Refraction-Corrected Transcranial Ultrasound Imaging Through the Human Temporal Window Using a Single Probe

Abstract

Transcranial ultrasound imaging (TUI) is a diagnostic modality with numerous applications, but unfortunately, it is hindered by phase aberration caused by the skull. In this article, we propose to reconstruct a transcranial B-mode image with a refraction-corrected synthetic aperture imaging (SAI) scheme. First, the compressional sound velocity of the aberrator (i.e., the skull) is estimated using the bidirectional headwave technique. The medium is described with four layers (i.e., lens, water, skull, and water), and a fast marching method calculates the travel times between individual array elements and image pixels. Finally, a delay-and-sum algorithm is used for image reconstruction with coherent compounding. The point spread function (PSF) in a wire phantom image and reconstructed with the conventional technique (using a constant sound speed throughout the medium), and the proposed method was quantified with numerical synthetic data and experiments with a bone-mimicking plate and a human skull, compared with the PSF achieved in a ground truth image of the medium without the aberrator (i.e., the bone plate or skull). A phasedarray transducer (P4-1, ATL/Philips, 2.5 MHz, 96 elements, pitch = 0.295 mm) was used for the experiments. The results with the synthetic signals, the bone-mimicking plate, and the skull indicated that the proposed method reconstructs the scatterers with an average lateral/axial localization error of 0.06/0.14 mm, 0.11/0.13 mm, and 1.0/0.32 mm,

Accelerated 2-D Real-Time Refraction-Corrected Transcranial Ultrasound Imaging

“…Several techniques exist to calculate these travel times in a layered medium. Eikonal solvers propagate the wavefront through the medium to calculate the travel-time from a array element or a virtual point source to an image pixel in the brain region [38,[41][42][43]. Another approach is iterative twopoint ray tracing, where the ray path providing the minimum travel time (Fermat's principle) is searched iteratively and accurately using a method for function minimization, for instance a ray bending approach using Brent's algorithm [39][40][41].…”

“…In this work, the same numerical model and experimental setups reported in [38] were used for the sake of comparison (except for evaluating the effect of pixel size on the resolution, discussed in Section II.D., and power Doppler imaging (shown in Figure 3(b))). For the numerical study, the k-Wave MATLAB toolbox was used in a two dimensional (2D) lossless medium, and the shear wave propagation was ignored [78].…”

“…A singleelement transmission sequence was used to excite the medium for generating the numerical and experimental B-mode images. A 4.2 mm-thick bone-mimicking plate (Sawbones, Pacific Research Laboratory, Inc., Vashon, WA) (see Figure 2(b) of [38]) having attenuation close to that in low-porosity cortical bone, a compressional wave speed in the image plane of 3000 m/s and mass density of 1640 kg/m 3 [41] was used in the first experiment. A sagittally-cut human skull (see Figure 2(c) of [38]) was used in the second experiment.…”

“…Another approach is using either ultrasound measurements [31][32][33] or CT/MRI scans of the skull to obtain the true geometry and sound speed of the temporal bone prior to image reconstruction, and then correct phase aberration and refraction during image reconstruction [34][35][36][37]. Recently, we have shown the feasibility of single-sided two-dimensional transcranial ultrasound through the human temporal window using a single handheld commercial probe, where the position, true geometry and sound speed of the bone layer were estimated for an accurate correction of phase aberration and refraction [38,39]. The same methodology was used before for in vivo imaging of the inner structure of the Accelerated 2D Real-Time Refraction-Corrected Transcranial Ultrasound Imaging Moein Mozaffarzadeh*, Eric Verschuur, Martin D. Verweij, Nico de Jong, Guillaume Renaud R radius and tibia bone [40,41].…”

“…While promising results were achieved, no fast-enough implementation is available for realtime transcranial imaging, i.e. enabling an imaging rate of at least 20 images per second, which limits its application in practice, especially for translation to 3D TUI [38,[42][43][44]. Realtime image reconstruction is essential in practice as it allows the operator to freely move the probe find the proper imaging window and imaging plane, and look for biomarkers related to a possible brain disorder in either B-mode or flow images.…”