ABSTRACT
Ultrasound is the best way to diagnose thyroid nodules. To discriminate benign and malignant nodules, calcification is an important characteristic. However, calcification in ultrasonic images cannot be extracted accurately because of capsule wall and other internal tissue. In this paper, deep learning was first proposed to extract calcification, and two improved methods were proposed on the basis of Alexnet convolutional neural network. First, adding the corresponding anti-pooling (unpooling) and deconvolution layers (deconv2D) made the network to be trained for the required features and finally extract the calcification feature. Second, modifying the number of convolution templates and full connection layer nodes made feature extraction more refined. The final network was the combination of two improved methods above. To verify the method presented in this article, we got 8 416 images with calcification, and 10 844 without calcification. The result showed that the accuracy of the calcification extraction was 86% by using the improved Alexnet convolutional neural network. Compared with traditional methods, it has been improved greatly, which provides effective means for the identification of benign and malignant thyroid nodules.
ABSTRACT
Objective To compare the kinematics of the knee joint of normal Chinese people and people after fixed-bearing posterior stabilized total knee arthroplasty while doing weight-bearing deep knee-bending using fluoroscopy analysis.Methods From June to December 2010,ten volunteers and ten patients with fixed-bearing posterior stabilized total knee arthroplasty (TKA) were required to perform weight-bearing deep knee-bending activity under surveillance of roentgenization,motion between femur and tibia was analyzed with interval of 15° according to two-to three-dimensional (2D-3D) registration technique.Results During weight-bearing deep knee-bending,the average weight-bearing maximal flexion was 136° in the normal group,which was significantly higher than the 125° in the TKA group.All 10 normal knees present posterior translation of femoral condyle during deep knee-bending,posterior translation of the lateral condyle was greater than medial condyle,thereby creating a medial pivot type of axial rotational pattern in which the tibia internally rotates relative to the femur as flexion increased.The average amount of posterior femoral translation of the medial condyle was 7.3±1.2 mm,whereas the lateral condyle translated posteriorly 19.3±3.1 mm.All knees experienced tibia rotated internally during progressive flexion,and the average amount of axial rotation for the ten subjects from 0° to 135° was 23.8±3.4°.From extension to maximal flexion,the average amount of posterior translation of medial condyle was 1.4±0.6 rmm,whereas the lateral condyle translated 6.4±1.7 mm in the posterior direction in the fixed-bearing posterior-stabilized TKA.The average amount of tibial internal rotation was 8.5°±3.4°.Conclusion In normal Chinese people,during knee flexion activities the lateral condyle experiences significantly more amount of posterior translation than the medial condyle,leading to the tibia present medial pivot internal rotation relative to the femur,and the PFC-Sigrna fixed-bearing posterior stabilized designs has the similar kinematics with normal knee,but the posterior condylar translation and tibial axial rotation are much less than the normal.
ABSTRACT
To solve the problem of imprecise positioning of feature point and of the feature data redundancy in facial expression recognition by active appearance models (AAM), the automatic adjustment of initial model for AAM fitting is proposed in this paper. The specific aims are to improve the precision of positioning and to more effectively reflect the variation of expressions by acquired features. The problem of feature selection is resolved by adopting quadratic mutual information and reducing the feature dimension. The support vector machine (SVM) classifier is used for expression recognition. The experimental results on CAS-PEAL facial expression database show that the proposed method effectively improves the performance of facial expression recognition, the maximum recognition rate being 83.33%.
Subject(s)
Humans , Algorithms , Computer Simulation , Facial Expression , Image Interpretation, Computer-Assisted , Methods , Models, Biological , Pattern Recognition, Automated , Methods , Signal Processing, Computer-AssistedABSTRACT
A method based on a symmetric region growing algorithm is presented for the segmentation of ultrasonic medical image. The method divides into three steps. First, according to the characteristic of the ultrasonic medical images, an adaptive weighted median filter is used to suppress speckle noise. Then, scan the digital image from the first row and grow regions from each scanned point by applying the growth criteria and combination criteria until all image pixels have been scanned. Examine the resulting regions using the seed criteria. If any point of a region satisfies the criteria for the region of interest region, assign the region to the resulting segmented image. The effectiveness of this method and a group of growth criteria as well as combination criteria applicable to ultrasonic medical image have been obtained by cardiac ultrasound image segmentation experiments. The experiment result shows that this method is good in the performance of the segmention of cardiac ultrasonic medical image.
Subject(s)
Humans , Algorithms , Echocardiography , Methods , Image Enhancement , Methods , Image Processing, Computer-AssistedABSTRACT
Speckle is the main reason which declines the quality of medical ultrasonic images. In this paper, the initial condition for the Downhill filter, a morphological reconstruction algorithm, is modified and applied in the speckle reduction. Firstly, the initial area and start position as the mark image was determined in the marker image. Then the modified Downhill filter was used in the ultrasonic marker image. The results showed that, in comparison with the other three filters, this modified Downhill filter, while maintaining the integrity of the contour,could reduce the speckle in the regions of cavity efficiently and rapidly.
Subject(s)
Humans , Algorithms , Image Enhancement , Methods , Image Processing, Computer-Assisted , Ultrasonography , MethodsABSTRACT
Lower contrast and speckle noise are the main reasons which decline the quality of medical ultrasonic images. In this paper, a new method is proposed to filter the speckle noises and enhance the contrast simultaneously. Anisotropic diffusion filtering method was firstly applied to filter images. Then the loss of information, which the contrast function of contrast enhancement model lies on, was obtained. Finally, the contrast can be enhanced by using enhancement model. Experimental results show that the proposed method not only removes the speckle noises effectively, but also enhances the contrast obviously. This method supplies an effective approach for improving the quality of medical ultrasonic images.
Subject(s)
Humans , Algorithms , Anisotropy , Image Enhancement , Image Processing, Computer-Assisted , Methods , Signal Processing, Computer-Assisted , Ultrasonography , MethodsABSTRACT
Ultrasound medical imaging has been widely applied in clinic diagnoses because of its real-time, non-invasiveness and convenience. However, it suffers from severe speckle noises. Until now, a lot of filtering algorithms have been proposed, but none of them is satisfactory. In this paper, four nonlinear filter methods, based on median filter, wavelet decomposition and anisotropic diffusion, are systematically reviewed. Finally, the prospect of the nonlinear filtering method is discussed.
Subject(s)
Humans , Algorithms , Image Processing, Computer-Assisted , Methods , Nonlinear Dynamics , UltrasonographyABSTRACT
This study aims to provide a computer-aided method for the diagnosis of fatty liver by B-scan ultrasonic imaging. Fatty liver is referred to the infiltration of triglycerides and other fats of the liver cells, which affected the texture of liver tissue. In this paper, some features including mean intensity ratio, as well as angular second moment, entropy and inverse differential moment of gray level co-occurrence matrix were extracted from B-scan ultrasonic liver images. Feature vectors which indicated two classes of images were created with the four features. Then we used kappa-means clustering algorithm, self-organized feature mapping (SOFM) artificial neural network and back-propagation (BP) artificial neural network to classify these vectors. The accuracy rate of kappa-means clustering algorithm was 100% for normal liver and 63.6% for fatty liver. The results of SOFM neural network showed that the accuracy rate was 84.8% for normal liver and 90.9% for fatty liver. The accuracy rate of neural network was 100% both for normal liver and fatty liver. This technology could detect the characteristics of B-scan images of normal liver and fatty liver more accurately. It could greatly improve the accuracy of the diagnosis of fatty liver.
Subject(s)
Humans , Diagnosis, Differential , Fatty Liver , Diagnostic Imaging , Image Processing, Computer-Assisted , Methods , Lung , Diagnostic Imaging , Neural Networks, Computer , Sensitivity and Specificity , UltrasonographyABSTRACT
Cell culture is one of the usual methods for studying living cell and tissue, the method presented in this paper is based on image processing and analyzing technology for activity estimation of mesenchymal stem cells (MSCs). The existing activity estimation methods are costly, complex and invasive. In this method, thresholding is used to preprocess image and to separate out the growth hallow. Then the area is calculated by counting the pixels of the growth hallow. The changes of the activity estimated by this method are similar to those by corresponding cellular experiments. Compared with the existing methods in biology, medicine or medical cellular science, this method is easier, faster, cost-effective and non-invasive. The proposed method has been proved to be efficient by primary experiments of MSCs.
Subject(s)
Animals , Rats , Cells, Cultured , Femur , Cell Biology , Image Processing, Computer-Assisted , Methods , Mesenchymal Stem Cells , Cell Biology , Rats, Sprague-DawleyABSTRACT
The purpose of this article is to evaluate the role of quantitative margin features in the computer-aided diagnosis of malignant and benign solid breast masses using sonographic imaging. The tumour was seperated by the expert. Three contour features circurity (C), area ratio (A) and length width ratio (LWR) was caculated from the tumour contour. Then back-propagation (BP) neural network with contour features was used to classify tumors into benign and malignant. Results from 119 ultrasonic images have been applied in this experiment. BP neural network yielded the following results: 89.7% and 73.5% respectively. The methods applied in this paper are helpful to raise the correctance of breast cancer diagnosis.
Subject(s)
Female , Humans , Breast Neoplasms , Diagnostic Imaging , Pathology , Image Processing, Computer-Assisted , Methods , Neural Networks, Computer , UltrasonographyABSTRACT
Objective To improve the key steps of direct volume rendering(DVR) method to obtain ideal semitransparent 3D visualization of medical images from 2D slices.Method An arbitrary data set was classified into a simple one and a complex one according to the boundaries existing within the data set,thereby a piecewise linear function was used to construct the transfer function(TF).Low-grayed voxels were regarded as a kind of material which did not radiate but do induce attenuation of sampling lights.The TF was adjusted according to the opacity distribution when one resampling process ends,and the rendering procedure was completed.Result Interior structures of tissues in different levels were rendered.Conclusion Multi-level visualization of 3D medical images for the semitransparent effect can be obtained by considering all the tissues in the data set as semitransparent materials.
ABSTRACT
Mitral regurgitation is one of the most serious heart diseases. With the development of up-to-date medical techniques, the ratio of successful operations in valvular repair and valvular replacement has been largely improved. Examinations before operation become extremely crucial. Accurate method is required in assessing the degree of mitral regurgitation to set down the corresponding treatment method. This paper reviews the evaluation methods of mitral valvular regurgitation provided in these years and presents comments on the application areas as well as the merits and disadvantages of those methods. Finally, the prospect of the method based on three-dimensional Doppler ultrasonographic imaging on mitral regurgitation is discussed.
Subject(s)
Humans , Echocardiography, Doppler, Color , Echocardiography, Three-Dimensional , Image Processing, Computer-Assisted , Mitral Valve Insufficiency , Diagnostic ImagingABSTRACT
The ultrasonic estimation of fetal weight at delivery is of important prognostic significance in obstetrical practice. The convertional regression formulas used for estimating fetal weight have the disadvantage of less reliability. In this study, we used the back propagation neural network (BP) to estimate Fetal Weight. Some input variables were adopted in constructing the BP model: biparietal diameter (BPD), cerebellum transverse diameter (TCD), abdominal circumference (AC), liver length (LL), femur length (FL), fetal thigh soft tissue thickness (FSTT), and gestational age (GA). The fetal weights of 109 singleton fetuses were estimated. In the training group and validation group, coincidence rates were 89.77% and 76.19% respectively. The results show that the estimation based on neural network is more accurate than that by regression method. GA, its unit is not week but day in our formulas, is very valuable in combination with other ultrasonic parameters on estimation.
Subject(s)
Female , Humans , Infant, Newborn , Pregnancy , Anthropometry , Methods , Birth Weight , Fetal Weight , Gestational Age , Neural Networks, Computer , Regression Analysis , Term BirthABSTRACT
Objective To improve the precision of the traditional segmentation of echocardiogram, by suppressing the influence from inherent speckle noises in medical ultrasonic images. Method An automatic segmentation method based on reconstructed morphology was proposed in this paper. First, the opening and closing operations by reconstruction were imposed to the ultrasonic image. Second, the top-hat operation was used to extract the bright and/or dark features and to find out the boundaries corresponding to these features, whereby implemented the automatic segmentation. Result The segmented echocardiogram had less artificial boundaries resulted from speckle noise, and could accurately be extracted the artery and ventricle. Conclusion The presented method can detect both dark and bright objects accurately, and the boundary has a fine continuity. In addition, the algorithm is also applicable to the extraction of sole bright/dark features, accordingly to reduce the complexity and time needed and to improve the accuracy.
ABSTRACT
The object of this study was to work at accurate axis registration and interpolation methods for multi-dimension reconstruction of rotary scanning ultrasonic medical images. At first, time-field curves of the images' axes were analyzed according to their characteristic points and the axial direction registration was realized. Similar matrix was used to find registration pixels line near the axes of two images. Auto-correlation function and Fourier spectrum were used to evaluate the effects of axes registration. Second, an interpolation method was studied for the special space distribution of rotary scanning images. Results of experiments indicate that the axes registration and interpolation methods were suitable to rotary scanning medical images. The quality of reconstruction can be greatly improved by registration-based interpolation methods.
Subject(s)
Humans , Algorithms , Echocardiography , Methods , Image Interpretation, Computer-Assisted , Methods , Image Processing, Computer-Assisted , MethodsABSTRACT
Shear-warp is a volume rendering technology based on object-order. It has the characteristics of high speed and high image quality by comparison with the conventional visualization technology. The authors introduced the principle of this algorithm and applied it to the visualization of 3-D data obtained by interpolating rotary scanning echocardiogram. The 3-D reconstruction of the echocardiogram was efficiently completed with high image quality. This algorithm has a prospective application in medical image visualization.
Subject(s)
Adult , Female , Humans , Algorithms , Computer Simulation , Echocardiography , Image Interpretation, Computer-Assisted , Methods , Image Processing, Computer-Assisted , Methods , Imaging, Three-Dimensional , Methods , Phantoms, ImagingABSTRACT
Based on the application of box-counting fractal model to the texture analysis of ultrasonic image after intravenous injection of Levovist acoustics contrast agent in a dog, this paper presents a method to calculate the fractal dimension(D) and assistant characteristic (C(L)) of the ultrasonic images of left ventricle. It was found that the D and C(L) changed regularly in the continued cardiac cycles, that is, the maximum value of D appears during diastasis, the minimum value appears during end-systole; as the ventricular systole begins, D changes from maximum to smallness, and as the diastole begins, D changes from smallness to maximum; the maximum value of C(L) appears around end-systole, the minimum value appears around diastasis; from the start of left ventricular systole to end-systole, C(l) shows a tendency to change from smallness to maximum, and from the start of diastole to diastasis, it shows a tendency to change from maximum to smallness. The changes of D and C(L) are very similar to the changes of left ventricular pressure. It is thus evident that there is correlation between the fractal texture characteristics of the contrast agent ultrasonic images and the changes of intracardiac pressure. The results demonstrate that it is possible to measure the intracardiac pressure noninvasively.