1,720,952 research outputs found
Multiple Measurement Vector Model for Sparsity-Based Vascular Ultrasound Imaging
No full textUltrasound imaging of the vasculature has major significance for the detection of cardiovascular diseases and cancer. However, limited spatial resolution or long acquisition times of existing techniques limit the visualization of the microvascular structures. Enforcing sparsity in the underlying vasculature as well as exploiting statistical independence between voxels have become prominent for fast super-resolution imaging. However, such a statistical independence may not be valid for all voxels and may hence lead to a distorted signal model. Here we present an image reconstruction method that exploits the sparsity of the vasculature data without distorting the original signal model. We employ a multiple measurement vector (MMV) model to enforce the joint sparsity over the images at different time instants. To reduce the computational complexity of obtaining the solution, the ℓ1-SVD method is applied to the MMV model. We demonstrate that our method improves spatial resolution and provides a clear separation between blood vessels. Although our method is slightly slower than existing approaches, it outperforms them in terms of image reconstruction quality.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work publicSignal Processing System
Spatial compression in ultrasound imaging
No full textHigh quality three dimensional ultrasound imaging is typically attained by increasing the amount of sensors, resulting in complex hardware. Compressing measurements before sensing addresses this problem, and could enable new clinical applications. We have developed an analogue compression technique, by positioning a plastic coding mask in front of the aperture, which distorts the ultrasound field by inducing varying local echo delays. This results in a compression of the spatial ultrasound field across the sensor surface, while retaining sufficient information for 3D imaging. Using only a single sensor, complementary measurements can be obtained by rotation of the sensor and the mask to increase the conditioning of the reconstruction problem. In this work, we study a method to optimize the shape of the coding mask. To this end, we define an approximate signal model that captures the ultrasound response of the mask, and use it to pose mask shape optimization as a sensor selection problem. We solve it by relaxing it to a convex problem, as well as by using a greedy selection method. Our simulation results show that these approaches are able to outperform the random design strategy, in particular when mask rotations are included in the problem.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing SystemsImPhys/Acoustical Wavefield Imagin
Calibration techniques for single-sensor ultrasound imaging with a coding mask
No full textWe consider a model-based ultrasound imaging scenario using a single transducer with a coding mask, and assume that the pulse-echo model is erroneously estimated, resulting in decreased imaging performance. Although the pulse-echo Green's function to each pixel has to be measured to obtain a good model, typically only forward-field measurements are obtained for better SNR, from which the pulse-echo Green's functions are estimated. However, if the transducer's receive transfer function is different from the transmit transfer function, the forward-field measurements do not incorporate the receive transfer function, resulting in an incorrect pulse-echo model. We propose two calibration techniques that start with this erroneous model, and update it using pulse-echo measurements. In the first technique we assume the calibration phantom is known a priori, whereas in the second technique we use multiple random calibration phantoms of which only the second-order statistics are assumed to be known beforehand. Both methods are able to significantly improve the pulse-echo model, strongly improving imaging performance. Our simulation results show that the first technique works best, since there is no uncertainty about the calibration image, whereas the blind calibration technique requires no exact knowledge of the calibration phantom, making it robust to positioning or manufacturing errors.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
Coding Mask Design for Single Sensor Ultrasound Imaging
No full textWe study the design of a coding mask for pulse-echo ultrasound imaging. We are interested in the scenario of a single receiving transducer with an aberrating layer, or ‘mask,’ in front of the transducer's receive surface, with a separate co-located transmit transducer. The mask encodes spatial measurements into a single output signal, containing more information about a reflector's position than a transducer without a mask. The amount of information in such measurements is dependent on the mask geometry, which we propose to optimize using an image reconstruction mean square error (MSE) criterion. We approximate the physics involved to define a linear measurement model, which we use to find an expression for the image error covariance matrix. By discretizing the mask surface and defining a discrete number of mask thickness levels per point on its surface, we show how finding the best mask can be posed as a variation of a sensor selection problem. We propose a convex relaxation in combination with randomized rounding, as well as a greedy optimization algorithm to solve this problem. We show empirically that both algorithms come close to the global optimum. Our simulations further show that the optimized masks have better a MSE than nearly all randomly shaped masks. We observe that an optimized mask amplifies echoes coming from within the region of interest (ROI), and strongly reduces the correlation between echoes of pixels within the ROI.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
Joint Estimation of Hemodynamic Response and Stimulus Function in Functional Ultrasound Using Convolutive Mixtures
No full textFunctional ultrasound (fUS) is an exciting new neuroimaging technique that is able to record brain activity similar to functional magnetic resonance imaging, yet with higher spatiotemporal resolution and at lower cost. We consider the problem of jointly estimating the underlying neural sources and the hemodynamic response function (HRF) from fUS recordings. We propose to model the measured voxel time-series as a convolutive mixture of multiple source signals and solve the blind deconvolution problem via block-term decomposition. This allows us to estimate both the source time courses and a different HRF for each voxel and source combination, which accounts for the variability of HRF across different brain regions and events respectively. The proposed approach is proven to be robust against noise via simulations and further validated on real fUS data by performing a visual experiment on a mouse. The obtained results show that the proposed method is able to recover the timings of the visual paradigm.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
Frequency Domain Joint Estimation of HRF and Stimulus from fUS Data
No full textTo better understand how brain signals are processed and even how the human mind works, analyzing the hemodynamic signal model is one of the most essential steps. In the CUBE group of Erasmus MC, functional ultrasound (fUS) data of a mouse’s brain is recorded. By using this fUS dataset, this thesis will solve the problem regarding the joint estimation of hemodynamic response function (HRF) and the underlying stimulus. Usually, hemodynamic responses are investigated in the time domain, while this thesis provides another perspectivefrom frequency domain signal processing. We consider the hemodynamic response as a convolutive signal mixture, then try to transform it into an instantaneous mixing model by converting the context into the frequency domain. By applying independent vector analysis (IVA), this estimation problem can be solved without facing permutation ambiguity which is a well-unknown problem regarding independent component analysis (ICA). Additional steps before and after IVA are also discussed so that a whole estimation road map is formed.Both simulation and experimental analysis are provided to validate this estimation algorithm. Results show that by using this method, both stimulus and HRF estimation can be achieved satisfyingly in a suitable experimental setting. This thesis provides insights and future potentials for IVA to be further investigated in neural signal processing problems.Electrical Engineering | Circuits and System
Efficient and Flexible Spatiotemporal Clutter Filtering of High Frame Rate Images Using Subspace Tracking
No full textCurrent methods to measure blood flow using ultrafast Doppler imaging often make use of a Singular Value Decomposition (SVD). The SVD has been shown to be an effective way to remove clutter signals associated with slow moving tissue. Conventionally, the SVD is calculated from an ensemble of frames, after which the first dominant eigenvectors are removed. The Power Doppler Image (PDI) is then computed by averaging over the remaining components. The SVD method is computationally intensive and lacks flexibility due to the fixed ensemble length. We propose a method, based on the Projection Approximation Subspace Tracking (PAST) algorithm, which is computationally efficient and allows us to sequentially estimate and remove the principal components, while also offering flexibility for calculating the PDI, e.g. by using any convolutional filter. During a functional ultrasound (fUS) measurement, the intensity variations over time for every pixel were correlated to a known stimulus pattern. The results show that for a pixel chosen around the location of the stimulation electrode, the PAST algorithm achieves a higher Pearson correlation coefficient than the state-of-the-art SVD method, highlighting its potential to be used for fUS measurements.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
Joint optimization of coding mask and scan positions for compressive single sensor imaging
No full textWe study the optimal design of an aperture coding mask, and the optimal sensing positions of a single ultrasound sensor with a scanning configuration. In previous works, we have shown that 3D ultrasound imaging is possible using a randomly shaped coding mask with randomly chosen sensing positions. Here we propose an optimization algorithm for the joint design of the coding mask and the sensing positions. We first define a linear measurement model and parameterize it with respect to the mask shape. To optimize the shape of the mask, we use a greedy descent algorithm to minimize the imaging MSE, assuming a Wiener estimate is used for image reconstruction. To optimize the sensing positions, we pre-define a set of such sensing positions by gridding the measurement plane, and regard each sensing position as a virtual sensor candidate. We then use a greedy sensor selection algorithm to find a good selection of sensing positions. To jointly optimize for both the mask and the sensing positions, we alternatingly optimize between them, keeping either the mask shape or the sensing positions fixed. Using simulations we show that the joint optimization results in better imaging performance than optimizing for the mask or the sensing positions alone, or using a completely random design.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
3D Carotid Artery Flow Imaging Using Compressive Sensing with a Spatial Coding Mask: A Simulation Study
No full textIt has been previously demonstrated that applying an aberrating mask for 2D compressive imaging using a low number of sensors (elements) can significantly improve image resolution, as evaluated via the point spread function. Here we investigate the potential to apply a similar approach for 3D flow monitoring. We conducted a 3D k-Wave simulation using a 5x5 sensor array coupled to a physical coding mask, performing B-mode and power Doppler imaging on a 3D carotid artery flow model. An approximately three times smaller lateral PSF was achieved at the cost of increased background clutter level and slightly increased axial PSF. A better definition of the vessel border and finer flow speckle were observed in power Doppler imaging. Our results suggest that 3D compressive imaging using a very low sensor count of 25 with spatial coding mask has the potential to monitor 3D carotid artery flow.Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Signal Processing System
High-speed Data Acquisition and Processing for Transcranial Functional Ultrasound Brain Imaging
No full textFunctional ultrasound is by now a well-established technique in the neuroscience community to measure brain activity. However, the transcranial application of functional ultrasound on humans, with the exception of the acoustic windows in the skull, remains a huge challenge because of the properties of the cranial bone the acoustic waves will reflect, refract, attenuate or will cause aberration. Therefore, the signal-to-noise ratio (SNR) of the incoming signal is too low for transcranial functional ultrasound (TCfUS). This thesis tries to overcome the SNR problem of TCfUS with the design of a 64-channel ultrasound acquisition system that is focused on achieving the highest SNR possible. This is achieved by placing the analog front-end (AFE) chips as near to the transducer elements as possible and by oversampling the incoming signal with a factor of 25, a theoretical increase of 15 dB of the signal-to-quantization-noise ratio over the current state-of-the-art is estimated. The proposed design is a receive-only system where the transmission of the ultrasound pulses is carried out by a separate ultrasound system. The design splits the acquisition system into a front-end and a back-end subsystem, where the front-end system is implemented using four AFE58JD48 analog front-end chips from Texas Instruments. From here the data samples are transported over fiber optics to the back-end subsystem, which consists of a VCK190 FPGA board from Xilinx, where the samples are processed and/or transported to a workstation for storage. Because the system organization differentiates from more conventional research ultrasound systems, a trade-off is introduced between processing and throughput, resulting in three processing configurations for the FPGA with each a different focus: raw RF data sampling, real-time processing, and hardware processing. Due to time and resource constraints, no measurements and results are available on the SNR and decimation. However, a theoretical exploration has been done on the expandability of the number of channels which was found to be 192 channels for the VCK190.https://ultrasoundbrainimaging.com/ Website CUBE https://neuro.nl/ Website department of Neuroscience at Erasmus MC https://neurocomputinglab.com/ Website Neuro Computing LabComputer Scienc
- …
