Search PubMed⌕ Search

Biomedical subjects

Yung-Nien Sun

Publications and source records attributed to Yung-Nien Sun.

5 recordsLinked to original sources
PubMed2006

Segmentation of kidney from ultrasound B-mode images with texture-based classification.

The segmentation of anatomical structures from sonograms can help physicians evaluate organ morphology and realize quantitative measurement. It is an important but difficult issue in medical image analysis. In this paper, we propose a new method based on Laws' microtexture energies and maximum a posteriori (MAP) estimation to construct a probabilistic deformable model for kidney segmentation. First, using texture image features and MAP estimation, we classify each image pixel as inside or outside the boundary. Then, we design a deformable model to locate the actual boundary and maintain the smooth nature of the organ. Using gradient information subject to a smoothness constraint, the optimal contour is obtained by the dynamic programming technique. Experiments on different datasets are described. We find this method to be an effective approach.

Humans↗
PubMed2005

Quantitatively characterizing the textural features of sonographic images for breast cancer with histopathologic correlation.

OBJECTIVE: In this study, quantitative characterization of sonographic image texture and its correlation with histopathologic findings was developed for facilitating clinical diagnosis. A statistical feature matrix was applied to quantify the texture difference (ie, the dissimilarity) of the sonographic images for malignant and benign breast tumors. METHODS: Thirty-three patients were recruited for this study. Imaging was performed on a commercially available sonographic imaging system in clinical use. The parameters used for image acquisition were kept the same during clinical examination. RESULTS: On the basis of dissimilarity values, 3 phenomena were noted in the relatively large malignancies studied. First, stellate carcinoma showed the least dissimilarity on sonographic images; second, circumscribed carcinoma showed the most dissimilarity; and third, malignant tissue mixed with fibrous and cellular parts (dense lymphocyte infiltration and prominent intraductal tumors) had dissimilarity values in between. Image textures with smaller dissimilarity values (especially for those values <4.4 in our study) are likely to be stellate carcinoma. CONCLUSIONS: From the experimental results, it is shown that the cellular and fibrous content with spatial distribution of breast masses determine the dissimilarity values on sonographic images. The dissimilarity may be used to quantitatively represent the image texture and is well correlated with the histopathologic description.

Adolescent↗
PubMed2002

Initial experience of using color kinesis in the diagnosis of coronary artery disease.

BACKGROUND: Color kinesis (CK) is a recently developed echocardiographic technique. This report describes our initial effort in the validation of the use of CK for the diagnosis of coronary artery disease (CAD). METHODS: Two-dimensional (2-D) echocardiography and CK were studied in 30 normal subjects and 24 CAD patients. Coronary angiography was performed in the 24 patients. Significant (> 70% luminal diameter stenosis) CAD was present in 18 patients (79%), all of whom had history of myocardial infarction. Regional fractional area change in each segment was displayed as a stacked color histogram. The histograms derived from these 30 normal subjects were averaged to obtain the normal pattern of left ventricular contraction; the mean value +/- 1 SD was considered the reference histogram. When the regional fractional area change deviated from this normal reference, this segment was considered as having regional wall motion abnormality. The detection of wall motion abnormalities by visual interpretation of 2-D echocardiography, reviewing the CK loop recording, and CK stacked histograms were compared. To assess the relationship of measurement of endocardial excursion of CK images, the width of the color band was measured at the midpoint of each segment along a line perpendicular to the cardiac border. The endocardial excursion measured by 2 independent observers was compared using linear regression analysis and calculation of intraclass correlation coefficient. RESULTS: The sensitivity and specificity for detection of CAD were 77.8% and 66.6%, respectively, for CK loop reviewing, 83.3% and 66.7% for CK stacked histogram analysis, and 77.8% and 83.3% for 2-D echocardiography. The overall accuracies for CAD detection were 75% for CK loop reviewing, 79.2% for CK stacked histogram analysis, and 79.2% for the 2-D echocardiography (not significant in all comparisons). The correlation of measurement of endocardial excursion from the CK images by 2-observers was good (r = 0.85, p < 0.01), and intraclass correlation coefficient was 0.99 (p < 0.0001). CONCLUSIONS: Our data demonstrate that both the CK loop reviewing and stacked histogram analysis were comparable to 2-D echocardiography for detecting CAD.

Aged↗
PubMedDate not supplied

Texture feature coding method for classification of liver sonography.

This paper introduces a new texture analysis method called texture feature coding method (TFCM) for classification of ultrasonic liver images. The TFCM transforms a gray-level image into a feature image in which each pixel is represented by a texture feature number (TFN) coded by TFCM. The TFNs obtained are used to generate a TFN histogram and a TFN co-occurrence matrix (CM), which produces texture feature descriptors for classification. Four conventional texture analysis methods that are gray-level CM, texture spectrum, statistical feature matrix and fractal dimension, are used also to classify liver sonography for comparison. The supervised maximum likelihood (ML) classifiers implemented by different type texture features are applied to discriminate ultrasonic liver images into three disease states that are normal liver, liver hepatitis and cirrhosis. The 30 liver sample images proven by needle biopsy are used to train the ML system that classify on a set of 90 test sample images. Experimental results show that the ML classifier together with TFCM texture features outperforms one with the four conventional methods with respect to classification accuracy.

Fractals↗