Matthew McCormick

Preliminary in vivo atherosclerotic carotid plaque characterization using the accumulated axial strain and relative lateral shift strain indices

In this paper, we explore two parameters or strain indices related to plaque deformation during the cardiac cycle, namely, the maximum accumulated axial strain in plaque and the relative lateral shifts between plaque and vessel wall under in vivo clinical ultrasound imaging conditions for possible identification of vulnerable plaque. These strain indices enable differentiation between calcified and lipidic plaque tissue utilizing a new perspective based on the stiffness and mobility of the plaque. In addition, they also provide the ability to distinguish between softer plaques that undergo large deformations during the cardiac cycle when compared to stiffer plaque tissue. Soft plaques that undergo large deformations over the cardiac cycle are more prone to rupture and to release micro-emboli into the cerebral bloodstream. The ability to identify vulnerable plaque, prone to rupture, would significantly enhance the clinical utility of this method for screening patients. We present preliminary in vivo results obtained from ultrasound radio frequency data collected over 16 atherosclerotic plaque patients before these patients undergo a carotid endarterectomy procedure. Our preliminary in vivo results indicate that the maximum accumulated axial strain over a cardiac cycle and the maximum relative lateral shift or displacement of the plaque are useful strain indices that provide differentiation between soft and calcified plaques.

Methods for robust in vivo strain estimation in the carotid artery.

A hierarchical block-matching motion tracking algorithm for strain imaging is presented. Displacements are estimated with improved robustness and precision by utilizing a Bayesian regularization algorithm and an unbiased subsample interpolation technique. A modified least-squares strain estimator is proposed to estimate strain images from a noisy displacement input while addressing the motion discontinuity at the wall-lumen boundary. Methods to track deformation over the cardiac cycle incorporate a dynamic frame skip criterion to process data frames with sufficient deformation to produce high signal-to-noise displacement and strain images. Algorithms to accumulate displacement and/or strain on particles in a region of interest over the cardiac cycle are described. New methods to visualize and characterize the deformation measured with the full 2D strain tensor are presented. Initial results from patients imaged prior to carotid endarterectomy suggest that strain imaging detects conditions that are traditionally considered high risk including soft plaque composition, unstable morphology, abnormal hemodynamics and shear of plaque against tethering tissue can be exacerbated by neoangiogenesis. For example, a maximum absolute principal strain exceeding 0.2 is observed near calcified regions adjacent to turbulent flow, protrusion of the plaque into the arterial lumen and regions of low echogenicity associated with soft plaques. Non-invasive carotid strain imaging is therefore a potentially useful tool for detecting unstable carotid plaque.

Anechoic sphere phantoms for estimating 3-D resolution of very-high-frequency ultrasound scanners

Two phantoms have been constructed for assessing performance of high-frequency ultrasound imagers. They also allow for periodic quality assurance tests and training technicians in the use of higher-frequency scanners. The phantoms contain eight blocks of tissue-mimicking material; each block contains a spatially random distribution of suitably small anechoic spheres having a small distribution of diameters. The eight mean sphere diameters are distributed from 0.10 to 1.09 mm. The two phantoms differ primarily in terms of the frequency dependence of the backscatter coefficient of the background material. Because spheres have no preferred orientation, all three (spatial) dimensions of resolution contribute to sphere detection on an equal basis; thus, the resolution is termed 3-D. Two high-frequency scanners are compared. One employs single-element (fixed focus) transducers (25 and 55 MHz), and the other employs variable focus linear arrays (20, 30, and 40 MHz). The depth range for detection of spheres of each size is determined corresponding to determination of 3-D resolution as a function of depth. As expected, the single-element transducers are severely limited in useful imaging depth ranges compared with the linear arrays. In this preliminary report, only one human observer analyzed images.

ITK: enabling reproducible research and open science

Reproducibility verification is essential to the practice of the scientific method. Researchers report their findings, which are strengthened as other independent groups in the scientific community share similar outcomes. In the many scientific fields where software has become a fundamental tool for capturing and analyzing data, this requirement of reproducibility implies that reliable and comprehensive software platforms and tools should be made available to the scientific community. The tools will empower them and the public to verify, through practice, the reproducibility of observations that are reported in the scientific literature. Medical image analysis is one of the fields in which the use of computational resources, both software and hardware, are an essential platform for performing experimental work. In this arena, the introduction of the Insight Toolkit (ITK) in 1999 has transformed the field and facilitates its progress by accelerating the rate at which algorithmic implementations are developed, tested, disseminated and improved. By building on the efficiency and quality of open source methodologies, ITK has provided the medical image community with an effective platform on which to build a daily workflow that incorporates the true scientific practices of reproducibility verification. This article describes the multiple tools, methodologies, and practices that the ITK community has adopted, refined, and followed during the past decade, in order to become one of the research communities with the most modern reproducibility verification infrastructure. For example, 207 contributors have created over 2400 unit tests that provide over 84% code line test coverage. The Insight Journal, an open publication journal associated with the toolkit, has seen over 360,000 publication downloads. The median normalized closeness centrality, a measure of knowledge flow, resulting from the distributed peer code review system was high, 0.46.

Bayesian Regularization Applied to Ultrasound Strain Imaging

Noise artifacts due to signal decorrelation and reverberation are a considerable problem in ultrasound strain imaging. For block-matching methods, information from neighboring matching blocks has been utilized to regularize the estimated displacements. We apply a recursive Bayesian regularization algorithm developed by Hayton et al. [Artif. Intell., vol. 114, pp. 125-156, 1999] to phase-sensitive ultrasound RF signals to improve displacement estimation. The parameter of regularization is reformulated, and its meaning examined in the context of strain imaging. Tissue-mimicking experimental phantoms and RF data incorporating finite-element models for the tissue deformation and frequency-domain ultrasound simulations are used to compute the optimal parameter with respect to nominal strain and algorithmic iterations. The optimal strain regularization parameter was found to be twice the nominal strain and did not vary significantly with algorithmic iterations. The technique demonstrates superior performance over median filtering in noise reduction at strains 5% and higher for all quantitative experiments performed. For example, the strain SNR was 11 dB higher than that obtained using a median filter at 7% strain. It has to be noted that for applied deformations lower than 1%, since signal decorrelation errors are minimal, using this approach may degrade the displacement image.

In vivo attenuation and equivalent scatterer size parameters for atherosclerotic carotid plaque: Preliminary results

We have previously reported on the equivalent scatterer size, attenuation coefficient, and axial strain properties of atherosclerotic plaque ex vivo. Since plaque structure and composition may be damaged during a carotid endarterectomy procedure, characterization of in vivo properties of atherosclerotic plaque is essential. The relatively shallow depth of the carotid artery and plaque enables non-invasive evaluation of carotid plaque utilizing high frequency linear-array transducers. We investigate the ability of the attenuation coefficient and equivalent scatterer size parameters to differentiate between calcified, and lipidic plaque tissue. Softer plaques especially lipid rich and those with a thin fibrous cap are more prone to rupture and can be classified as unstable or vulnerable plaque. Preliminary results were obtained from 10 human patients whose carotid artery was scanned in vivo to evaluate atherosclerotic plaque prior to a carotid endarterectomy procedure. Our results indicate that the equivalent scatterer size obtained using Farans scattering theory for calcified regions are in the 120-180 microm range while softer regions have larger equivalent scatterer size distribution in the 280-470 microm range. The attenuation coefficient for calcified regions as expected is significantly higher than that for softer regions. In the frequency bandwidth ranging from 2.5 to 7.5 MHz, the attenuation coefficient for calcified regions lies between 1.4 and 2.5 dB/cm/MHz, while that for softer regions lies between 0.3 and 1.3 dB/cm/MHz.

An Approach to Unbiased Subsample Interpolation for Motion Tracking

Accurate subsample displacement estimation is necessary for ultrasound elastography because of the small deformations that occur and the subsequent application of a derivative operation on local displacements. Many of the commonly used subsample estimation techniques introduce significant bias errors. This article addresses a reduced bias approach to subsample displacement estimations that consists of a two-dimensional windowed-sinc interpolation with numerical optimization. It is shown that a Welch or Lanczos window with a Nelder–Mead simplex or regular-step gradient-descent optimization is well suited for this purpose. Little improvement results from a sinc window radius greater than four data samples. The strain signal-to-noise ratio (SNR) obtained in a uniformly elastic phantom is compared with other parabolic and cosine interpolation methods; it is found that the strain SNR ratio is improved over parabolic interpolation from 11.0 to 13.6 in the axial direction and 0.7 to 1.1 in the lateral direction for an applied 1% axial deformation. The improvement was most significant for small strains and displacement tracking in the lateral direction. This approach does not rely on special properties of the image or similarity function, which is demonstrated by its effectiveness with the application of a previously described regularization technique.

Building bridges between cellular and molecular structural biology

The integration of cellular and molecular structural data is key to understanding the function of macromolecular assemblies and complexes in their in vivo context. Here we report on the outcomes of a workshop that discussed how to integrate structural data from a range of public archives. The workshop identified two main priorities: the development of tools and file formats to support segmentation (that is, the decomposition of a three-dimensional volume into regions that can be associated with defined objects), and the development of tools to support the annotation of biological structures. DOI: http://dx.doi.org/10.7554/eLife.25835.001

Absolute backscatter coefficient estimates of tissue-mimicking phantoms in the 5–50 MHz frequency range

Absolute backscatter coefficients in tissue-mimicking phantoms were experimentally determined in the 5-50 MHz frequency range using a broadband technique. A focused broadband transducer from a commercial research system, the VisualSonics Vevo 770, was used with two tissue-mimicking phantoms. The phantoms differed regarding the thin layers covering their surfaces to prevent desiccation and regarding glass bead concentrations and diameter distributions. Ultrasound scanning of these phantoms was performed through the thin layer. To avoid signal saturation, the power spectra obtained from the backscattered radio frequency signals were calibrated by using the signal from a liquid planar reflector, a water-brominated hydrocarbon interface with acoustic impedance close to that of water. Experimental values of absolute backscatter coefficients were compared with those predicted by the Faran scattering model over the frequency range 5-50 MHz. The mean percent difference and standard deviation was 54% ± 45% for the phantom with a mean glass bead diameter of 5.40 μm and was 47% ± 28% for the phantom with 5.16 μm mean diameter beads.

User-Guided Segmentation of Multi-modality Medical Imaging Datasets with ITK-SNAP

ITK-SNAP is an interactive software tool for manual and semi-automatic segmentation of 3D medical images. This paper summarizes major new features added to ITK-SNAP over the last decade. The main focus of the paper is on new features that support semi-automatic segmentation of multi-modality imaging datasets, such as MRI scans acquired using different contrast mechanisms (e.g., T1, T2, FLAIR). The new functionality uses decision forest classifiers trained interactively by the user to transform multiple input image volumes into a foreground/background probability map; this map is then input as the data term to the active contour evolution algorithm, which yields regularized surface representations of the segmented objects of interest. The new functionality is evaluated in the context of high-grade and low-grade glioma segmentation by three expert neuroradiogists and a non-expert on a reference dataset from the MICCAI 2013 Multi-Modal Brain Tumor Segmentation Challenge (BRATS). The accuracy of semi-automatic segmentation is competitive with the top specialized brain tumor segmentation methods evaluated in the BRATS challenge, with most results obtained in ITK-SNAP being more accurate, relative to the BRATS reference manual segmentation, than the second-best performer in the BRATS challenge; and all results being more accurate than the fourth-best performer. Segmentation time is reduced over manual segmentation by 2.5 and 5 times, depending on the rater. Additional experiments in interactive placenta segmentation in 3D fetal ultrasound illustrate the generalizability of the new functionality to a different problem domain.