Jeremy D. Gill

Segmentation of carotid artery in ultrasound images: Method development and evaluation technique

Segmentation of carotid artery lumen in two-dimensional and three-dimensional ultrasonography is an important step in computerized evaluation of arterial disease severity and in finding vulnerable atherosclerotic plaques susceptible to rupture causing stroke. Because of the complexity of anatomical structures, noise as well as the requirement of accurate segmentation, interactions are necessary between observers and the computer segmentation process. In this paper a segmentation process is described based on the deformable model method with only one seed point to guide the initialization of the deformable model for each lumen cross section. With one seed, the initial contour of the deformable model is generated using the entropy map of the original image and mathematical morphology operations. The deformable model is driven to fit the lumen contour by an internal force and an external force that are calculated, respectively, with geometrical properties of deformed contour and with the image gray level features. The evaluation methodology using distance-based and area-based metrics is introduced in this paper. A contour probability distribution (CPD) method for calculating distance-based metrics is introduced. The CPD is obtained by generating contours of the lumen using a set of possible seed locations. The mean contour can be compared to a manual outlined contour to provide accuracy metrics. The variance computed from the CPD can provide metrics of local and global variability. These metrics provide a complete performance evaluation of an interactive segmentation algorithm and a means for comparing different algorithm settings.

Segmentation of carotid artery in ultrasound images

Segmentation of carotid artery lumen in 2D and 3D ultrasonography is an important step in evaluating arterial disease severity and finding vulnerable artherosclerotic plaques susceptible to rupture causing stroke. Because of the complexity of anatomical structures, noise as well as the requirement of accurate segmentation, interactions are necessary between observers and computer segmentation process. We describe a segmentation algorithm based on a discrete dynamic model approach with only one seed point to guide the initialization of the deformable model for each lumen cross-section. With one seed, the initial contour of the deformable model is generated using the entropy map of the original image and mathematical morphology operations. The deformable model is driven to fit the lumen contour by an internal force and an external force that are calculated respectively with geometrical properties of deformed contour and with the image gray level features. We also introduce a set of metrics based on a contour probability distribution function for evaluating the accuracy and variability of the interactive segmentation algorithm. These metrics provide a complete performance evaluation of an interactive segmentation algorithm and a means for comparing different algorithm settings. Seven images of the common, internal and external carotid arteries were chosen to test the segmentation algorithm. The average position error and average variability of the boundary segmentation result are 0.2 mm and 0.25 mm.

3D ultrasound imaging: applications in image-guided therapy and biopsy

Abstract The use of conventional 2D ultrasonography to view 3D anatomy limits our ability to plan and guide interventional procedures. CT and MRI have been used in planning and guiding these procedures due to their ability to provide 3D images with accurate depiction of anatomy. Recent development of 3D ultrasound imaging techniques that are capable of acquiring B-mode, color Doppler and power Doppler images, has allowed the development of image-guided therapy and surgery approaches. Because ultrasound imaging is an inexpensive and compact imaging modality, it is particularly suited for applications in prostate therapy, breast biopsy and monitoring of disease progression in response to therapy.

Analysis of geometrical distortion and statistical variance in length, area, and volume in a linearly scanned 3-D ultrasound image

A linearly scanned three-dimensional (3-D) ultrasound imaging system is considered. The transducer array is initially oriented along the x axis and aimed in the y direction. After being tilted by an angle /spl theta/ about the x axis, and then swiveled by an angle /spl phi/ about the y axis, it is translated in the z direction, in steps of size d, to acquire a series of parallel two-dimensional (2-D) images. From these, the 3-D image is reconstructed, using the nominal values of the parameters (/spl phi/, /spl theta/, d). Thus, any systematic or random errors in these, relative to their actual values (/spl phi//sub 0/, /spl theta//sub 0/, d/sub 0/), will respectively cause distortions or variances in length, area, and volume in the reconstructed 3-D image, relative to the 3-D object. Here, the authors analyze these effects. Compact linear approximations are derived for the relative distortions as functions of the parameter errors, and hence, for the relative variances as functions of the parameter variances. Also, exact matrix formulas for the relative distortions are derived for arbitrary values of (/spl phi/, /spl theta/, d) and (/spl phi//sub 0/, /spl theta//sub 0/, d/sub 0/). These were numerically compared to the linear approximations and to measurements from simulated 3-D images of a cubical object and real 3-D images of a wire phantom. In every case tested, the theory was confirmed within experimental error (0.5%).

Segmentation of ulcerated plaque: a semi-automatic method for tracking the progression of carotid atherosclerosis

A semi-automatic method for segmentation of three-dimensional carotid vascular ultrasound (US) images is presented. The method is based on a dynamic balloon model represented by a triangulated mesh. The mesh is manually placed within the interior of the carotid arteries, then is driven outward until it reaches the vessel wall by applying an inflation force to the mesh. Once the mesh is in close proximity to the vessel wall, it is further deformed using an image-based force, in order to better localize the boundary. The authors examine the ability of the segmentation method to segment in vivo 3D US images of the carotid arteries. They furthermore examine its ability to distinguish subtle changes in vessel morphology, with the ultimate goal being to detect the progression or regression of atherosclerotic plaque. Two nearly identical common carotid vessel phantoms were imaged using a three-dimensional US imaging system, automatically registered, then segmented using the semi-automatic algorithm. The fabrication of the two phantoms was identical except for the inclusion of a hemispherical ulceration cut into one of the vessels. The two segmented surfaces were compared by determining the distance between them at each point along one of the two surfaces. Since the 3D US images had been previously registered, the two segmented surfaces are expected to overlap everywhere except near the region of ulceration. This was confirmed to within a 0.3 mm error.

Technique for evaluation of semiautomatic segmentation methods

In this paper we describe an evaluation technique that quantifies both the accuracy and variability in semiautomatic segmentation algorithms. The particular interest of the study is the evaluation of an active contour method for 2-D carotid artery lumen segmentation in ultrasound images. The active contour method used is known as the Geometrically Deformed Model (GDM). This segmentation method to be evaluated requires a single seed to be placed in the target region by the operator. The evaluation approach is based on the contour probability distribution (CPD), which is obtained by generating contours of the object using a set of possible seed locations. A contour matching procedure provides local displacement measures between any two contours, which in turn allow the calculation of the local CPD of a group of contours. The mean contour can be compared to an operator defined contour to provide accuracy measurements, and the variance can provide measures of local and global variability. The evaluation results from multiple images can be pooled to generate statistics for a more complete evaluation of a semi- automatic segmentation method.

Development and evaluation of a semiautomatic 3D segmentation technique of the carotid arteries from 3D ultrasound images

In this paper, we report on a semi-automatic approach to segmentation of carotid arteries from 3D ultrasound (US) images. Our method uses a deformable model which first is rapidly inflated to approximately find the boundary of the artery, then is further deformed using image-based forces to better localize the boundary. An operator is required to initialize the model by selecting a position in the 3D US image, which is within the carotid vessel. Since the choice of position is user-defined, and therefore arbitrary, there is an inherent variability in the position and shape of the final segmented boundary. We have assessed the performance of our segmentation method by examining the local variability in boundary shape as the initial selected position is varied in a freehand 3D US image of a human carotid bifurcation. Our results indicate that high variability in boundary position occurs in regions where either the segmented boundary is highly curved, or the 3D US image has poorly defined vessel edges.

MP3 compression of doppler ultrasound signals

The effect of lossy, MP3 compression on spectral parameters derived from Doppler ultrasound (US) signals was investigated. Compression was tested on signals acquired from two sources: 1. phase quadrature and 2. stereo audio directional output. A total of 11, 10-s acquisitions of Doppler US signal were collected from each source at three sites in a flow phantom. Doppler signals were digitized at 44.1 kHz and compressed using four grades of MP3 compression (in kilobits per second, kbps; compression ratios in brackets): 1400 kbps (uncompressed), 128 kbps (11:1), 64 kbps (22:1) and 32 kbps (44:1). Doppler spectra were characterized by peak velocity, mean velocity, spectral width, integrated power and ratio of spectral power between negative and positive velocities. The results suggest that MP3 compression on digital Doppler US signals is feasible at 128 kbps, with a resulting 11:1 compression ratio, without compromising clinically relevant information. Higher compression ratios led to significant differences for both signal sources when compared with the uncompressed signals.

Analysis of MP3-compressed Doppler ultrasound quadrature signals

The effect of lossy, MP3 (MPEG-Layer 3) compression on clinically important Doppler parameters - derived from spectral analysis of Doppler ultrasound signals - was investigated. Ten, 10-second acquisitions of gated Doppler ultrasound signal were collected in a phantom perfused with a pulsatile flow waveform. Doppler data were collected using two sample volume lengths - 1.5 mm and 10 mm. The in- phase and quadrature Doppler signals were digitized at 44.1 kHz and compressed using four grades of signal compression (with corresponding compression ratios given in brackets): uncompressed, 128 kbits/s (11:1), 64 kbits/s (44:1). The digital audio signals were identically processed with a Fourier analysis program that provided an estimate of the instantaneous Doppler frequency (velocity) spectrum and derived parameters such as peak velocity, mean velocity, spectral width, total integrated power, and ratio of spectral power from negative and positive velocities. Analysis of variance indicated there were no significant differences (p>0.05) observed in the peak or mean velocities, spectral width, or the power ratio derived from 128 kbits/s and 64 kbits/s audio signals when compared to the uncompressed audio signals (both sample volume lengths) and the 128 kbits/s audio signals (10 mm sample volume length). However, for the 32 kbits/s audio signals, significant differences (p<0.001) were found in all of the studied parameters.

Segmentation of ulcerated plaque: evaluation and optimization of a semiautomatic method for tracking the progression of carotid atherosclerosis

A semi-automatic method for segmenting carotid lumen and plaque from three-dimensional vascular ultrasound (US) images has been developed. We examine its ability to distinguish changes in carotid vessel and plaque surface morphology, such as those caused by plaque ulceration. Two stenosed vessel phantoms were imaged using a 3D US imaging system. The phantoms were identical except for the inclusion of a hemispherical cut in the side of one of the vessels, in order to simulate the development of an ulceration. Ultrasound images of the phantoms were segmented using our algorithm, then the resulting surfaces were registered to one another using a rigid-body iterative closest point (ICP) algorithm. The volume of ulceration was determined by finding the difference between the two segmented surfaces in a region of interest surrounding the ulceration. Since the true volume of the ulceration was known a priori, an optimization strategy was used to tune the deformable model to better segment the ulceration. Analysis of ulceration volume as a function of the deformable models parameters show that 1) large ulcerations are easily identified in our test case, and 2) the model is well behaved with respect to its parameters, suggesting that an automatic strategy for volumetric optimization is feasible.