Acoustic Mapping Research Articles

Distributed extended Kalman filtering for acoustic simultaneous localization and environment mapping

The localization accuracy of Acoustic Simultaneous Localization and Mapping (ASLAM) often suffers from environmental noise and room reverberation. To solve this problem, an ASLAM method based on distributed extended Kalman filter is proposed. Specifically, the distance between the sound source and the robot is effectively estimated by the gating mechanism method in the case of unknown location of the sound source. Then, the Time Difference of Arrival (TDOA) between robots, the Direction of Arrival (DOA) and the distance between the sound source and the robot are presented as observation. Next, the robots' trajectories and the sound source's position are tracked by multiple extended Kalman filters. Finally, the ASLAM localization is completed by fusing the sound source and the robots' positioning information through average consensus algorithm, respectively. The proposed method can efficiently promote the ASLAM localization accuracy in complex acoustic environments. The experimental results verify the effectiveness of the proposed method.

Kaiwi shoreline basalts fed by the west rift zone of East Molokaʻi

Bathymetry and acoustic imagery swath mapping, along with observations and samples from four manned submersible and four ROV dives, confirm that a seafloor slope break on the northern approaches to Kaiwi Channel, between the islands of Oʻahu and Molokaʻi, Hawaiʻi is a former shoreline, now submerged ∼800 m below present sea level. Subaerially emplaced, low-relief basaltic lavas above the slope break transition to submarine morphologies below. The entire region has been tilted about 1° to the SSE (150°), and is cut by an 8–15 m-high, north-facing scarp, 100–400 m south of the slope break. The distribution of platy, table-top, and rarer mounded branching corals indicates the former presence of fringing reefs around low-relief paleo-islands. We infer that the regional tilt resulted from loading by younger Hawaiian volcanoes, compounded by flexural uplift and back tilting away from the unloaded footwall of a flank landslide to the north.Basalt samples collected from both above and below the slope break have petrography, chemical composition, and age (1.64–1.80 Ma) indicating correlation with the (late-shield) Lower Member of the East Molokaʻi Volcanics, rather than with the more proximal volcano of West Molokaʻi. The most likely source of the Kaiwi basalts is a submarine ridge (rift zone) that extends northwest away from ʻĪlio Point on West Molokaʻi. Although the submarine ridge was previously assumed to be an extension of West Molokaʻi's northwest rift, we conclude that regional bathymetry and gravity are consistent with this feature being an extension of the west rift of East Molokaʻi. A corallary of this interpretation is that the shoreline slope break (SSB 7 of Taylor, 2019) in this area is distinct from and younger than the southern SSB 7 formed on West Molokaʻi volcano (∼1.65 Ma vs. ∼1.8 Ma).

Integrated methodology for gas content assessment and prediction in shallow muddy lake sediments: acoustic mapping and correlation analysis

This paper provides a step-by-step description of integrated methodology for quantification and prediction of gas (methane, CH4) content dynamics in shallow aquatic sediments under changing spatial and temporal conditions. Presence of gas bubbles even in small concentrations significantly affects sediment compressibility, which in turn decreases sound speed in sediment. Our integrated methodology consists of two basic steps. In the first step, free gas content is evaluated by acoustic applications based on the sound speed inferred from the reflection coefficient from gassy bottom. The experimental bottom reflections are registered and compared to the simulated ones, using a geoacoustic inversion technique. The best match between the model and the experiment provides sediment sound speed estimate, which is converted into free gas content using a basic relation. In the second step, a multivariate linear regression is fitted for gas content and closed form expression of gas content dependence on the following predictors, which change spatially and temporally over the aquatic ecosystem, is obtained: 1) water depth, 2) short-leaving CH4 production rate peaks fueled by punctuated organic matter deposition; and 3) CH4 bubble dissolution rates.•Gas content and sound speed in the sediment are estimated via the geoacoustic inversion technique by matching the experimentally recorded and simulated bottom reflections•Only single source and receiver are required for the acoustic methodology•A multivariate linear regression is fitted for gas content to indicate its dependence on various predictors that change spatially and temporally over the lake

Uncertainty analysis of Altantic salmon fish scale’s acoustic impedance using 30 MHz C-Scan measurements

Understanding the biomechanics of fish scales is crucial for their survival and adaptation. Ultrasonic C-scan measurements offer a promising tool for non-invasive characterization, however, existing literature lacks uncertainty analysis while evaluating acoustic impedance. This article presents an innovative integration of uncertainty into the analytical framework for estimating stochastic specific acoustic impedance of salmon fish scale through ultrasonic C-scans. In this study, the various types of uncertainties arising due to variation in biological structures and aging, measurement errors, and analytical noises are combined together in the form of uncertain reflectance. This uncertain reflectance possesses a distribution which is derived using a theory of waves by assuming suitable stochasticity in wavenumber. This distribution helps in development of a stochastic-specific acoustic impedance map of the scales which demonstrates the possible deviations of impedance from mean value depending on uncertainties. Furthermore, maximal overlap discrete wavelet transform is employed for efficient time–frequency deconvolution and Kriging for spatial data interpolation to enhance the robustness of the impedance map, especially in scenarios with limited data. The framework is validated by accurately estimating the specific acoustic impedance of well-known materials like a pair of target medium (polyvinylidene fluoride) and reference medium (polyimide), achieving over 90% accuracy. Moreover, the accuracy of the framework is found superior when compared with an established approach in the literature. Applying the framework to salmon fish scales, we obtain an average specific acoustic impedance of 3.1 MRayl along with a stochastic map visualizing the potential variations arising from uncertainties. Overall, this work paves the way for more accurate and robust studies in fish scale biomechanics by incorporating a comprehensive uncertainty analysis framework.

Characterizing prey fields in humpback whale foraging areas of southern British Columbia

Humpback whales ( Megaptera novaeangliae) use southern British Columbia waters to feed, but the type and quantity of prey in many areas used for feeding is unknown. We conducted active acoustic prey mapping in 55 small grid-surveys in two regions off Vancouver Island. We quantitatively compared fish and zooplankton-dominated biomass in known feeding areas with and without foraging humpback whales, and qualitatively described the prey characteristics of the foraged areas. Surveys of the water column suggest that, on average, humpback whale foraging was associated more with increased zooplankton than fish biomass. Prey characteristics varied between the two regions (∼500 km apart), but there was no significant difference in mean backscatter strength in the actively foraged areas between the two regions. Frequency differencing discriminated between the dominant taxa in the water column, but potential epipelagic prey (<10 m) would have been omitted from analysis. However, average depth at the maximum acoustic prey detections was significantly deeper when whales were present (84 m) versus absent (60 m), suggesting predominantly subsurface foraging opportunities suitable to prey mapping.

3D Acoustic Map Analysis of the National Theatre of Zagreb

Rapid technological advances in recent decades have led researchers to refine the accuracy of their studies. In the field of acoustics, the impact of new devices is noticeable, especially in the investigations of cultural heritage buildings. The selection of a seat in theatres and concert halls has always been a concern, since the live experience of artistic performance depends on the quality of hearing and sight view. This paper deals with the elaboration of 360° acoustic maps made in the National Theatre of Zagreb, one of the opera theatres investigated with the Sipario project. The analysis of the main acoustic parameters has been carried out, starting with site measurements describing the acoustic response at various representative points of the main hall by covering the audience area. In addition, acoustic maps have been created for some selected positions based on a 3-degree-of-freedom (3dof) technique that allows a panoramic visualization of the impulse responses (IRs). This methodology completes the determination of early and late reflections that contribute to the acoustic quality of a place. In addition to the interest of experts in acoustics, this methodology can also be adopted by music lovers who can find a reasonable explanation for seat selection when booking their tickets.

Multivariate mapping of seabed grain size parameters in the Bay of Fundy using convolutional neural networks

High-resolution seabed sediment information is critical for a range of marine spatial planning applications in multi-use shelf environments. To establish this information for the Bay of Fundy, Canada, legacy seabed sediment measurements were obtained from regional data compilations, and eight parameters describing the grain size were modelled across the extent of the bay using high resolution acoustic seafloor mapping and oceanographic datasets. This was achieved using a purpose-made convolutional neural network configured for geospatial modelling of multivariate grain size parameters. Shared information between the response parameters enabled model training with partially complete observations from the varied legacy data sources, and an explicit multiscale model architecture ensured that environmental predictors were implemented at appropriate scales for modelling each parameter. This avoids typical exhaustive exploration and selection of scale-specific predictor sets that often precede model building. Compositional grain size parameters were additionally accommodated using appropriate output activation functions, providing an efficient alternative to compositional data transformation and imputation. Results agreed well with our current understanding of the surficial geology of the bay, and cross-validation was used to quantitatively evaluate map predictions. Of the eight predicted parameters, the mean grain size and mud (clay and silt) fractions were predicted with high accuracy (> 50% variance explained); the accuracy of grain size skewness was comparatively low (24% variance explained). Exploration of variable importance suggested that compiled acoustic backscatter was the most important environmental variable for predicting the grain size, but that geographic information describing the latitude and longitude within the bay was also highly useful. We hypothesize an interaction between these variables that enables location-specific prediction. Data layers of predicted grain size parameter values are made available for further sedimentological and ecological exploration, and for marine spatial planning activities within the bay.

Improved skull bone acoustic property homogenization for fast transcranial ultrasound simulations

Transcranial ultrasound simulations are crucial to optimize and secure ultrasound interventions in brain therapy, depending on the patient skull. When performing such simulations, accurate modeling of the skull is essential, although very challenging, because of the inter/intra sample property variability. Simulations based on semi-analytical methods require a homogeneous description of the skull. Averaging the acoustic property maps derived from the CT scan does not modify the focus shift, but it leads to an overestimation of the pressure field amplitude. The purpose of this work is to provide a homogenization method that compensates for this amplitude overestimation. First, the skull acoustic property maps are segmented into a three-layer medium to represent the different types of skull bone (cortical – trabecular – cortical). Then, equivalent properties are computed so as to minimize the time of flight and transmission coefficient errors between the three-layer medium and the one-layer equivalent medium. This method was validated using 3D simulations with CIVA Healthcare and k-Wave and has proven to be very efficient.

Petra Anamnetic: acoustic territories in southern Jordan

ABSTRACT This paper discusses the Petra Anamnetic acoustic mapping project. Juxtaposing audio recordings made in the World Heritage Site of Petra and the neighbouring tourist town of Wadi Musa, Jordan, the piece interrogates how sound might provide insights into the complex frictions existing between the native Bedouins and the foreign travelers they have become dependent upon for their economic survival.

Breast Tomographic Ultrasound: The Spectrum from Current Dense Breast Cancer Screenings to Future Theranostic Treatments.

This review provides unique insights to the scientific scope and clinical visions of the inventors and pioneers of the SoftVue breast tomographic ultrasound (BTUS). Their >20-year collaboration produced extensive basic research and technology developments, culminating in SoftVue, which recently received the Food and Drug Administration's approval as an adjunct to breast cancer screening in women with dense breasts. SoftVue's multi-center trial confirmed the diagnostic goals of the tissue characterization and localization of quantitative acoustic tissue differences in 2D and 3D coronal image sequences. SoftVue mass characterizations are also reviewed within the standard cancer risk categories of the Breast Imaging Reporting and Data System. As a quantitative diagnostic modality, SoftVue can also function as a cost-effective platform for artificial intelligence-assisted breast cancer identification. Finally, SoftVue's quantitative acoustic maps facilitate noninvasive temperature monitoring and a unique form of time-reversed, focused US in a single theranostic device that actually focuses acoustic energy better within the highly scattering breast tissues, allowing for localized hyperthermia, drug delivery, and/or ablation. Women also prefer the comfort of SoftVue over mammograms and will continue to seek out less-invasive breast care, from diagnosis to treatment.

Is it possible to develop a digital twin for noise monitoring in manufacturing?

Noise monitoring is important in the context of manufacturing because it can help maintain a safe and healthy workspace for employees. Current approaches for noise monitoring in manufacturing are based on acoustic sensors, whose measured sound pressure levels (SPL) are shown as bar/curve charts and acoustic heat maps. In such a way, the noise emission and propagation process is not fully addressed. This paper proposes a digital twin (DT) for noise monitoring in manufacturing using augmented reality (AR) and the phonon tracing method (PTM). In the proposed PTM/AR-based DT, the noise is represented by 3D particles (called phonons) emitting and traversing in a spatial domain. Using a mobile AR device (HoloLens 2), users are able to visualize and interact with the noise emitted by machine tools. To validate the feasibility of the proposed PTM/AR-based DT, two use cases are carried out. The first use case is an offline test, where the noise data from a machine tool are first acquired and used for the implementation of PTM/AR-based DT with different parameter sets. The result of the first use case is the understanding between the AR performance of HoloLens 2 (frame rate) and the setting of the initial number of phonons and sampling frequency. The second use case is an online test to demonstrate the in-situ noise monitoring capability of the proposed PTM/AR-based DT. The result shows that our PTM/AR-based DT is a powerful tool for visualizing and assessing the real-time noise in manufacturing systems.

Optimization of 3D Passive Acoustic Mapping Image Metrics: Impact of Sensor Geometry and Beamforming Approach.

Three-dimensional passive acoustic mapping (PAM) with matrix arrays typically suffers from high demands on the receiving electronics and high computational load. In our study, we investigated, both numerically and experimentally, the influence of matrix array aperture size, element count, and beamforming approaches on defined image metrics. With a numerical Vokurka model, matrix array acquisitions of cavitation signals were simulated. In the experimental part, two 32 × 32 matrix arrays with different pitches and aperture sizes were used. After being reconstructed into 3D cavitation maps, defined metrics were calculated for a quantitative comparison of experimental and numerical data. The numerical results showed that the enlargement of the aperture from 5 to 40 mm resulted in an improvement of the full width at half maximum (FWHM) by factors of 6 and 13 (in lateral and axial dimension, respectively). A larger array sparsity influenced the point spread function (PSF) only slightly, while the grating lobe level (GLL) remained more than 12 dB below the main lobe. These results were successfully experimentally confirmed. To further reduce the GLL caused by array sparsity, we adapted a non-linear filter from optoacoustics for use in PAM. In combination with the delay, multiply, sum, and integrate (DMSAI) algorithm, the GLL was decreased by 20 dB for 64-element reconstructions, resulting in levels that were identical to the fully populated matrix reconstruction levels.

Subharmonics are not the full story—Searching for acoustic signatures of cavitation and other fantastic beasts

Cavitation induced bio-effects are being exploiting in a wide range of applications from physiotherapy to brain surgery. The acoustic emissions generated by bubble activity are extremely useful in enabling real time treatment monitoring. The relationship between the spectral characteristics of these emissions and bubble dynamics is, however, complex. Here we report data from experiments with simultaneous ultra-high-speed optical imaging and passive acoustic mapping of individual microbubbles exposed to 50 cycles of ultrasound at 0.5 MHz with varying peak negative pressures. The spectral content of the acoustic emissions from individual microbubble was compared to the bubble dynamics observed by the imaging. As expected from prior work, both the number of discrete harmonics and broadband content in the emissions increased with increasing amplitude of bubble oscillation. There was no clear correlation, however, between the presence of ultra and sub-harmonic components and bubble behaviour. Indeed, these components were frequently absent. Moreover, phenomena such as microjetting, fragmentation and coalescence, that could produce very different effects in tissue, were indistinguishable acoustically. The results thus indicate that the definition of cavitation thresholds or doses should be very carefully considered depending on the therapeutic effect (or avoidance of unwanted bioeffects) required for a particular application.

Real time cavitation monitoring during ultrasound-mediated drug delivery in normothermically perfused human tumour-bearing livers

In the fight against a broad spectrum of human diseases, cavitation techniques show great promise for overcoming physical barriers that lead to suboptimal uptake of passively administered therapeutics. However, small animal testing of candidate therapies remains a poor predictor of clinical success. Here we demonstrate ultrasound-mediated drug delivery in normal and tumour-bearing human livers infused with protein-based cavitation nuclei (PCaN). Whole and partial human livers were obtained immediately from hepatectomy surgeries and were normothermically sustained using a clinically approved perfusion system (OrganOx Metra). Ultrasound was applied using a 0.5 MHz focused source (Sonic Concepts H107) and was monitored with a calibrated linear array (ATS L7-4) for real time structural and cavitational imaging implemented on an array controller (Verasonics Vantage 256). Specifically, the therapy process was monitored using passive acoustic mapping (PAM) of broadband cavitation emissions, employing a non-adaptive beamformer that deconvolves the array point spread function. Levels of fluorescently labelled drugs incorporated in and co-administered with the PCaN were quantified in blood and tissue samples collected during and following treatment, respectively. This presentation highlights PAM observations of broadband cavitation persistence and drug delivery in untargeted and targeted tissues, including the first ever experiments in tumour-bearing human livers.

High-resolution acoustic mapping of tunable gelatin-based phantoms for ultrasound tissue characterization.

Background: The choice of gelatin as the phantom material is underpinned by several key advantages it offers over other materials in the context of ultrasonic applications. Gelatin exhibits spatial and temporal uniformity, which is essential in creating reliable tissue-mimicking phantoms. Its stability ensures that the phantom's properties remain consistent over time, while its flexibility allows for customization to match the acoustic characteristics of specific tissues, in addition to its low levels of ultrasound scattering. These attributes collectively make gelatin a preferred choice for fabricating phantoms in ultrasound-related research. Methods: We developed gelatin-based phantoms with adjustable parameters and conducted high-resolution measurements of ultrasound wave attenuation when interacting with the gelatin phantoms. We utilized a motorized acoustic system designed for 3D acoustic mapping. Mechanical evaluation of phantom elasticity was performed using unconfined compression tests. We particularly examined how varying gelatin concentration influenced ultrasound maximal intensity and subsequent acoustic attenuation across the acoustic profile. To validate our findings, we conducted computational simulations to compare our data with predicted acoustic outcomes. Results: Our results demonstrated high-resolution mapping of ultrasound waves in both gelatin-based phantoms and plain fluid environments. Following an increase in the gelatin concentration, the maximum intensity dropped by 30% and 48% with the 5MHz and 1MHz frequencies respectively, while the attenuation coefficient increased, with 67% more attenuation at the 1MHz frequency recorded at the highest concentration. The size of the focal areas increased systematically as a function of increasing applied voltage and duty cycle yet decreased as a function of increased ultrasonic frequency. Simulation results verified the experimental results with less than 10% deviation. Conclusion: We developed gelatin-based ultrasound phantoms as a reliable and reproducible tool for examining the acoustic and mechanical attenuations taking place as a function of increased tissue elasticity and stiffness. Our experimental measurements and simulations gave insight into the potential use of such phantoms for mimicking soft tissue properties.

3-D Passive Cavitation Imaging Using Adaptive Beamforming and Matrix Array Transducer With Random Apodization.

With the development of promising cavitation-based treatments, the interest in cavitation monitoring with passive acoustic mapping (PAM) is significantly increasing. While most of studies regarding PAM are performed in 2-D, 3-D imaging modalities are getting more attention relying on either custom-made or commercial matrix probes. Unless specific phased-arrays are used for a specific application, limitations due to probe apertures often results in poor performances of the 3-D mapping, due to the use of a delay-and-sum (DAS) classic beamformer, which results in strong artifacts and large main lobe sizes. In this article, 3D-PAM is achieved by performing adaptive beamforming in the frequency domain (FD) in 3-D, and using a random sparse apodization of a commercial matrix array driving only 256 elements among the 1024 available. It reduces the computation time and makes use of only one 256-channel research platform. Three beamformers have been implemented in 3-D and in the FD: the DAS beamformer, which corresponds to the beamformer used in previous 3D-PAM studies, the robust capon beamformer (RCB), an adaptive algorithm widely used in 2D-PAM for its high performances, and the MidWay (MW) beamformer, an adaptive algorithm with a computation complexity equivalent to the one of DAS. These algorithms are evaluated both in simulations and experiments with a harmonic source at different positions, and are also applied to real cavitation signals. The results show that, in the case of matrix arrays of small aperture such as generic commercial matrix probes, the DAS beamformer leads to large main lobe sizes, while adaptive beamformers largely improve the performances of the mapping. The low computation time and its parameter-free character make MW beamformer a good compromise for 3D-PAM applications. It thus appears that a random sparse apodization combined with adaptive beamforming is a good solution to achieve high-performance 3D-PAM with manageable devices.

Towards better differentiation of archaeological objects based on geomorphometric features of a digital elevation model, the case of the Old Oder Canal

AbstractLimited visibility in the underwater environment often restricts opportunities for archaeological prospection. Especially in reservoirs with a high content of suspended solids, methods based on acoustics prove to be extremely useful. This study represents the first high‐resolution acoustic mapping and archaeological prospections of the Old Oder Canal, which has extremely poor visibility. The study site is located near the town of Krosno Odrzańskie in Poland. The town is one of the country's most significant river crossings and settlements of mediaeval origin (including its wooden bridges). The following research objectives were identified: (1) exploration of the Old Oder Canal, including underwater acoustics and archaeological prospections; (2) analysis and interpretation of the study area based on acquired datasets; and (3) evaluation of secondary features of the river bathymetry for identification of archaeological objects. Possible locations of archaeological objects were determined based on analysis and interpretation of multibeam echosounder measurements of the riverbed. Fieldwork allowed structural elements of mediaeval bridges to be found and dendrochronological sampling performed. Feature selection analyses allowed the determination and evaluation of geomorphometric attributes, combining the characteristics of the discovered objects and diagnostics in order better to differentiate archaeological remains. Proposed secondary features may facilitate archaeological explorations in difficult environments.

Optimization of passive acoustic bedload monitoring in rivers by signal inversion

Abstract. Recent studies have shown that hydrophone sensors can monitor bedload flux in rivers by measuring the self-generated noise (SGN) emitted by bedload particles when they impact the riverbed. However, experimental and theoretical studies have shown that the measured SGN depends not only on bedload flux intensity but also the propagation environment, which differs between rivers. Moreover, the SGN can propagate far from the acoustic source and be well measured at distant river positions without bedload transport. It has been shown that this dependency of the measured SGN data on the propagation environment can significantly affect the performance of monitoring bedload flux by hydrophone techniques. In this article, we propose an inversion model to solve the problem of the SGN propagation and integration effect. In this model, we assume that the riverbed acts as SGN source areas with intensity proportional to the local bedload flux. The inversion model locates the SGN sources and calculates their corresponding acoustic power by solving a system of linear algebraic equations, accounting for the actual measured cross-sectional acoustic power (acoustic mapping) and attenuation properties. We tested the model using data from measured bedload SGN profiles (acoustic mapping with a drift boat) and bedload flux profiles (direct sampling with an Elwha sampler) acquired during two field campaigns conducted in 2018 and 2021 on the Giffre river in the French Alps. Results confirm that the bedload flux measured at different verticals on the river cross-section correlates more with the inversed acoustic power than measured acoustic power. Moreover, it was possible to fit data from the two field campaigns with a common curve after inversion, which was not possible with the measured acoustic data. The results of the inversion model, compared to measured data, show the importance of considering the propagation effect when using the hydrophone technique and offer new perspectives for the calibration of bedload flux with SGN in rivers.

Passive Acoustic Mapping for Convex Arrays With the Helical Wave Spectrum Method.

Passive acoustic mapping (PAM) has emerged as a valuable imaging modality for monitoring the cavitation activity in focused ultrasound therapies. When it comes to imaging in the human abdomen, convex arrays are preferred due to their large acoustic window. However, existing PAM methods for convex arrays rely on the computationally expensive delay-and-sum (DAS) operation limiting the image reconstruction speed when the field-of-view (FOV) is large. In this work, we propose an efficient and frequency-selective PAM method for convex arrays. This method is based on projecting the helical wave spectrum (HWS) between cylindrical surfaces in the imaging field. Both the in silico and in vitro experiments showed that the HWS method has comparable image quality and similar acoustic cavitation source localization accuracy as the DAS-based methods. Compared to the frequency-domain and time-domain DAS methods, the time-complexity of the HWS method is reduced by one order and two orders of magnitude, respectively. A parallel implementation of the HWS method realized millisecond-level image reconstruction speed. We also show that the HWS method is inherently capable of mapping microbubble (MB) cavitation activity of different status, i.e., no cavitation, stable cavitation, or inertial cavitation. After compensating for the lens effects of the convex array, we further combined PAM formed by the HWS method and B-mode imaging as a real-time dual-mode imaging approach to map the anatomical location where MBs cavitate in a liver phantom experiment. This method may find use in applications where convex arrays are required for cavitation activity monitoring in real time.

An Inverse Method Using Cross-Spectral Matrix Fitting for Passive Cavitation Imaging.

High-intensity focused ultrasound (HIFU) can produce cavitation, which requires monitoring for specific applications such as sonoporation, targeted drug delivery, or histotripsy. Passive acoustic mapping has been proposed in the literature as a method for monitoring cavitation, but it lacks spatial resolution, primarily in the axial direction, due to the absence of a time reference. This is a common issue with passive imaging compared to standard pulse-echo ultrasound. In order to improve the axial resolution, we propose an adaptation of the cross spectral matrix fitting (CMF) method for passive cavitation imaging, which is based on the resolution of an inverse problem with different regularizations that promote sparsity in the reconstructed cavitation maps: Elastic Net (CMF-ElNet) and sparse Total Variation (CMF-spTV). The results from both simulated and experimental data are presented and compared to state-of-the-art approaches, such as the frequential delay-and-sum (DAS) and the frequential robust capon beamformer (RCB). We show the interest of the method for improving the axial resolution, with an axial full width half maximum (FWHM) divided by 3 and 5 compared to RCB and DAS, respectively. Moreover, CMF-based methods improve contrast-to-noise ratio (CNR) by more than 15 dB in experimental conditions compared to RCB. We also show the advantage of the sparse Total Variation (spTV) prior over Elastic Net (ElNet) when dealing with cloud-shaped cavitation sources, that can be assumed as sparse grouped sources.