imaging includes: a focused therapy transducer configured for generating a burst of therapy ultrasound waves that generate cavitation bubbles at a target region that is a focal region of the tissue, where the cavitation bubbles are sub-mm scale gas bubbles, and where, in absence of additionally generated therapy ultrasound waves, the gas bubbles undergo a dissolving process. The system also includes an imaging probe configured for generating imaging ultrasound pulses after the gas bubbles are generated by the therapy transducer at the target region and within a time period of the dissolving process. The system also includes a controller that is configured for: acquiring returning ultrasound signals as imaging planes, where the imaging planes at least partially correspond to a space occupied by the gas bubbles; composing ultrasound images as B-scan images based on the returning ultrasound signals; filtering the B-scan images, where the filtering is configured for isolating representations of dissolving gas bubbles from representations of surrounding tissue; and determining spatial and temporal attributes of the sub-mm scale gas bubbles. The complete disclosure of all patents, patent applications, and publications, and electronically available material cited herein are incorporated by reference in their entirety. Supplementary materials referenced in publications (such as supplementary tables, supplementary figures, supplementary materials and methods, and / or supplementary 6 3915-P1337WO.UW experimental data) are likewise incorporated by reference in their entirety. In the event that any inconsistency exists between the disclosure of the present application and the disclosure(s) of any document incorporated herein by reference, the disclosure of the present application shall govern. The foregoing detailed description and examples have been given for clarity of understanding only. No unnecessary limitations are to be understood therefrom. The invention is not limited to the exact details shown and described, for variations obvious to one skilled in the art will be included within the invention defined by the claims. The description of embodiments of the disclosure is not intended to be exhaustive or to limit the disclosure to the precise form disclosed. While the specific embodiments of, and examples for, the disclosure are described herein for illustrative purposes, various equivalent modifications are possible within the scope of the disclosure. DESCRIPTION OF THE DRAWINGS The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, where: FIGURE 1 is a schematic diagram of an ultrasound system for in accordance with an embodiment of the present technology; FIGURE 2 is a schematic diagram of an ultrasound system for in accordance with an embodiment of the present technology; FIGURE 3 is a graph of tissue displacement caused by ultrasound system in accordance with an embodiment of the present technology; FIGURE 4 is a graph of cavitation bubbles size as a function of time in accordance with an embodiment of the present technology; FIGURE 5 represents dynamic mode decomposition (DMD) filtering in accordance with an embodiment of the present technology; FIGURES 6A-6D are graphs representing input data for test ultrasound imaging in accordance with embodiments of the present technology; 7 3915-P1337WO.UW FIGURE 7 is a graph of bubble amplitude decay in accordance with an embodiment of the present technology; FIGURES 8A-8D are graphs representing processed test data of FIGURES 6A-6D based on DMD filtering in accordance with an embodiment of the present technology; FIGURES 9A-9D are graphs representing processed test data of FIGURES 6A-6D based on SVD filtering; FIGURES 10A and 10B are graphs representing processed test data of FIGURES 6A-6D based on IIR filtering; and FIGURES 11A-11F are graphs of ultrasound imaging test results using DMD filtering in accordance with an embodiment of the present technology. DETAILED DESCRIPTION While several embodiments have been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the claimed subject matter. Example devices, methods, and systems are described herein. It should be understood the words “example,” “exemplary,” and “illustrative” are used herein to mean “serving as an example, instance, or illustration.” Any embodiment or feature described herein as being an “example,” being “exemplary,” or being “illustrative” is not necessarily to be construed as preferred or advantageous over other embodiments or features. The example embodiments described herein are not meant to be limiting. It will be readily understood aspects of the present disclosure, as generally described herein, and illustrated in the figures, can be arranged, substituted, combined, separated, and designed in a wide variety of different configurations, all of which are explicitly contemplated herein. FIG.1 is a schematic diagram of an ultrasound system 1000 in accordance with an embodiment of the present technology. The ultrasound system 1000 includes an ultrasound probe 100 having a therapy transducer and an imaging transducer. The ultrasound probe 100 may be controlled by a controller (e.g., a computer) 600 having suitable software and commands for controlling the ultrasound. A monitor 500 can display images 62 of the target 8 3915-P1337WO.UW tissue of the patient that are obtained, for example, by an imaging transducer of the ultrasound probe 100 or by a separate imaging device. FIG. 2 is a schematic diagram of an ultrasound system for in accordance with an embodiment of the present technology. The illustrated pHIFU system is a test setup using ex vivo bovine tongue tissue 62, however, other types of test tissue may be used in different embodiments. In one embodiment, a 1.5 MHz HIFU therapy transducer 120 (8 cm aperture size, 6 cm focal distance and a 2 cm circular central opening) is controlled by high power driving electronics 122, which may be a custom built 12-channel class D amplifier system. A 64-element ATL P7-4 ultrasound imaging probe 140 is mounted at the center opening of the HIFU therapy transducer 120 and connected to Verasonics V1 control electronics 142 to acquire planewave Doppler imaging data (e.g., B-scan data). The driving electronics 122 and control electronics 142 may be collectively referred to as a controller for simplicity and brevity. In other embodiments, the driving electronics 122 and control electronics 142 may be parts of the controller 600. The transducer assembly 120 / 140 is mounted in a tank 300 filled with degassed, deionized water 320. The ex vivo tissue sample 62 (representing a tissue of a patient) is placed in front of the transducer in a holder attached to a 3D positioning system 200 (e.g. Velmex Inc., Bloomfield, NY, USA), such that the HIFU focus is at a depth of 1 cm. Dash lines 52 indicate focusing of the HIFU waves onto a focal area 54. In some embodiments, the pHIFU pulsing protocol includes pulse duration 1 ms, followed by pulse repetition frequency (PRF) 1 Hz. In some embodiments, pHIFU pulses are delivered at low duty cycle (e.g., <5%) and relatively low frequency (e.g., around 1 MHz) to avoid thermal effects and promote cavitation. During each HIFU pulse, the ultrasound (US) imaging probe may be operated in passive mode at a sampling rate of 20 MHz, to record broadband emissions from cavitation bubbles. In some embodiments, the received signals are then bandpass and comb-filtered to remove backscattered HIFU harmonics and are summed across the 64 channels with beamforming delays corresponding to the location of HIFU focus. The binary indicator of cavitation occurrence may be obtained from the resulting passive cavitation detection (PCD) signal, and it can be used to confirm the cavitation activity of the acquired dataset. 9 3915-P1337WO.UW In the illustrated embodiment, each HIFU pulse is immediately followed by a 5 MHz planewave Doppler ensemble consisting of 14 pulses at pulse repetition frequency (PRF) of 3 kHz. However, in different embodiments, other numbers of pulses and PRFs are also possible. Two different HIFU output levels are selected to achieve in situ focal pressure levels above and below the cavitation threshold. The in situ peak positive and negative pressures may be calculated using hydrophone measurements in water using attenuation coefficient of 0.044 Np / cm with 1 cm depth in tissue. The resulting peak positive and negative pressure are determined as 20.9 / -9.4 MPa, and 72.3 / -12.7 MPa, respectively. In some embodiments, a total of 30 and 60 HIFU treatment pulses are delivered to each focus location for non-cavitation and cavitation test cases, respectively. FIG.3 is a graph of tissue displacement caused by ultrasound system in accordance with an embodiment of the present technology. In some embodiments, the choice of the bovine tongue as the test tissue 62 is based on its anatomical structure, characterized by the interweaving of muscle bundles with fatty tissue, which results in a good cavitation probability. In some experiments, the tissues are trimmed, the fibrous surface are cut, degassed in saline for 1 hour in a desiccant chamber, and embedded in low melting point agarose gel (e.g., UltraPure Agarose; Invitrogen) to facilitate its placement in the holder. In operation, the test tissue 62 is characterized by a relatively slow motion (e.g., frequencies below 1 Hz, below 5 Hz, below 10 Hz, or similar) after being impacted by the pHIFU pulses. The displacement of tissue 62 is marked by arrow 56. The displacement graph includes two time periods: a rapid increase of the tissue motion when pHIFU pulse is applied (HIFU ON) up to a maximum amplitude of the tissue motion, followed by a gradual decrease in the amplitude of the tissue motion when HIFU does not transmit ultrasound (HIFU OFF). Gas bubbles at the target focal area of the tissue are generated during the HIFU ON period. Ultrasound imaging based on the gas bubbles is described with reference to FIG.4 below. FIG.4 is a graph of cavitation bubbles size as a function of time in accordance with an embodiment of the present technology. With the inventive technology, DMD processing includes capturing changes in the planewave Doppler imaging pulses backscattered from the HIFU focal area immediately after each HIFU pulse. Without being bound to theory, 10 3915-P1337WO.UW the changes in backscattering are expected to result from two phenomena: slow elastic tissue rebound motion induced by acoustic radiation force (ARF) associated with the HIFU pulse (as shown in FIG. 3); and rapid dissolution of the cavitation bubbles of varying size and at varying rate (as shown in FIG.4). In particular, immediately after the pHIFU pulse (HIFU ON), large gas bubbles (also referred to as cavitation bubbles) 64 are generated in the test tissue. After the pHIFU pulse is transmitted, the HIFU OFF period begins at about time t1. During the HIFU OFF period, the gas bubbles 64 rapidly decay and ultimately disappear. The gas bubbles 64 are characterized by their (decreasing) size and by shapes that groupings of the gas bubbles create up to the disappearance of the gas bubbles. In some embodiments, multiple bursts of the imaging ultrasound are transmitted toward the target during the HIFU OFF time at, for example, t1, t2, …, tfin time, resulting in the corresponding n imaging data sets. The process may repeat after a period of time with another pHIFU pulse (HIFU ON), closely followed by another set of imaging ultrasound transmittance and receiving during the following HIFU OFF period. As a result, the ultrasound images are obtained outside of the application of high amplitude pHIFU signal (HIFU ON), but still while the rapidly dissolving gas bubbles are in existence and capable of being imaged. An example processing of the ultrasound images is discussed with reference to FIG.5 below. In some embodiments, the therapy transducer may operate at a frequency of about 1 MHz. In some embodiments, the therapy transducer operates based on a duty cycle of less than 5% or less than 10%. In some embodiments, the therapy transducer generates bursts of therapy ultrasound waves, where the adjacent bursts of the therapy ultrasound waves are separated by a quiet period of time during which the therapy transducer does not generate the therapy ultrasound waves. As explained above, the gas bubbles are generated during the bursts of the therapy ultrasound waves. In some embodiments, the gas bubbles decay during quiet periods of time during which the therapy transducer does not generate the therapy ultrasound waves, while the imaging transducer operates at least partially during the quiet periods of time. In some embodiments, the dissolving process of the gas bubbles includes collapsing the gas bubbles due to a pressure exerted by the surrounding tissue on the gas bubbles. Correspondingly, the earlier growth of the gas bubbles damages the surrounding tissue. 11 3915-P1337WO.UW FIG. 5 represents dynamic mode decomposition (DMD) filtering in accordance with an embodiment of the present technology. In many embodiments, the inventive ultrasound imaging technology has to distinguish between the motion of the tissue itself (slow motion taking place in slow time) and the motion of the gas bubbles (fast motion taking place in fast time). In some embodiments, the captured changes in reconstructed images over slow time can be represented by a low-dimensional set of patterns, with each of these patterns associated with specific factors (e.g. cavitation or tissue motion) contributing to the observed changes. Additionally, linear approximation of the system dynamics may be used as the simplest case, described by coupled linear differential equations: =, (1)where is a matrix of backscattered pulses that reflects the dynamics of tissue and bubbles, and is a linear operator that revels the dynamics of the matrix . The linear approximation is expected to be adequate due to the limited number of temporal acquisition points (i.e. the number of pulses in the Doppler ensemble) within the bubbles’ lifespan. In some embodiments, we use the most basic implementation, also known as the exact DMD. Here, the data for pre-processing of DMD is constructed similarly to the general matrix formulation of SVD clutter filtering used in conventional Doppler imaging. The input data are +1 2D scan-converted in-phase and quadrature (I / Q) components acquired by Verasonics in slow time (labeled as , , , ), where +1 is the numberof pulses in Doppler ensemble. The total number of pixels in a single image ( , =0, 1, ,) is with grid size corresponding to the one wavelength of Doppler ensemble pulse (~0.3 mm). As illustrated in FIG. 5, each 2D I / Q dataset is reshaped into a column vector by attaching one column under the other in left-to-right order, and stacked into and matrices each having columns and one time step shifted from one another (i.e. contains to and contains to ). Then, the discrete time version of Eq. (1) could be written as 12 3915-P1337WO.UW= , (2)where = ( ) and is time step in slow time. Notably, the number ofrows ( ) of the matrices and , corresponding to the total number of pixels in aplanewave Doppler image, is on the order of 10 in this study. It is usually much largerthan the number of columns ( ), which is on the order of 10.In (2), the matrix fully describes the temporal evolution of the image dataset . However, the direct calculation of , as defined by =argmin = , (3) given that the dimensions of matrix are × . In Eq. (3), argmin denotes the process of determining the matrix at whichthe argument attains its minimum, is the Frobenius norm, and † indicates the pseudo- inverse. Instead of this direct computation, the exact DMD provides a reduced order set ofexact eigenvalues and eigenvectors of matrix . Specifically, a rank- approximation ofeigen-decomposition of can be expressed as (4) where is an × matrix, its columns of length are eigenvectors of , isan × diagonal matrix in which each diagonal element is an eigenvalue of , and thesuperscript denotes the inverse of the given matrix. The objective of DMD is to derive matrices and . Below is a summary of the corresponding procedure: a) Compute the SVD of the reconstructed dataset . =. (5)where the columns of and are right and left singular vectors with orthogonality that represent spatial modes and temporal coefficients, respectively, is a diagonal matrix 13 3915-P1337WO.UW that has at most non-zero singular values, and the superscript asterisk denotes the complex conjugate transpose. b) Truncate the SVD matrices to rank- by selecting the first singular values in adescending order and corresponding columns of and . ( ). Similar to SVD, thedetermination of the best-optimized rank is a crucial step in DMD. For example, if the rank is too low, there is a risk of intertwining tissue and bubble modes, and if it is too high, meaningless clutter modes could complicate the interpretation of the modes and identifying modes associated with bubbles. In this study, is determined when the summation of first singular values exceeded 97% of that of all singular values ( >0.97 ,where is a kthsingular value or diagonal term of matrix ). . (6) Here the subscript indicates rank- truncation of given matrix, such that thedimension of , , and are × , × , and × , respectively. Then, approximatecould be rewritten by substituting Eq. (6) into Eq. (3), as . (7) c) Compute the reduced order × matrix , which is a similarity transformationmatrix of . == . (8)d) Compute the eigen-decomposition of . =, (9)14 3915-P1337WO.UW where is the matrix whose columns are eigenvectors of and is the diagonal matrix, whose diagonal elements are the eigenvalues of . Since and are related by matrix similarity, they share the same eigenvalues , but different eigenvectors. e) Compute eigenvector matrix using Eqs, (4), (7), (8), and (9): == … . (10)of ). The diagonal elements of and column vectors of obtained from (9) and (10) represent the exact eigenvalue and corresponding eigenvector (DMD mode shapes) of matrix . The DMD mode shapes can be reshaped as 2D images, which are expected to correspond to physically interpretable modes such as planewave images of tissue and bubbles (Fig.2). The continuous-time eigenvalue matrix , defined as 0… 00 … 0 indicates how each DMD mode shape evolves in time. Here, is ithcontinuous- time eigenvalue (i.e. ithdiagonal element of ). The negative real part of corresponds to the temporal decay rate; the imaginary part corresponds to the angular frequency or rate of change. Mode Classification and Reconstruction Over the course of pHIFU treatment, the behavior of the transient bubbles at the focus is expected to change as the tissue becomes more disrupted. To identify and monitor these changes, the discrete-time eigenvalues (i.e. diagonal elements of ) are classified and tracked on the complex plane using k-means clustering method, which is a widely used method that groups dataset into k clusters by minimizing the distance between each data point and its cluster’s centroid. Since the discrete-time eigenvalues for exponentially 15 3915-P1337WO.UW decaying dynamics are bounded by a unit circle on the complex plane, they are moresuitable for clustering than continuous-time eigenvalues ( ). The classified eigenvaluesand corresponding DMD mode shapes are then characterized with respect to the three metrics: frequency, temporal decay rate, and maximum power level of the DMD modeshape defined as = (| | ). Based on the metrics, each DMD mode could beassociated with the dynamics of tissue or bubbles, and selective reconstruction of those modes could be performed. A bubble mode reconstruction, for example, can be expressed as () = , , (12) ,of selected modes that are interpreted ascomponents of bubble mode and is ithelement of initial mode amplitude vector relatedto the initial image (i.e. = ).Connection of DMD Modes to Conventional Doppler Variables Elimination of certain DMD modes can serve as a wall filter in conventional Doppler processing. For each DMD bubble mode, the consecutive bubble images in (12)(e.g. at = and =2 ) have a phase shift of { } relative to each otherover all the where { } denotes the imaginary part of . This means thatDMD processing directly provides the phase shift necessary for color Doppler estimation of velocity associated with each mode. Specifically, the angular frequency of a DMD mode(i.e. { })corresponds to Doppler phase shift, from which Doppler velocity is calculatedas: ={ }4 (13) where is the sound speed in tissue and indicates the frequency of planewave Doppler imaging pulses. This also implies that a positive DMD frequency corresponds to motion directed towards the HIFU transducer, whereas a negative frequency corresponds to motion away from it. 16 3915-P1337WO.UW Therefore, ultrasound images are composed as B-scan images based on the returning ultrasound signals, while filtering of the B-scan images operates to isolate representations of dissolving gas bubbles from representations of surrounding tissue. A person of ordinary skill would understand that in the context of this inventive technology, the B-scan images are ‘planewave images’ or ‘flash images.’ That is, referring back to FIG. 4, these B-scan images imply that each of imaging pulse (i.e., one of the t1– tfinimaging pulses) generates one image. Based on such filtering and discrimination among the representations of dissolving gas bubbles and those of the surrounding tissue, the inventive technology can determine spatial and temporal attributes of the sub-mm scale gas bubbles. FIGS. 6A-6D are graphs representing input data for test ultrasound imaging in accordance with embodiments of the present technology. FIGS.6A-6D represent numerical examples for testing DMD filter performance vs. SVD and IIR high pass filter. The input data consisting of four components of simulated 2D scan data: a square image with highamplitude (A), low frequency (f), and low decay rate ( ), simulating tissue backscatter (S1of FIG. 6A), left and right-side circles inside the square with low amplitude, high frequency, and high decay rate (S2of FIG. 6B and S3of FIG. 6C), simulating backscatter from bubbles, and a low amplitude random noise without any coherence in space and time (S4of FIG.6D). These ultrasound imaging tests are described below. FIG. 7 is an illustration of the amplitude decay for each component of input data across slow time. The amplitude was expressed as signal to noise ratio (SNR) in dB, as compared to the random noise level. That is, FIG.7 shows the SNR changes of each modeover slow time. The brightness of the two area of bubbles ( and ) rapidly decayed tobelow the random noise level within a few sampling points (3 sampling points (~ 1 ) and 6 sampling points (~2.2 ), respectively), which corresponded to decay time for residual cavitation bubbles. Three different methods are applied to the input data described above: DMD, SVD, and IIR high pass filter. In SVD, the data processing followed the same procedure as that described for DMD until step b). IIR high pass filter with the cut-off frequency of 275 Hz and order 3 is utilized to eliminate the image of background tissue (i.e., ). To assess the 17 3915-P1337WO.UW robustness of the DMD from random noise , the simulation is repeated n=1000 times with random noise regenerated every time. Mean and standard deviation of the frequency and temporal decay rate for each DMD mode, as measured by each of the processing methods, are obtained and compared to ground truth. Evaluation of DMD Performance in a Numerical Example The performance of DMD is evaluated in a numerically simulated time series of images, meant to represent a simplistic motion from tissue and bubbles with PRF and number of pulses matching the experimental arrangement. The data are processed with DMD filter and by commonly used wall filters in conventional Doppler processing – SVD and IIR high pass filter – for side-by-side comparison. As shown in FIGS.6A-6D, the data is a combination of four shapes , , , and : a square of uniform brightness, two circles with radially cosine distribution of brightness, and random noise, respectively. Each of them except random noise is spatiotemporally (also referred to as “spatially and temporally”) coherent – brightness of all points of the shape had the same behavior in time.The input brightness data [ ] is thus defined as[ ] = ( ) , (14) where is the initial brightness amplitude of each shape, is frequency ( = 50Hz, = 500 Hz, = 550 Hz, and = 0 Hz), is temporal decay rate ( = 0.1 , =2 , = 5 , and = 0 ), is time step ( = 0.33 , whichcorresponded to the experimentally allowable PRF of 3 kHz), and is sample number orthe virtual imaging pulse number (1 14). At each time step, random noise isrepeatedly generated to ensure that there is no spatial or temporal coherence. The initial brightness amplitudes, frequencies, and temporal decay rates of the modes are chosen to approximate the ultrasound images of slowly moving tissue which typically dominates the image amplitude, and that of a collection of bubbles which are quickly changing anddissolving. The initial brightness amplitudes of each shape ( , , , and ) are 2, 1, 1,and 0.01, respectively. For tissue containing two bubble areas ( and ), the amplitudesare designed to have signal-to-noise ratio (SNR) of 40 dB vs random noise level ( ), and18 3915-P1337WO.UWthat of tissue ( ) is set to be twice larger than or . Figure 3(b) shows the SNR changesof each mode over slow time. The brightness of the two area of bubbles ( and ) rapidlydecayed to below the random noise level within a few sampling points (3 sampling points (~ 1 ) and 6 sampling points (~2.2 ), respectively), which corresponded to a previously reported decay time for residual cavitation bubbles

[0012] . Three different methods were applied to the input data described above: DMD, SVD, and IIR high pass filter. In SVD, the data processing followed the same procedure as that described for DMD until step b). IIR high pass filter with the cut-off frequency of 275 Hz and order 3 was utilized to eliminate the image of background tissue (i.e. ). To assess the robustness of the DMD from random noise , the simulation was repeated n=1000 times with random noise regenerated every time. Mean and standard deviation of the frequency and temporal decay rate for each DMD mode, as measured by each of the processing methods, were obtained and compared to ground truth. The results of this numerical simulation are discussed below. FIGS.8A-8D are graphs representing processed test data of FIGS.6A-6D based on DMD filtering in accordance with an embodiment of the present technology. In particular,four DMD modes along with their corresponding frequency ( ) and temporal decay rate( ) are presented. These images represent B-scan data in the practical ultrasound imaging.Stated differently, these images serve as in silico representation or in silico model of B- mode data. FIGS.9A-9D are graphs representing processed test data of FIGS.6A-6D based on SVD filtering. In particular, the first four columns of SVD matrix with the correspondingsingular values ( ) are presented.FIGS. 10A and 10B are graphs representing processed test data of FIGS. 6A-6D based on IIR filtering. In particular, the filter is an IIR high pass filter. For DMD (FIGS.8A-8D), each mode image, and corresponding values of and were very consistent across repeated simulation with regenerated random noise. Table 1 below shows the mean and standard deviation of and estimated by DMD. 19 3915-P1337WO.UW

[0002] As seen, DMD filtering acts to effectively decompose spatially coherent patternsand provids and estimates with high accuracy, within 0.6% error vs ground truth. Themodes 1, 2, and 3 closely corresponded to the shapes of input data without overlapping with each other. However, mode 3 exhibited more contamination by the random noise compared to modes 1 or 2. This was expected, as the amplitude of mode 3 brightness exceeded that of random noise for only 3 out of 14 sampling points. In some embodiments, the spatial and temporal attributes of the bubbles include an average time required for decay of the gas bubbles. In other embodiments, the spatial and temporal attributes of the gas bubbles include a rate of decrease of sizes of the gas bubbles. In some embodiments, the spatial and temporal attributes comprise rates of decay and angular frequency that corresponds to a Doppler phase shift of the ultrasound wave by the gas bubbles. In some embodiments, the spatial and temporal attributes include at least one spatial distribution of the gas bubbles. In some embodiments, the at least one spatial distribution of the gas bubbles comprises a shape, while the brightness of all points of the shape has a same behavior in time. In contrast with the above DMD results, SVD was not able to separate modes 2 and3 (“bubble modes”), and, as seen in FIGS.9A-9D, these modes overlap. The singular valueof mode 1 (“tissue mode”) was an order of magnitude higher than those of modes 2 and 3, which is consistent with the input dataset where mode 1 had the highest initial amplitude and the lowest . 20 3915-P1337WO.UW FIGS. 10A and 10B show the result of applying IIR high pass filter to the data, meant to eliminate mode 1, leaving only modes 2 and 3, and the result of subtraction of the filtered data from the input data, meant to reconstruct mode 1. Because the appropriate cut- off frequency to filter out mode 1 was known a priori, the separation of the “bubble modes” from “tissue mode” was successful, and the images of the modes had minimal overlap. FIGS. 11A-11F are graphs of ultrasound imaging test results using DMD filtering in accordance with an embodiment of the present technology. In particular, FIG. 11A represents a raw B-scan image. FIGS. 11B-11F depict representative example of DMD modes corresponding to a HIFU pulse that induced cavitation in the ex vivo tissue. The illustrated DMD modes of Bubble Doppler obtained from pHIFU exposure of ex vivo tissue with cavitation. The box on the B-mode image of FIG.11A highlights the region of interest for bubble Doppler. Five DMD modes were identified, and the corresponding images are presented with frequencies and decay rates. The intensity bar scale was normalized to the maximum amplitude value over all DMD modes. HIFU was delivered from the top of the images to the target point indicated by V-shaped dashed line of FIGS.11B-11F. DMD processing consistently yielded five modes for those pHIFU exposures that represented the dynamics of tissue and residual bubbles immediately following each pulse. The first DMD mode displayed speckles over the entire ROI, with brighter hyperechoic regions at the focus. This mode had the lowest frequency (0.006 kHz) and temporal decay rate (0.01 ), indicating stationary backscatter from tissue and persistent residual bubbles that did not change within the imaging time window. The second mode was similar to the first mode in speckle distribution but had a higher temporal decay rate. The third and fourth modes depicted the speckles around the focus, with higher frequencies and temporal decay rates than the first two modes and lower amplitude level (<40% of the second). The Doppler speeds corresponding to those two modes were 2.4 and 3.7 cm / s with opposite directions, likely linked with ARF-displaced tissue motion similarly to second and third modes in Fig 5. The fifth mode showed a backscatter pattern highly localized to the focal region, and the corresponding frequency, temporal decay rate, and Doppler speed were the highest among all modes (0.785 kHz, 3.23 , and 12.1 cm / s). This mode was thus hypothesized to be associated with quickly dissolving cavitation bubbles. Indeed, previous 21 3915-P1337WO.UW studies have shown that most cavitation bubbles in pHIFU exposures dissolved within 1-2 following the pulse, which is consistent with of this mode. The amplitude ratio between the first and fifth mode was initially -2 dB but reached -30 dB after 1 . Further, many cavitation bubbles with different radius and dissolution rate are expected to be present within the resolution cell of planewave image. Bubbles dissolving at different rates may be interpreted by DMD as rapidly changing backscattered signal. In conclusion, the effectiveness of DMD to identify spatially coherent patterns and quantify their temporal dynamics was demonstrated in a numerically simulated image dataset and compared to the performance of SVD and IIR wall filtering. Then DMD was applied to planewave Doppler datasets acquired during pHIFU exposures of ex vivo tissue. In the simulation study DMD outperformed the SVD and IIR filter in terms of quality of pattern decomposition and accuracy of the temporal dynamics metrics associatedwith them–frequency ( ) and temporal decay rate ( ). The values of and obtainedthrough DMD were remarkably accurate for all modes, despite the simulation’s input data containing two limitations that are inherent in the experimental implementation. First, the total duration of Doppler ensemble in slow time (i.e.4.7 for 14 sampling points with a 3 kHz PRF) was insufficient to encompass even a single cycle of 50 Hz, which was designated as the frequency for the square shape. Second, the SNR for the two circles were intentionally reduced to levels below random noise after 6 and 3 sampling points, respectively, as shown in Fig.3(b). In pHIFU exposures of ex vivo tissue, DMD modes and their (or corresponding Doppler speed) and were found to be physically interpretable and consistent with prior studies. In pHIFU exposures without cavitation (Fig. 5), the three DMD modes could be identified as stationary tissue (first mode) and ARF-displaced axial tissue motion in both directions (second and third modes). This is because the tissue, immediately after the termination of HIFU pulse, is expected to move towards the HIFU transducer, and then to rebound away within the total acquisition duration (i.e. 4.7 ), as seen prior study

[0030] . Similarly, each DMD mode in pHIFU exposures resulting in cavitation (Fig 7) can be identified as stationary tissue (first and second modes), ARF-induced axial tissue motion in both directions (third and fourth modes), and transient bubbles (fifth mode). This may 22 3915-P1337WO.UW be attributed to the fact that some larger size, more stable bubbles moved together with tissue and potentially amplified the ARF displacement. In the bubble mode, the temporal decay rate can be interpreted as an average rate of bubble dissolution within a resolution cell of a planewave image. However, the physical interpretation of frequency , or the corresponding Doppler speed, is not as straightforward, considering that residual bubbles are not expected to move or oscillate after the HIFU pulse. A possible explanation is that the speed of the surface of shrinking bubbles within the resolution cell is synchronized, leading to a phase shift in slow time. The DMD then interprets this as motion. The traceability of tissue and bubble modes is another advantage of DMD. For instance, the k-means clustering on complex planes allows one to track each DMD mode. This ability is particularly useful for quantitative imaging guidance in pHIFU therapy, as it allows for tracking of bubble characteristics, which could be physically interpretable in terms of inflicted tissue damage. For example, the decrease in frequency and temporal decay rate for cavitation bubble mode could be interpreted as a sign of bubble coalescence and decrease in surrounding tissue stiffness. The utility of DMD was numerically tested for a relatively narrow application – as a type of wall filter in Doppler-based detection, mapping and quantitation of cavitation bubbles induced by pHIFU. However, we anticipate this technique to be useful in a number of other time-resolved ultrasound imaging tasks, e.g. vascular imaging, characterization of flow patterns in large veins and arteries, and monitoring of tissue ablative interventions over time. In some embodiments, the clustering method meant to track the different modes over the course of pHIFU exposure may not necessarily include automatic identification of the cluster corresponding to the cavitation bubbles. To automate the labeling process, several algorithms could be utilized, including the eigenvalue thresholding (using and ), mode image-based metrics (e.g. contrast-to-noise ratio), or machine learning techniques. Second, the appropriate selection for rank- truncation using SVD, described in step b) of DMD processing section, also affects the decomposition quality of DMD. As described above, a preliminary empirical investigation was performed to identify the most 23 3915-P1337WO.UW effective truncation method for fast-dissolving bubbles. In different embodiments, such an investigation may need to be repeated if there are changes in acquisition parameters, suchas the number of imaging pulses ( ) used and Doppler PRF. Even though the abovenumerical studies are based on pHIFU exposures of stationary ex vivo tissue, other application in the in vivo environment with more complicated analysis and increased number of modes are also possible. Furthermore, the above-described advantages of the DMD filtering are expected to stand out as even more pronounced. Specific elements of any foregoing embodiments can be combined or substituted for elements in other embodiments. Moreover, the inclusion of specific elements in at least some of these embodiments may be optional, wherein further embodiments may include one or more embodiments that specifically exclude one or more of these specific elements. Furthermore, while advantages associated with certain embodiments of the disclosure have been described in the context of these embodiments, other embodiments may also exhibit such advantages, and not all embodiments need necessarily exhibit such advantages to fall within the scope of the disclosure. As used herein and unless otherwise indicated, the terms “a” and “an” are taken to mean “one”, “at least one” or “one or more”. Unless otherwise required by context, singular terms used herein shall include pluralities and plural terms shall include the singular. Unless the context clearly requires otherwise, throughout the description and the claims, the words ‘comprise’, ‘comprising’, and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in the sense of “including, but not limited to”. Words using the singular or plural number also include the plural and singular number, respectively. Additionally, the words “herein,” “above,” and “below” and words of similar import, when used in this application, shall refer to this application as a whole and not to any particular portions of the application. Unless otherwise indicated, all numbers expressing quantities of components, molecular weights, and so forth used in the specification and claims are to be understood as being modified in all instances by the term “about.” Accordingly, unless otherwise indicated to the contrary, the numerical parameters set forth in the specification and claims are approximations that may vary depending upon the desired properties sought to be 24 3915-P1337WO.UW obtained by the present invention. At the very least, and not as an attempt to limit the doctrine of equivalents to the scope of the claims, each numerical parameter should at least be construed in light of the number of reported significant digits and by applying ordinary rounding techniques. In the context of this disclosure, the term “about,” approximately” and similar means + / - 5% of the stated value. Notwithstanding that the numerical ranges and parameters setting forth the broad scope of the invention are approximations, the numerical values set forth in the specific examples are reported as precisely as possible. All numerical values, however, inherently contain a range necessarily resulting from the standard deviation found in their respective testing measurements. All headings are for the convenience of the reader and should not be used to limit the meaning of the text that follows the heading, unless so specified. All of the references cited herein are incorporated by reference. Aspects of the disclosure can be modified, if necessary, to employ the systems, functions, and concepts of the above references and application to provide yet further embodiments of the disclosure. These and other changes can be made to the disclosure in light of the detailed description. It will be appreciated that, although specific embodiments of the invention have been described herein for purposes of illustration, various modifications may be made without deviating from the spirit and scope of the invention. Accordingly, the invention is not limited except as by the claims. 25 3915-P1337WO.UW

Claims

CLAIMS What is claimed is:

1. A method for imaging transient cavitation bubbles in a tissue of a patient using ultrasound imaging, the method comprising: generating a burst of therapy ultrasound waves by a focused therapy transducer; generating cavitation bubbles by the burst of therapy ultrasound waves at a target region that is a focal region of the tissue, wherein the cavitation bubbles are sub-mm scale gas bubbles, wherein, in absence of additionally generated therapy ultrasound waves, the gas bubbles undergo a dissolving process; after generating a plurality of the gas bubbles at the target region and within a time period of the dissolving process, generating imaging ultrasound pulses by an imaging probe; acquiring returning ultrasound signals as imaging planes, wherein the imaging planes at least partially correspond to a space occupied by the gas bubbles; composing ultrasound images as B-scan images based on the returning ultrasound signals; filtering the B-scan images, wherein the filtering is configured for isolating representations of dissolving gas bubbles from representations of surrounding tissue; and determining spatial and temporal attributes of the sub-mm scale gas bubbles.

2. The method of claim 1, wherein the filtering the ultrasound images comprises applying a dynamic mode decomposition (DMD) filter.

3. The method of claim 2, wherein the dynamic mode decomposition filter at least partially filters out motion of the surrounding tissue. 26 3915-P1337WO.UW4. The method of claim 2, wherein the spatial and temporal attributes comprise an average time required for decay of the gas bubbles.

5. The method of claim 2, wherein the spatial and temporal attributes comprise a rate of decrease of sizes of the gas bubbles.

6. The method of claim 2, wherein the spatial and temporal attributes comprise rates of decay and angular frequency that corresponds to a Doppler phase shift of the ultrasound wave by the gas bubbles.

7. The method of claim 2, wherein the spatial and temporal attributes comprise at least one spatial distribution of the gas bubbles.

8. The method of claim 7, wherein the at least one spatial distribution of the gas bubbles comprises a shape, and wherein brightness of all gas bubbles of the shape has a same behavior in time.

9. The method of claim 1, wherein the therapy transducer is a pulsed high- intensity focused ultrasound (pHIFU).

10. The method of claim 9, wherein the therapy transducer operates at a frequency of about 1 MHz.

11. The method of claim 9, wherein the therapy transducer operates based on a duty cycle of less than 5%.

12. The method of claim 9, wherein the therapy transducer generates bursts of the therapy ultrasound waves, and wherein adjacent bursts of the therapy ultrasound waves are separated by a quiet period of time during which the therapy transducer does not generate the therapy ultrasound waves. 27 3915-P1337WO.UW13. The method of claim 12, wherein the gas bubbles are generated during the bursts of the therapy ultrasound waves.

14. The method of claim 13, wherein the gas bubbles decay during quiet periods of time during which the therapy transducer does not generate the therapy ultrasound waves.

15. The method of claim 14, wherein the imaging transducer operates at least partially during the quiet periods of time.

16. The method of claim 1, wherein the dissolving process of the gas bubbles comprises collapsing the gas bubbles due to a pressure exerted by the surrounding tissue on the gas bubbles.

17. The method of claim 16, wherein growth of the gas bubbles damages the surrounding tissue.

18. The method of claim 2, further comprising reconstructing of a bubble mode, wherein the reconstructing is expressed as: () = ,where , indicates an index of selected modes that are interpreted ascomponents of bubble mode, and is ithelement of initial mode amplitude vector that is related to an initial image .

19. A system for imaging transient cavitation bubbles in a tissue of a patient using ultrasound imaging, the system comprising: a focused therapy transducer configured for generating a burst of therapy ultrasound waves that generate cavitation bubbles at a target region that is a focal region of the tissue, 28 3915-P1337WO.UWwherein the cavitation bubbles are sub-mm scale gas bubbles, and wherein, in absence of additionally generated therapy ultrasound waves, the gas bubbles undergo a dissolving process; an imaging probe configured for generating imaging ultrasound pulses after the gas bubbles are generated by the therapy transducer at the target region and within a time period of the dissolving process; and a controller configured for: acquiring returning ultrasound signals as imaging planes, wherein the imaging planes at least partially correspond to a space occupied by the gas bubbles; composing ultrasound images as B-scan images based on the returning ultrasound signals; filtering the B-scan images, wherein the filtering is configured for isolating representations of dissolving gas bubbles from representations of surrounding tissue; and determining spatial and temporal attributes of the sub-mm scale gas bubbles.

20. The system of claim 19, wherein the filtering the ultrasound images comprises applying a dynamic mode decomposition (DMD) filter.

21. The system of claim 20, wherein the dynamic mode decomposition filter at least partially filters out motion of the surrounding tissue.

22. The system of claim 20, wherein the spatial and temporal attributes comprise an average time required for decay of the gas bubbles or a rate of decrease of sizes of the gas bubbles.

23. The system of claim 20, wherein the spatial and temporal attributes comprise rates of decay and angular frequency that corresponds to a Doppler phase shift of the ultrasound wave by the gas bubbles. 29 3915-P1337WO.UW24. The system of claim 20, wherein the spatial and temporal attributes comprise at least one spatial distribution of the gas bubbles.

25. The system of claim 24, wherein the at least one spatial distribution of the gas bubbles comprises a shape, and wherein brightness of all gas bubbles of the shape has a same behavior in time.

26. The system of claim 19, wherein the therapy transducer is a pulsed high- intensity focused ultrasound (pHIFU).

27. The system of claim 26, wherein the therapy transducer operates at a frequency of about 1 MHz.

28. The system of claim 26, wherein the therapy transducer operates based on a duty cycle of less than 5%.

29. The system of claim 26, wherein the therapy transducer generates bursts of the therapy ultrasound waves, and wherein adjacent bursts of the therapy ultrasound waves are separated by a quiet period of time during which the therapy transducer does not generate the therapy ultrasound waves.

30. The system of claim 29, wherein the gas bubbles are generated during the bursts of the therapy ultrasound waves.

31. The system of claim 30, wherein the gas bubbles decay during quiet periods of time during which the therapy transducer does not generate the therapy ultrasound waves.

32. The system of claim 31, wherein the imaging transducer operates at least partially during the quiet periods of time. 30 3915-P1337WO.UW33. The system of claim 19, wherein the dissolving process of the gas bubbles comprises collapsing the gas bubbles due to a pressure exerted by the surrounding tissue on the gas bubbles.

34. The system of claim 33, wherein growth of the gas bubbles damages the surrounding tissue.

35. The system of claim 20, further comprising reconstructing of a bubble mode, wherein the reconstructing is expressed as: () = ,index of selected modes that are interpreted ascomponents of bubble mode, and is ithelement of initial mode amplitude vector that is related to an initial image . 31 3915-P1337WO.UW

Images