arXiv subjects

Dan Adam

Publications and source records attributed to Dan Adam.

6 recordsLinked to original sources

arXiv2018

High quality ultrasonic multi-line transmission through deep learning

Frame rate is a crucial consideration in cardiac ultrasound imaging and 3D sonography. Several methods have been proposed in the medical ultrasound literature aiming at accelerating the image acquisition. In this paper, we consider one such method called \textit{multi-line transmission} (MLT), in which several evenly separated focused beams are transmitted simultaneously. While MLT reduces the acquisition time, it comes at the expense of a heavy loss of contrast due to the interactions between the beams (cross-talk artifact). In this paper, we introduce a data-driven method to reduce the artifacts arising in MLT. To this end, we propose to train an end-to-end convolutional neural network consisting of correction layers followed by a constant apodization layer. The network is trained on pairs of raw data obtained through MLT and the corresponding \textit{single-line transmission} (SLT) data. Experimental evaluation demonstrates significant improvement both in the visual image quality and in objective measures such as contrast ratio and contrast-to-noise ratio, while preserving resolution unlike traditional apodization-based methods. We show that the proposed method is able to generalize well across different patients and anatomies on real and phantom data.

cs.CV↗

arXiv2018

High frame-rate cardiac ultrasound imaging with deep learning

Cardiac ultrasound imaging requires a high frame rate in order to capture rapid motion. This can be achieved by multi-line acquisition (MLA), where several narrow-focused received lines are obtained from each wide-focused transmitted line. This shortens the acquisition time at the expense of introducing block artifacts. In this paper, we propose a data-driven learning-based approach to improve the MLA image quality. We train an end-to-end convolutional neural network on pairs of real ultrasound cardiac data, acquired through MLA and the corresponding single-line acquisition (SLA). The network achieves a significant improvement in image quality for both $5-$ and $7-$line MLA resulting in a decorrelation measure similar to that of SLA while having the frame rate of MLA.

cs.CV↗

arXiv2017

SUSHI: Sparsity-based Ultrasound Super-resolution Hemodynamic Imaging

Identifying and visualizing vasculature within organs and tumors has major implications in managing cardiovascular diseases and cancer. Contrast-enhanced ultrasound scans detect slow-flowing blood, facilitating non-invasive perfusion measurements. However, their limited spatial resolution prevents the depiction of microvascular structures. Recently, super-localization ultrasonography techniques have surpassed this limit. However, they require long acquisition times of several minutes, preventing the detection of hemodynamic changes. We present a fast super-resolution method that exploits sparsity in the underlying vasculature and statistical independence within the measured signals. Similar to super-localization techniques, this approach improves the spatial resolution by up to an order of magnitude compared to standard scans. Unlike super-localization methods, it requires acquisition times of only tens of milliseconds. We demonstrate a temporal resolution of 25Hz, which may enable functional super-resolution imaging deep within the tissue, surpassing the temporal resolution limitations of current super-resolution methods, e.g. in neural imaging. The sub-second acquisitions make our approach robust to motion artifacts, simplifying in-vivo use of super-resolution ultrasound.

physics.med-ph↗

arXiv2016

Fast Vascular Ultrasound Imaging with Enhanced Spatial Resolution and Background Rejection

Ultrasound super-localization microscopy techniques presented in the last few years enable non-invasive imaging of vascular structures at the capillary level by tracking the flow of ultrasound contrast agents (gas microbubbles). However, these techniques are currently limited by low temporal resolution and long acquisition times. Super-resolution optical fluctuation imaging (SOFI) is a fluorescence microscopy technique enabling sub-diffraction limit imaging with high temporal resolution by calculating high order statistics of the fluctuating optical signal. The aim of this work is to achieve fast acoustic imaging with enhanced resolution by applying the tools used in SOFI to contrast-enhance ultrasound (CEUS) plane-wave scans. The proposed method was tested using numerical simulations and evaluated using two in-vivo rabbit models: scans of healthy kidneys and VX-2 tumor xenografts. Improved spatial resolution was observed with a reduction of up to 50% in the full width half max of the point spread function. In addition, substantial reduction in the background level was achieved compared to standard mean amplitude persistence images, revealing small vascular structures within tumors. The scan duration of the proposed method is less than a second while current superlocalization techniques require acquisition duration of several minutes. As a result, the proposed technique may be used to obtain scans with enhanced spatial resolution and high temporal resolution, facilitating flow-dynamics monitoring. Our method can also be applied during a breath-hold, reducing the sensitivity to motion artifacts.

q-bio.TO↗

arXiv2015

Using Anatomical Markers for Left Ventricular Segmentation of Long Axis Ultrasound Images

Left ventricular segmentation is essential for measuring left ventricular function indices. Segmentation of one or several images requires an initial guess of the contour. It is hypothesized here that creating an initial guess by first detecting anatomical markers, would lead to correct detection of the endocardium. The first step of the algorithm presented here includes automatic detection of the mitral valve. Next, the apex is detected in the same frame. The valve is then tracked throughout the cardiac cycle. Contours passing from the apex to each valve corner are then found using a dynamic programming algorithm. The resulting contour is used as an input to an active contour algorithm. The algorithm was tested on 21 long axis ultrasound clips and showed good agreement with manually traced contours. Thus, this study demonstrates that detection of anatomic markers leads to a reliable initial guess of the left ventricle border.

cs.CV↗

arXiv2015

Stable Support Recovery of Stream of Pulses with Application to Ultrasound Imaging

This paper considers the problem of estimating the delays of a weighted superposition of pulses, called stream of pulses, in a noisy environment. We show that the delays can be estimated using a tractable convex optimization problem with a localization error proportional to the square root of the noise level. Furthermore, all false detections produced by the algorithm have small amplitudes. Numerical and in-vitro ultrasound experiments corroborate the theoretical results and demonstrate their applicability for the ultrasound imaging signal processing.

cs.IT↗