-
[show abstract]
[hide abstract]
ABSTRACT: After 20 years of innovation in techniques that specifically image the biomechanical properties of tissue, the evolution of elastographic imaging can be viewed from its infancy, through a proliferation of approaches to the problem to incorporation on research and then clinical imaging platforms. Ultimately this activity has culminated in clinical trials and improved care for patients. This remarkable progression represents a leading example of translational research that begins with fundamentals of science and engineering and progresses to needed improvements in diagnostic and monitoring capabilities applied to major categories of disease, surgery and interventional procedures. This review summarizes the fundamental principles, the timeline of developments in major categories of elastographic imaging, and concludes with recent results from clinical trials and forward-looking issues.
Physics in Medicine and Biology 01/2011; 56(1):R1-R29. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, a two-dimensional (2D) quantitative sonoelastographic technique for estimating local shear wave speeds from slowly propagating shear wave interference patterns (termed crawling waves) is presented. Homogeneous tissue- mimicking phantom results demonstrate the ability of quantitative sonoelastographic imaging to accurately reconstruct the true underlying shear wave speed distribution as verified using mechanical measurements. From heterogeneous phantoms containing a 5 or 10 mm stiff inclusion, results indicate that increasing the estimator kernel size increases the transition zone length about boundaries. Contrast-to-noise ratio (CNR) values from quantitative sonoelastograms obtained in heterogeneous phantoms reveal that the 2D quantitative sonoelastographic imaging technique outperforms the one-dimensional (ID) precursor in terms of image noise minimization and contrast enhancement. Experimental results from an embedded porcine liver specimen with an induced radiofrequency ablation (RFA) lesion validate 2D quantitative sonoelastographic imaging in tissue. Overall, 2D quantitative sonoelastography was shown to be a promising new imaging method to characterizing the shear wave speed distribution in elastic materials.
Ultrasonics Symposium, 2007. IEEE; 12/2007
-
B. Castaneda,
K. Hoyt,
M. Zhang,
D. Pasternack,
L. Baxter,
P. Nigwekar,
A. di Sant'Agnese,
J. Joseph,
J. Strang,
D.J. Rubens, K.J. Parker
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we evaluate the performance of sonoelastography for prostate cancer detection. Ultrasound (US) B-mode and sonoelastographic volumes were acquired from five prostate glands ex vivo. Additionally, one more gland was imaged in vivo using a transrectal US probe. Semi-automatic algorithms were used to segment the surface of the gland from the B-mode volume and the tumors from sonoelastographic data. To assess the detection performance, three dimensional (3D) sonoelastographic findings were compared in size and position to 3D histological data. Sonoelastography detected seven out of nine cancers in the ex vivo prostate glands and two out of three malignant masses in the in vivo experiment. Overall, 3D sonoelastography has shown potential for prostate cancer detection albeit based on limited data.
Ultrasonics Symposium, 2007. IEEE; 12/2007
-
[show abstract]
[hide abstract]
ABSTRACT: Radiofrequency ablation (RFA) is a minimally invasive thermal therapy that is under investigation as an alternative to surgery for treating liver tumors. Currently, there is a need to monitor the process of lesion creation to guarantee complete treatment of the diseased tissue. In a previous study, sonoelastography was used to detect and measure RFA lesions during exposed liver experiments in a porcine model in vivo. Manual outlining of these lesions in the sonoelastographic images is challenging due to a lack of boundary definition and artifacts formed by respiratory motion and perfusion. As a result, measuring the lesions becomes a time-consuming process with high variability. This work introduces a semi-automatic segmentation algorithm for sonoelastographic data based on level set methods. This algorithm aims to reduce the variability and processing time involved in manual segmentation while maintaining comparable results. For this purpose, eleven RFA lesions are created in five porcine livers exposed through a midline incision. Three independent observers perform manual and semi-automatic measurements on the in vivo sonoelastographic images. These results are compared to measurements from gross pathology. In addition, we assess the feasibility of performing sonoelastograhic measurements transcutaneously. The procedure previously described is repeated with three more lesions without exposing the liver. Overall, the semi-automatic algorithm outperforms manual segmentation in accuracy, speed, and repeatability. These results suggest that sonoelastography in combination with the segmentation algorithm has the potential to be used as a complementary technique to conventional ultrasound for thermal ablation monitoring and follow-up imaging.
Ultrasonics Symposium, 2007. IEEE; 12/2007
-
[show abstract]
[hide abstract]
ABSTRACT: A quantitative sonoelastographic technique for skeletal muscle tissue characterization is introduced. Experimental data was collected in both ex vivo bovine and in vivo human skeletal muscle tissue. Crawling wave sonoelastographic data was processed using a quantitative technique for estimating local shear wave speed distributions. Results on ex vivo skeletal muscle samples demonstrate shear wave anisotropy and existence of fast and slow shear waves corresponding to propagation parallel and perpendicular to muscle fibers. Comparison of relative frequency-dependent changes between shear wave speed estimates for both shear wave propagation parallel and perpendicular to muscle fibers suggests increased viscoelastic effects for the former. Preliminary sonoelastographic data from two healthy human subjects was acquired in the relaxed rectus femoris muscles. Results demonstrate that quantitative elasticity data can be reproducibly acquired in vivo. Overall, preliminary results are encouraging and quantitative sonoelastography may prove clinically feasible for the in vivo characterization of skeletal muscle in health and disease.
Ultrasonics Symposium, 2007. IEEE; 11/2007
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we introduce and evaluate a novel sonoelastographic technique for imaging shear velocity distributions from propagating shear wave interference patterns (termed crawling waves). A mathematical relationship between local crawling wave spatial phase derivates and shear velocity is presented with phase derivatives estimated using an autocorrelation-based technique. Results from homogeneous phantoms illustrate the ability of sonoelastographic shear velocity imaging to accurately quantify the true shear velocity distribution as verified using time-of-flight measurements. Results from a heterogeneous phantom reveal the ability of sonoelastographic shear velocity imaging to distinguish a stiff circular inclusion with shear velocity contrast comparable to that measured using mechanical testing techniques. High contrast visualization of focal carcinomas in an in-vitro prostrate specimen demonstrates the feasibility of this novel sonoelastographic imaging technique in tissue
Ultrasonics Symposium, 2006. IEEE; 11/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Three methods for extraction and quantitative measurement of features in CT and MR images are examined: hand tracing, semi-automated tracing using the livewire graph search algorithm, and extraction using a geometrically constrained region growth algorithm. Extracted structures are evaluated in terms of volume, cross-sectional area, and major axis in plane. Reproducibility, time required for extraction, and accuracy of each of these metrics is measured for each of the extraction methods using both phantoms and clinical lung tumor data.
Image Processing. 2002. Proceedings. 2002 International Conference on; 02/2002
-
[show abstract]
[hide abstract]
ABSTRACT: This paper describes fusion of three-dimensional (3-D) ultrasound (US) and magnetic resonance imaging (MRI) data sets, without the assistance of external fiducial markers or external position sensors. Fusion of these two modalities combines real-time 3-D ultrasound scans of soft tissue with the larger anatomical framework from MRI. The complementary information available from multiple imaging modalities warrants the development of robust fusion capabilities. We describe the data acquisition, specialized algorithms, and results for 3-D fused data from phantom studies and in vivo studies of the normal human vasculature and musculoskeletal systems.
IEEE Transactions on Medical Imaging 05/2001; 20(4):354-9. · 3.64 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A novel method for the automated 3D extraction and measurement of soft-tissue lesions in CT imagery is presented. The extraction is carried out using a hybrid algorithm that incorporates elements from both competitive region growth and deformable template techniques. This algorithm is tested against manual tracing and is shown to provide significantly improved performance using three performance metrics: speed, precision and accuracy
Computer-Based Medical Systems, 2001. CBMS 2001. Proceedings. 14th IEEE Symposium on; 02/2001
-
[show abstract]
[hide abstract]
ABSTRACT: We present a method for combining histological data with 3D
ultrasound of excised prostate tissue using surface registration. With
this technique, 3D prostate cancer lesions can be properly located and
visualized within a B-scan volume for tissue characterization
comparisons. Three prostate specimens were scanned with a GE Logiq 700
(Expert series) to obtain 2D B-scan sequences. The prostate boundaries
were manually segmented from each sequence and reconstructed into 3D
volumes. Specimens were fixed, sectioned into slabs, then mounted whole
onto slides. Cancerous lesions were outlined by a pathologist. The
slides were photographed with an Optronics Spot digital camera using
Image-Pro Plus software on a PC. The gland surface was manually
segmented from 2D histology images and reconstructed into a volume.
Various reconstruction problems were addressed, such as specimen
shrinkage due to the fixing and staining processes. The fusion algorithm
translates and rotates one surface volume in 3D to find the best surface
overlap. The resulting geometric transform is used to reorient the
original image volume. The displacement error was determined by
measuring the urethra offsets in final volume cross-sections. The offset
distance ranged from 1.25 mm to 3.45 mm with an average of 2.36 mm.
Another measure to gauge volume alignment is to calculate the ratio of
overlapping voxels to total combined voxels(intersection/union). For a
perfect case, this ratio will be 1. The ratios for the three cases
ranged in value from 0.774 to 0.845
Ultrasonics Symposium, 2001 IEEE; 02/2001
-
[show abstract]
[hide abstract]
ABSTRACT: The fusion of multiple ultrasound scans, taken from different
orientations and different times, can enhance interventional procedures
and can quantify response to therapy. However, fusion of ultrasound
scans requires precision alignment of recognizable fiducial markers. We
present the results of volume registration of liver tissue and phantom
material to test the accuracy of a semi-automated registration
algorithm. Both linear tracking and magnetic “Flock of
Birds” (FOE) position sensing are employed to register sequential
frames of 2D ultrasound into 3D space. Major vessels and/or
interventional probes are utilized as internal fiducial markers for a
correlation registration algorithm. US scans were obtained from a
volunteer with a known hemangioma, a benign, stable, echogenic liver
tumor. Complementary, non-parallel scans were then fused to determine
hemangioma co-registration displacement error. The average central
displacement error was 4.1 mm. To further test our algorithm, we
obtained perpendicular scans of a radio-frequency (RF) ablation probe
within a tissue phantom. The volumes were registered and fused
successfully, with an average displacement error of 1.1 mm. In
conclusion, internal structures such as vessels and interventional
probes provide adequate markers for volume fusion, producing
misregistration errors less than 7.4 mm
Ultrasonics Symposium, 2000 IEEE; 11/2000
-
[show abstract]
[hide abstract]
ABSTRACT: A technique for reducing vibration artifacts encountered when
using pure tone vibration in vibration amplitude sonoelastography is
presented. Multi-tone signals were designed to synchronize with the
pulse repetition frequency and packet size of the scanner. When applied
to phantoms with known uniform elasticity it was found that noise levels
using the multi-tone signals were 28 to 35% less than when pure tone
vibration was used. Images of a hard phantom lesion made using
multi-tone vibration compared favorably with those using pure tone
vibration
Ultrasonics Symposium, 2000 IEEE; 11/2000
-
[show abstract]
[hide abstract]
ABSTRACT: Sonoelastography is an ultrasound imaging technique where low amplitude, low-frequency shear waves (less than 0.1 mm displacement and less than 1 kHz frequency) are propagated through internal organs, while real-time Doppler techniques are used to image the resulting vibration pattern. When a discrete hard inhomogeneity, such as a tumour, is present within a region of soft tissue, a decrease in the vibration amplitude will occur at its location. This forms the basis for tumour detection using sonoelastography. For three-dimensional (3D) imaging the acquisition of sequential tomographic slices using this technique, combined with image segmentation, enables the reconstruction, quantification and visualization of tumour volumes. Sonoelastography and magnetic resonance images (MRI) of a tissue phantom containing a hard isoechoic inclusion are compared to evaluate the accuracy of this method. The tumour delineation from sonoelastography was found to have good agreement with the tumour from MRI except for a bleeding at one of its ends. Although sonoelastography is still in an experimental phase, the principles behind this imaging modality are explained and some practical aspects of acquiring sonoelastography images are described. Results from a 3D sonoelastography reconstruction of a tissue mimicking phantom and an ex vivo whole prostate specimen are presented.
Physics in Medicine and Biology 07/2000; 45(6):1477-94. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A novel iterative approach is presented to estimate Young's modulus in homogeneous soft tissues using vibration sonoelastography. A low-frequency (below 100 Hz) external vibration is applied and three or more consecutive frames of B-scan image data are recorded. The internal vibrational motion of the soft tissue structures is calculated from 2D displacements between pairs of consecutive frames, which are estimated using a mesh-based speckle tracking method. An iterative forward finite element approach has been developed to reconstruct Young's modulus from the measured vibrational motion. This is accomplished by subdividing the 2D image domain into sample blocks in which Young's modulus is assumed to be constant. Because the finite element equations are internally consistent, boundary values other than displacement are not required. The sensitivity of the results to Poisson's ratio and the damping coefficient (viscosity) is investigated. The approach is verified using simulated displacement data and using data from tissue-mimicking phantoms.
Physics in Medicine and Biology 07/2000; 45(6):1495-509. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Fully developed speckle patterns observed in coherent imagery are characterized by a Rayleigh-distributed envelope amplitude. Non-Rayleigh distributions are observed in many cases, such as when the number of scatterers in a resolution cell is small or scatterers are organized with some periodicity. Distributions resulting from the assumption of random scatterer phase (random walk models) have been used to describe the speckle amplitude in these cases, leading to K, Rician, and homodyned-K amplitude distributions. An alternative is to incorporate nonrandom phase implicitly by adopting models that directly describe the spatial placement of point scatterers. We examine the consequences of assuming that scattering is described in one dimension by a stationary renewal process in which the arrival times are the locations of ideal point scatterers, the interscatterer distances are drawn from a gamma distribution, and the scatterer amplitudes are allowed to be correlated in space. This model has been called the marked regularity model because variations of the model parameters can generate spatial distributions ranging from clustered to random to nearly periodic. We will demonstrate that all of the non-Rayleigh distributions generated by the previous random phase models can also be generated by the marked regularity model, and we show under what conditions the different distributions will result. We also demonstrate that the regularity model is inherently capable of describing certain sparse scattering conditions. Therefore, the model can represent many cases and provide an intuitively pleasing description of the spatial placement of the scatterers.
IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control 02/1999; 46(4):867-74. · 1.69 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors have derived a “frameless fusion”
technique to integrate volumetric images using internal vasculature as
the fiducial markers. They demonstrate the feasibility of ultrasound (US
to US) liver image fusion using internal vascular landmarks from in-vivo
data obtained on separate scans taken minutes apart or weeks apart. The
US volumes include scans from varying orientations on a single volunteer
with a 5-8 MHz curvilinear transducer (GE Logiq 700 MR) attached to a
motorized, hand-held track. Power Doppler US volumes were segmented to
extract and reconstruct the portal vein structure as 1 mm cubed voxels.
The vessel volumes were then rendered with commercial 3D software and
correlated as a rigid body by a semi-automated, in-house program. After
the vessel volumes were fused, the corresponding grayscale volumes were
re-positioned and fused to create overlapping images. Volume
registration was assessed in the grayscale data by measuring alignment
of diaphragm and vessel walls (portal and hepatic vein) on 6 slices at
25-75% through the volumes. The fused data from the same day and after
six weeks displayed an average 2 mm error centrally and 3-4 mm at the
far field. Potential uses for merged or fused 3D images of the liver
include the reconstruction of the whole organ for treatment planning,
and lesion follow-up
Ultrasonics Symposium, 1999. Proceedings. 1999 IEEE; 02/1999
-
[show abstract]
[hide abstract]
ABSTRACT: With blue noise halftoning of color images, the issues of Moire
patterns in conventional ordered dither techniques are replaced by the
concerns of blue noise quality of overlaid blue noise patterns. We need
to produce a high quality halftone image for many combinations of color
planes, as well as for any single color plane. In this paper, we studied
the properties and power spectrum characteristics of combined blue noise
patterns. An approach is proposed which is based on digital filter
techniques and can be used to generate a set of jointly-blue noise
masks. The masks are individually blue noise and also produce blue noise
patterns when the individual patterns are combined in the usual sense of
color overlay
Image Processing, 1999. ICIP 99. Proceedings. 1999 International Conference on; 02/1999
-
[show abstract]
[hide abstract]
ABSTRACT: By exploiting the correlation of ultrasound speckle patterns that result from scattering by underlying tissue elements, two-dimensional tissue motion theoretically can be recovered by tracking the apparent movement of the associated speckle patterns. Speckle tracking, however, is an ill-posed inverse problem because of temporal decorrelation of the speckle patterns and the inherent low signal-to-noise ratio of medical ultrasonic images. This paper investigates the use of an adaptive deformable mesh for nonrigid tissue motion recovery from ultrasound images. The nodes connecting the mesh elements are allocated adaptively to stable speckle patterns that are less susceptible to temporal decorrelation. We use the approach of finite element analysis in manipulating the irregular mesh elements. A novel deformable block matching algorithm, making use of a Lagrange element for higher-order description of local motion, is proposed to estimate a nonrigid motion vector at each node. In order to ensure that the motion estimates are admissible to a physically plausible solution, the nodal displacements are regularized by minimizing the strain energy associated with the mesh deformations. Experiments based on ultrasound images of a tissue-mimicking phantom and a muscle undergoing contraction, and on computer simulations, have shown that the proposed algorithm can successfully track nonrigid displacement fields.
IEEE Transactions on Medical Imaging 01/1999; 17(6):945-56. · 3.64 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Vibration sonoelastography has been developed for the detection of hard lesions in relatively soft tissue. The basic concept is to propagate low-amplitude and low-frequency shear waves (with displacements below 0.1 mm and frequencies typically below 1000 Hz) through deep organs, and displaying the vibration response in real-time using advanced color Doppler imaging techniques. A hard inhomogeneity, such as a tumor, will produce a localized disturbance in the vibration pattern, forming the basis for detection even when the tumor is isoechoic on B-scan images. This paper focuses on the important quantitative issues concerning the detectability or inherent contrast of lesions. The specific factors of lesion size, relative stiffness and vibration frequency are studied using theoretical models, finite element methods and experimental measurements on tissue-mimicking materials. The results indicate that detectability increases with vibration (shear wave) frequency; however, loss mechanisms ultimately limit the penetration of higher vibration frequencies (in the kHz range).
Ultrasound in Medicine & Biology 12/1998; 24(9):1437-47. · 2.29 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The 3D reconstruction of high contrast anatomical structures has been widely explored. However, a class of important clinical problems involves the motion of very complex musculoskeletal structures including the joints, hence a 4D reconstruction is desired. Practical difficulties with 4D reconstruction with MRI include the time required for data acquisition, the resolution required for visualization of small but critical structures, the gross inhomogeneities of field coil response, the degree of noise present with the signal and the extreme low-contrast details between some distinct anatomical structures. The authors present a comprehensive approach to 4D musculoskeletal imagery that address the above challenges. Specific MRI imaging protocols; segmentation, motion estimation and motion tracking algorithms are developed and applied to render complex 4D musculoskeletal systems. Applications of the approach include the analysis of the rotation of the upper arm and the knee extension
Biomedical Image Analysis, 1998. Proceedings. Workshop on; 07/1998
