Erasmia Lyka

Sum-of-harmonics method for improved narrowband and broadband signal quantification during passive monitoring of ultrasound therapies

Passive Acoustic Mapping (PAM) enables real-time monitoring of ultrasound therapies by beamforming acoustic emissions emanating from the ultrasound focus. Reconstruction of the narrowband or broadband acoustic emissions component enables mapping of different physical phenomena, with narrowband emissions arising from non-linear propagation and scattering, non-inertial cavitation or tissue boiling, and broadband (generally, of significantly lower amplitude) indicating inertial cavitation. Currently, accurate classification of the received signals based on pre-defined frequency-domain comb filters cannot be guaranteed because varying levels of leakage occur as a function of signal amplitude and the choice of windowing function. This work presents a time-domain parametric model aimed at enabling accurate estimation of the amplitude of time-varying narrowband components in the presence of broadband signals. Conversely, the method makes it possible to recover a weak broadband signal in the presence of a dominant harmonic or other narrowband component. Compared to conventional comb filtering, the proposed sum-of-harmonics method enables PAM of cavitation sources that better reflect their physical location and extent.

Real-time three-dimensional passive cavitation detection for clinical high intensity focused ultrasound systems

Bubble activity during High Intensity Focussed Ultrasound (HIFU) surgery has been linked with desirable effects, such as an enhanced heat deposition caused by inertial cavitation, and undesirable effects, such as lesion migration caused by boiling bubbles. There is presently no reliable way of achieving spatiotemporal monitoring of cavitation activity during clinical HIFU treatments. In the present work, a near-acoustically- transparent two-dimensional 32-element PVDF array was designed and mounted on the therapy transducer of a clinical HIFU device (Model JC200, Chongqing Haifu) to enable detection of acoustic emissions arising from cavitation during therapy. The signal detected by each of the elements was digitized and processed in real time on a Graphical Processing Unit (GPU), and beamformed using our previously described Passive Acoustic Mapping (PAM) algorithm to produce real-time three-dimensional (3D) maps of cavitation activity with a frame rate in excess of 5 Hz. The system was initially validated in agar-based tissue-mimicking materials, demonstrating that the displayed volume of cavitation activity agreed with predictions based on in situ pressure calibrations. The system was further validated during clinical HIFU treatments of kidney tumour, liver tumour and uterine fibroid ablation, and was found to enable accurate localization of the HIFU focus at sub-lesioning intensities.

Multi-source passive acoustic source localisation

Accurate acoustic source localisation (ASL) has significant potential to improve diagnostic ultrasound imaging of small vessel structures. Recently, super-resolution imaging techniques for both active and passive source mapping have been developed; however, these approaches assume the presence of a single source within the point spread function (PSF) of the system. In reality, multiple sources may be present, for example, when high concentrations of contrast agent are used. We use a two-step approach to localise multiple sources within the PSF of a clinically relevant passive cavitation detection system. First, we apply a Blind Source Separation (BSS) technique known as Independent Component Analysis, which relies on higher-order statistics to separate and reconstruct signals originating from independent sources. Second, we determine the time-difference-of-arrival of the separated signals on each receiver, and perform ASL by fitting a polynomial using least squares regression. Simulation and experimental results demonstrate that with the BSS-ASL combination, multiple sources aligned axially within the PSF of a single array located 70 mm from the focus can be localised with sub-millimetre resolution. We verify sources as distinct using Passive Acoustic Maps generated from a perpendicular linear array. Further work is necessary to ascertain performance limits and to validate the technique in vivo.Accurate acoustic source localisation (ASL) has significant potential to improve diagnostic ultrasound imaging of small vessel structures. Recently, super-resolution imaging techniques for both active and passive source mapping have been developed; however, these approaches assume the presence of a single source within the point spread function (PSF) of the system. In reality, multiple sources may be present, for example, when high concentrations of contrast agent are used. We use a two-step approach to localise multiple sources within the PSF of a clinically relevant passive cavitation detection system. First, we apply a Blind Source Separation (BSS) technique known as Independent Component Analysis, which relies on higher-order statistics to separate and reconstruct signals originating from independent sources. Second, we determine the time-difference-of-arrival of the separated signals on each receiver, and perform ASL by fitting a polynomial using least squares regression. Simulation and experimental ...

Passive acoustic mapping of extravasation for vascular permeability assessment

Prior research has demonstrated that Passive Acoustic Mapping (PAM) enables real-time monitoring of cavitation activity occurring within the vasculature to achieve drug delivery and/or opening of the Blood Brain Barrier. In the present work, we focus on whether sub-micron cavitation nuclei can be imaged once extravasated. This would provide a means of determining both vascular permeability before or after ultrasound exposure, and real-time monitoring of successful drug delivery. A key challenge in achieving these objectives is the spatial resolution of PAM. A novel bistatic setup was used to achieve sub-millimetre resolution both axially and transversely to two imaging arrays. A vertically oriented flow channel in a tissue mimicking phantom was placed at the focus of two perpendicular confocal HIFU transducers, each with a coaxial linear imaging array. Sequential acoustic excitation at 0.5 or 1.55 MHz was used to first extravasate, and then re-excite the nuclei once extravasated. The lower frequency creat...

Passive acoustic mapping of cavitation during shock wave lithotripsy

Passive acoustic mapping (PAM) has previously been used to localize inertial cavitation during high intensity focused ultrasound. Here, this technique has been applied to shock wave lithotripsy (SWL), a non-invasive procedure whereby kidney stones are fragmented. Conventional diagnostic ultrasound probes were used to detect acoustic emissions during SWL. Signals consisted of reverberation sound from the incident shock wave followed, several hundred microseconds later, by emissions from cavitation collapses. Time-gating was used to isolate the cavitation signals, which were then processed using PAM to create spatial maps of the cavitation activity. Experiments in water indicated the spatial resolution was an ellipsoidal volume 5mm long by 1mm wide. Experiments were carried out in ex vivo pig kidneys and it was observed that cavitation was initiated in the region of the focus but moved laterally by up to 10mm and during treatment exhibited a general migration towards the source. These results suggest that P...

Optimal beamforming using higher order statistics for passive acoustic mapping

Passive Acoustic Mapping (PAM) of sources of nonlinear acoustic emissions has been extensively investigated for monitoring ultrasound therapies. Optimal data-adaptive beamforming algorithms, such as Robust Capon Beamformer (RCB), were readily proposed as a means of improving source localization, accounting simultaneously for array configuration and calibration errors. RCB, however, assumes that signal samples follow a Gaussian distribution. Aiming at improving the spatial resolution of PAM, especially in the axial direction with respect to the array, we propose an alternative beamforming approach, Robust Beamforming by Linear Programming (RLPB). This method makes no assumptions on the statistical distribution of the received signals, and exploits not only the variance but also higher-order-statistics (HOS) of the received signals. Performance evaluation on simulated and in vitro experimental data suggests improvement in spatial resolution on the order of 20% and 15% in the axial and transverse directions ...

Cavitation enhanced drug delivery in-vivo using combined B-mode guidance and real-time passive acoustic mapping: Challenges and results

Inertial cavitation nucleated by nano-scale sonosensitive particles (SSPs) at modest peak negative pressures (~1 MPa at 500 kHz) and monitored by passive acoustic mapping (PAM) has been recently shown to improve the dose and distribution of anti-cancer agents during ultrasound (US) enhanced delivery (Myers 2016, Kwan 2015). As applications of therapy monitoring using PAM have advanced rapidly including its use in clinical trials, means of validating the performance of PAM in-vivo remains a major focus of efforts. For drug delivery, PAM should not only quickly and reliably detect and localize desired and undesired cavitation, but it should provide some predictor of successful delivery. In-vivo experiments using PAM in subcutaneous tumor implanted murine models across a range of cancer cell lines (HEPG2, SKOV, EMT6, CT-26) demonstrate the detection of inertial cavitation by SSPs in the target regions when sonicated by US, but no cavitation with US alone. Additionally when SSPs are co-administered with an on...

A sum-of-harmonics time-domain method to distinguish harmonic and broadband signals during passive acoustic mapping of ultrasound therapies

Passive Acoustic Mapping (PAM), a novel technique for real-time monitoring of ultrasound-based therapies, performs passive beamforming of nonlinear acoustic emissions simultaneously received on an array of transducers. These emissions can be classified according to their frequency content into harmonic, indicative of nonlinear scattering, stable cavitation or tissue boiling, and broadband, indicative of inertial cavitation. However, the magnitude of coherent reflections often greatly exceeds that of incoherent broadband emissions arising from inertial cavitation and the ability to distinguish these two components is, thus, key to successful detection and mapping of inertial cavitation. We propose a novel time-domain-based filter that uses a parametric model in order to estimate a time-varying amplitude harmonic component in the presence of a lower-amplitude broadband signal. Performance evaluation on simulated and experimental data has shown that the model is able to accurately detect a time-varying ampli...