Showing papers by "Arun K. Thittai published in 2019"

Compressed Sensing with Gaussian Sampling Kernel for Ultrasound Imaging

Abstract: Recently, compressed sensing (CS) has been applied to ultrasound imaging for either data reduction or frame rate improvement. However, there are no detailed reports yet on strategies for lateral undersampling of channel data in conventional focused beamforming (CFB) and its recovery exploiting the CS approach. We propose a strategic lateral undersampling approach for channel data using the Gaussian sampling scheme and compare it with a direct extension of the often-used uniform undersampling reported for axial undersampling to the lateral direction and 2-D random sampling reported in the literature. As opposed to the reported 2-D random undersampling, we explore undersampling of channel data in the lateral direction by acquiring radiofrequency data from only a reduced number of chosen receive elements and subjecting these data to further undersampling in the axial direction. The effect of the sampling schemes on CS recovery was studied using data from simulations and experiments for various lateral and axial undersampling rates. The results suggest that CS-recovered data from the Gaussian distribution-based channel data subsampling yielded better recovery and contrast in comparison to those obtained from the often-used uniform distribution-based undersampling. Although 90% of the samples from the original data using the proposed sampling scheme were discarded, the contrast of the CS-recovered image was comparable to that of the reference image. Thus, CS with the proposed Gaussian sampling scheme for channel data subsampling not only reduces the data size significantly, but also strategically uses only a few active receive elements in the process; thus, it can provide an attractive option for the affordable point-of-care ultrasound system.

Enhancing Image Quality of Photoacoustic Tomography Using Sub-Pitch Array Translation Approach: Simulation and Experimental Validation

Abstract: Objective : The purpose of this paper is to present a more convenient and practical alternate way of increasing the lateral discrete array sampling while using a typical λ pitch linear array transducer at receive for photoacoustic tomography (PAT) application. Methods: We have employed a linear translation-based approach, in which the array transducer is translated by sub-pitch amount to create an augmented RF frame data having denser lateral spatial sampling. The denser λ /2 and λ /4 pitch data were reconstructed and compared against conventional λ pitch reconstructed PAT image using simulation and tissue mimicking phantom experiments in terms of improvements in resolution and contrast. Results: The results from experiments demonstrate a 34.48% improvement in lateral resolution (LR), measured in terms of full-width at half-maximum of the lateral profile of point spread function, and a maximum of 7-dB improvement in contrast is achieved while using a λ /2-pitch configuration when compared to the conventional λ -pitch configuration. Conclusion: It was demonstrated that λ /2- and λ /4-pitch configurations result in better LR and contrast than λ -pitch configuration. Significance: Based on the results obtained, the proposed method has the potential to serve as an easy-to-integrate and simple way of achieving better image quality without requiring to increase the system complexity with existing transducer array probe technology in regular clinical scanners.

Deep-Learning based Identification of Frames Containing Foetal Gender Region During Early Second Trimester Ultrasound Scanning

Abstract: Sex-selective abortions, based on ultrasound foetal scanning of pregnant women, is a rampant problem in developing countries like India. Although prenatal gender screening has been outlawed, poor enforcement of it has abetted a steady rise in their number. Currently, the B-scans are displayed on the screen in real-time and the operator can make out the gender from the displayed image. Therefore, enforcement of the secrecy has not been effective and has only led to limitations on the usage of the technology. Hence, it may be useful to develop a method that can identify frames containing gender-indicative features in real-time and block it out automatically from the display, thus, preventing unauthorized viewing. In this work, deep learning-based techniques have been explored to detect images containing the gender-defining features among the entire set of images in cine-loop, with a conforming accuracy averaging above 80%.

Lateral Resolution Improvement in Ultrasound Imaging System using Compressed Sensing: Initial Results

Abstract: Compressed-Sensing (CS) has been applied to ultrasound imaging to reduce data or to reduce the data acquisition time. There appears to be no report that uses CS framework to reduce the number of active receive elements in Conventional Focused Beamforming (CFB). Thus, in our previous work, a novel undersampling scheme based on Gaussian distribution was investigated and reported for reducing the number of active receive elements and data in CFB. In this paper, we exploit the Gaussian sampling based CS framework to improve the lateral resolution (LR) of the ultrasound system without increasing the system’s complexity and cost. A notable difference from our previous work being the use of waveatom as the sparsifying basis, instead of 2D-Fourier basis, and analysis of the proposed framework for different receive aperture sizes. Simulation data for this study were generated using Field II, and experimental data were acquired from an in-vitro cyst phantom using Verasonics V-64 ultrasound scanner. The results indicate that the proposed framework of choosing a limited number of receive elements from a receive aperture length that is three or four times the corresponding active aperture size obtained from the same number of consecutive receive elements yields nearly twice an improvement in LR and about 27% increase in contrast to that of CFB reference image. Thus, the findings suggest a possibility to improve the LR of the current ultrasound system without increasing the system complexity, which will be significant for affordable point-of-care ultrasound systems.

Noise Reduction in Inherently low-SNR PLD-based PAT images

Abstract: High energy solid state lasers such as Nd:YAG are most commonly used as source of illumination in Photoacoustic (PA) imaging. However, recently, pulsed laser diodes (PLD)are being increasingly explored as a potential alternative to solid state sources due to its advantages such as portability, high pulse repetition rate and affordability. However, PLD has low energy per pulse which results in weaker signal, especially, from deep-seated photoacoustic targets. Hence, additional noise from external sources, such as, electronic noise, jitter, etc. have more pronounced detrimental effect on the quality of the reconstructed PA images resulting in low Signal-to-Ratio (SNR) values. Recently, several works have been reported to reduce the effect of noise by taking advantage of the high pulse-repetition rate offered by PLD and performing averaging. In this work we demonstrate that by strategically employing noise-reduction filter on the raw data prior to reconstruction stage yields significant improvement in SNR compared to mere averaging (∼61%) in PLD-based PAT systems.

Ultrasound-Based Regularized Log Spectral Difference Method For Monitoring Microwave Hyperthermia

Abstract: The feasibility of using normalized cumulative difference attenuation (NCDA) map for tracking the spatial and temporal evolution of temperature during microwave hyperthermia experiment on in-vitro phantoms is explored in this study. The NCDA maps were estimated from the beamformed ultrasound radio frequency (RF) data using a regularized log spectral difference (RLSD) technique. The NCDA maps were estimated at different time instants for the entire period of the experiment. The contour maps of the NCDA and the ground truth temperature map, obtained using an infra-red(IR) thermal camera corresponding to the ultrasound imaging plane, showed that NCDA was able to locate the axial and lateral co-ordinates of the hotspot with the error of < 1.5 mm axially and < 0.1 mm laterally. The error in the estimated hotspot area was less than 8 %. This preliminary in-vitro study suggests that NCDA maps estimated using RLSD may have potential in evaluating the spatio-temporal evolution of temperature and may help in the development of ultrasound-based image-guided temperature monitoring system for microwave hyperthermia.

Regularized Nakagami Ultrasound Imaging for Microwave Hyperthermia monitoring

Abstract: The study involves development of ultrasound-based regularized Nakagami imaging (RNI) that improves the quality of Nakagami image. The feasibility of RNI to image the spatial and temporal evolution of hotspot during microwave (MW) hyperthermia experiment was explored on in-vitro polyacrylamide gel (PAG)-agar based phantoms. The normalized cumulative differential regularized Nakagami (NCDRN) maps were estimated from the envelope of beamformed ultrasound radiofrequency (RF) data using proposed RNI technique. The NCDRN maps were estimated at different time instants for the entire duration of the experiment. The experiments were carried out on phantoms at power level of 12 W fed to the microwave antenna. The contour maps of the NCDRN and the ground truth temperature map, obtained using an infra-red (IR) thermal camera corresponding to ultrasound imaging plane, showed that NCDRN was able to locate the axial and lateral co-ordinates of the hotspot with an error of < 1.5 mm axially and < 0.4 mm laterally. This preliminary in-vitro study demonstrates that NCDRN maps estimated using the regularized Nakagami imaging may have potential in imaging and evaluating the spatio-temporal evolution of hotspot and may help in the development of ultrasound-based image guided hotspot monitoring system for microwave hyperthermia.

Enhancing Depth of Penetration in PLD-Based Photoacoustic Imaging: Preliminary Results

Abstract: In most cases, high energy solid state lasers such as Nd:YAG are used as source of illumination for Photoacoustic Tomography (PAT). However the bulkiness, high cost and low pulse repetition frequency (PRF) poses a challenge in translating this technology to an affordable clinical imaging option at bed-side. Pulsed Laser Diodes (PLD) on the other hand is portable, inexpensive and offers high PRF. However, the achievable depth of penetration using PLD is much lower than the solid state lasers. In this work, we demonstrate the feasibility of using sub-pitch translation approach on the receive-side ultrasound transducer to increase the depth sensitivity in PAT imaging system while using PLD as a source of illumination. The preliminary results obtained from experiments suggest that the higher density data obtained by augmenting raw RF lines from λ/2 positions of a linear array transducer provides better signal strength from deeper located targets and thereby increasing the depth of penetration by about 15% that reaches up to a depth of 14.3 mm.

Quantitative quasi-static ultrasound elastography using reference layer: Ex-vivo study

Abstract: It is well-documented in the literature that changes in tissue elasticity are generally correlated with disease condition. In the case of diffuse liver disease, the elasticity of the liver reduces progressively. However, this change does not clearly manifest in conventional ultrasound examinations. Although quasi-static elastography is popular in clinical applications where qualitative assessment of relative tissue stiffness is enough, its potential is relatively underutilized in liver imaging due to the need for quantitative stiffness value. Recently, it was demonstrated that using a reference layer of known stiffness, one could produce quantitative modulus elastograms of the target tissue using quasi-static elastography using simulations and phantom experiments. Here, we examined the performance of this approach on ex-vivo goat liver samples and compare the estimated modulus values to that obtained from indentation measurements. The results suggest that using this approach of reference layer yields Young’s modulus values within 10% error compared to the ground truth.

Axial Super-Resolution in Ultrasound Imaging

Abstract: A fundamental challenge in non-destructive evaluation using ultrasound is to accurately estimate the thicknesses of different layers or cracks present in the object being probed. This inherently corresponds to localizing the point-sources of the reflections from the received signal. Conventional signal processing techniques cannot resolve reflectors whose spacing is below the axial resolution limit, which is of the order of the wavelength of the probing pulse. The objective of this paper is to demonstrate axial super-resolution capability using both simulated and experimental ultrasound data. We show that the ultrasound reflections could be modelled effectively as FRI signals, which can be sampled at sub-Nyquist rates. The FRI sampling method brings the reconstruction problem within a parametric estimation framework, for which efficient high-resolution spectral estimation techniques are available. We experimentally demonstrate that the proposed technique is able to resolve the thicknesses of layers of custom designed Agarose phantoms that are up to 2.25 times below the conventional resolution limit.