Jinhua Shao

A snake-based method for segmentation of intravascular ultrasound images and its in vivo validation

Image segmentation for detection of vessel walls is necessary for quantitative assessment of vessel diseases by intravascular ultrasound. A new segmentation method based on gradient vector flow (GVF) snake model is proposed in this paper. The main characteristics of the proposed method include two aspects: one is that nonlinear filtering is performed on GVF field to reduce the critical points, change the morphological structure of the parallel curves and extend the capture range; the other is that balloon snake is combined with the model. Thus, the improved GVF and balloon snake can be automatically initialized and overcome the problem caused by local energy minima. Results of 20 in vivo cases validated the accuracy and stability of the segmentation method for intravascular ultrasound images.

A 2D strain estimator with numerical optimization method for soft-tissue elastography.

Elastography is a bioelasticity-based imaging modality which has been proved to be a potential evaluation tool to detect the tissue abnormalities. Conventional method for elastography is to estimate the displacement based on cross-correlation technique firstly, then strain profile is calculated as the gradient of the displacement. The main problem of this method arises from the fact that the cross-correlation between pre- and post-compression signals will be decreased because of the signals compression-to-deformation. It may constrain the estimation of the displacement. Numerical optimization, as an efficient tool to estimate the non-rigid deformation in image registration, has its potential to achieve the elastogram. This paper incorporates the idea of image registration into elastography and proposes a radio frequency (RF) signal registration strain estimator based on the minimization of a cost function using numerical optimization method with Powell algorithm (NOMPA). To evaluate the proposed scheme, the simulation data with a hard inclusion embedded in the homogeneous background is produced for analysis. NOMPA can obtain the displacement profiles and strain profiles simultaneously. When compared with the cross-correlation based method, NOMPA presents better signal-to-noise ratio (SNR, 32.6+/-1.5 dB vs. 23.8+/-1.1 dB) and contrast-to-noise ratio (CNR, 28.8+/-1.8 dB vs. 21.7+/-0.9 dB) in axial normal strain estimation. The in vitro experiment of porcine liver with ethanol-induced lesion is also studied. The statistic results of SNR and CNR indicate that strain profiles by NOMPA performs better anti-noise and target detectability than that by cross-correlation based method. Though NOMPA carry a heavier computational burden than cross-correlation based method, it may be an useful method to obtain 2D strains in elastography.

Ultrasound elastography of ethanol-induced hepatic lesions: in vitro study.

OBJECTIVE To study the value of ultrasound elastography in evaluation of ethanol-induced lesions of liver. METHODS Alcohol with a dose of 2 ml was injected into a fresh porcine liver under ultrasound guidance to create stiff necrosis. Then freehand elastography of the lesion from the identical scan plane was obtained with SONOLINE Antares system using VF10-5 probe at about every 30 seconds till 6 minutes later. The original high quality radiofrequency data were acquired through an ultrasound research interface which was provided by the ultrasound system. Then, corresponding elastograms were produced offline using cross-correlation technique and compared with gross pathology findings. RESULTS Gray-scale sonogram showed a hyperechoic area with acoustic shadow below appeared immediately after alcohol injection. The hyperechoic area tended to be diffuse and its boundary to be illegible with time. On the contrary, the ethanol-induced lesion in elastogram appeared as a low strain hard region surrounded by high strain soft hepatic tissues, with clear but irregular boundaries. Sequential elastograms with the sketched lesion boundaries showed that the lesion area increased in the first 3 minutes after ethanol injection, and then reached a plateau which corresponding to gross specimen. CONCLUSION Ultrasound elastography is capable of detecting and evaluating the diffusion of ethanol-induced hepatic lesion, and more sensitive and accurate than routine sonography.

Elastographic Evaluation of the Temporal Formation of Ethanol-Induced Hepatic Lesions Preliminary In Vitro Results

The purpose of this study was to evaluate the temporal formation of ethanol‐induced hepatic lesions using ultrasound elastography.

High frame rate and high line density ultrasound imaging for local pulse wave velocity estimation using motion matching: A feasibility study on vessel phantoms

Pulse wave imaging (PWI) is an ultrasound-based method to visualize the propagation of pulse wave and to quantitatively estimate regional pulse wave velocity (PWV) of the arteries within the imaging field of view (FOV). To guarantee the reliability of PWV measurement, high frame rate imaging is required, which can be achieved by reducing the line density of ultrasound imaging or transmitting plane wave at the expense of spatial resolution and/or signal-to-noise ratio (SNR). In this study, a composite, full-view imaging method using motion matching was proposed with both high temporal and spatial resolution. Ultrasound radiofrequency (RF) data of 4 sub-sectors, each with 34 beams, including a common beam, were acquired successively to achieve a frame rate of ∼507 Hz at an imaging depth of 35 mm. The acceleration profiles of the vessel wall estimated from the common beam were used to reconstruct the full-view (38-mm width, 128-beam) image sequence. The feasibility of mapping local PWV variation along the artery using PWI technique was preliminarily validated on both homogeneous and inhomogeneous polyvinyl alcohol (PVA) cryogel vessel phantoms. Regional PWVs for the three homogeneous phantoms measured by the proposed method were in accordance with the sparse imaging method (38-mm width, 32-beam) and plane wave imaging method. Local PWV was estimated using the above-mentioned three methods on 3 inhomogeneous phantoms, and good agreement was obtained in both the softer (1.91±0.24 m/s, 1.97±0.27 m/s and 1.78±0.28 m/s) and the stiffer region (4.17±0.46 m/s, 3.99±0.53 m/s and 4.27±0.49 m/s) of the phantoms. In addition to the improved spatial resolution, higher precision of local PWV estimation in low SNR circumstances was also obtained by the proposed method as compared with the sparse imaging method. The proposed method might be helpful in disease detections through mapping the local PWV of the vascular wall.

Evaluating the Significance of Viscoelasticity in Diagnosing Early-Stage Liver Fibrosis with Transient Elastography

Transient elastography quantifies the propagation of a mechanically generated shear wave within a soft tissue, which can be used to characterize the elasticity and viscosity parameters of the tissue. The aim of our study was to combine numerical simulation and clinical assessment to define a viscoelastic index of liver tissue to improve the quality of early diagnosis of liver fibrosis. This is clinically relevant, as early fibrosis is reversible. We developed an idealized two-dimensional axisymmetric finite element model of the liver to evaluate the effects of different viscoelastic values on the propagation characteristics of the shear wave. The diagnostic value of the identified viscoelastic index was verified against the clinical data of 99 patients who had undergone biopsy and routine blood tests for staging of liver disease resulting from chronic hepatitis B infection. Liver stiffness measurement (LSM) and the shear wave attenuation fitting coefficient (AFC) were calculated from the ultrasound data obtained by performing transient elastography. Receiver operating curve analysis was used to evaluate the reliability and diagnostic accuracy of LSM and AFC. Compared to LSM, the AFC provided a higher diagnostic accuracy to differentiate early stages of liver fibrosis, namely F1 and F2 stages, with an overall specificity of 81.48%, sensitivity of 83.33% and diagnostic accuracy of 81.82%. AFC was influenced by the level of LSM, ALT. However, there are no correlation between AFC and Age, BMI, TBIL or DBIL. Quantification of the viscoelasticity of liver tissue provides reliable measurement to identify and differentiate early stages of liver fibrosis.

Subtraction elastography for the evaluation of ablation-induced lesions: a feasibility study

Different noninvasive or minimally invasive therapeutic ablation procedures can produce tissue necrosis associated with local-stiffness increase. Although elastography has been proved as a potential evaluation tool for many kinds of ablation-induced lesions, the application of subtraction technique in elastography to enhance the visualization of the ablation lesions has rarely been reported. In this paper, subtraction elastography is proposed to evaluate the ablation-induced lesions. Three models are constructed to simulate different kinds of ablated inclusions. The simulation results showed that subtraction elastography is superior to conventional elastography in detecting the ablation-induced lesions with higher signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). The artifacts induced by elastographic signal processing algorithms can be largely reduced in subtraction elastography. In addition, subtraction elastography is less influenced by the stiff background and can provide more reliable boundary information about the lesion than conventional elastography. Furthermore, the feasibility of subtraction elastography is validated by an in vitro experiment of ethanol-induced hepatic lesions. The preliminary results of this work suggest that subtraction elastography may be a good option for the evaluation of ablationinduced lesions.

New Noninvasive Assessment of Liver Fibrosis in Chronic Hepatitis B Maximal Accumulative Respiration Strain

Objective. A novel parameter acquired from conventional B‐mode sonographic videos was introduced in this study, and its diagnostic accuracy for evaluation of hepatic fibrosis was investigated. Methods. Twenty‐eight patients with chronic hepatitis B and 8 patients with hepatic cysts in the right lobe (controls) were enrolled. B‐mode sonographic videos of hepatic motion under the ensisternum in the sagittal plane were captured during peaceful breathing. Maximal accumulative respiration strain (MARS) values of hepatic tissue were obtained after image analysis. METAVIR scoring after liver biopsy was considered the standard. First, the relationship between MARS and the fibrotic stage was studied; and second, receiver operating characteristic (ROC) curves were used to assess the accuracy of MARS for evaluation of the fibrotic stage. Results. When the transducer was placed in the sagittal imaging plane under the ensisternum during the whole respiratory period, the hepatic tissue motion was almost in the same plane. The MARS values (mean ± SD) were 29.44% ± 10.44% in the F0 group (no fibrosis; n = 8), 19.30% ± 9.10% in the F1 group (portal fibrosis without septa; n = 8), 18.09% ± 7.36% in the F2–F3 group (portal fibrosis with few septa or numerous septa without cirrhosis; n = 12), and 14.16% ± 4.18% in the F4 group (cirrhosis; n = 8). The Spearman correlation coefficient between MARS and the fibrotic stage was 0.516 (P = .001). The diagnostic accuracy rates, expressed as areas under the ROC curves, were 0.87 for mild fibrosis (F ≥ 1), 0.72 for substantial fibrosis (F ≥ 2), and 0.75 for cirrhosis (F = 4). Conclusions. Maximal accumulative respiration strain attained from B‐mode sonographic videos of hepatic tissue is a new, convenient, economical, and promising noninvasive parameter for assessment of hepatic fibrosis in patients with chronic hepatitis B.

A regularization-free Young's modulus reconstruction algorithm for ultrasound elasticity imaging

Ultrasound elasticity imaging aims to reconstruct the distribution of elastic modulus (e.g., Youngs modulus) within biological tissues, since the value of elastic modulus is often related to pathological changes. Currently, most elasticity imaging algorithms face a challenge of choosing the value of the regularization constant. We propose a more applicable algorithm without the need of any regularization. This algorithm is not only simple to use, but has a relatively high accuracy. Our method comprises of a nonrigid registration technique and tissue incompressibility assumption to estimate the two-dimensional (2D) displacement field, and finite element method (FEM) to reconstruct the Youngs modulus distribution. Simulation and phantom experiments are performed to evaluate the algorithm. Simulation and phantom results showed that the proposed algorithm can reconstruct the Youngs modulus with an accuracy of 63~85%.

Performance comparison of rigid and affine models for motion estimation using ultrasound RF signals: Simulations and phantom experiments

Rigid model-based and non-rigid model-based methods are two main groups of space-domain methods of tissue motion estimation. Affine model is one of the commonly used non-rigid models. The performances of the rigid model and affine model have not been compared on ultrasound radio-frequency (RF) signals. In this study, three methods, i.e., the normalized cross-correlation method with rigid model (NCC), the optical flow method with rigid model (OFRM) and optical flow method with affine model (OFAM), are compared using ultrasound RF signals, rather than B-mode images used in previous studies. Simulations and phantom experiments are conducted to make the comparison. The results show that the affine model-based method (i.e., OFAM) obtains the lowest root-mean-square error (RAISE) or registration error among all the methods. The affine model is demonstrated to be superior to the rigid model in motion estimation based on RF signals.