Kyung-Sook Yang

Changes in Magnetic Resonance Imaging Signal Intensity of Transplanted Meniscus Allografts Are Not Associated With Clinical Outcomes

PURPOSE To evaluate changes in intrameniscal signal intensity (IMSI) of transplanted allografts during the first year after meniscus allograft transplantation (MAT) by use of serial magnetic resonance imaging, as well as to analyze the relation between IMSI and clinical outcome. METHODS This prospective study involved 43 patients who underwent MAT between 2006 and 2007 after diagnosis of total or subtotal meniscectomized knees. The mean patient age at the time of surgery was 35.8 years (range, 17 to 46 years). Allografts were assessed by conventional magnetic resonance imaging performed at 6 weeks and 3, 6, and 12 months after MAT. The ratio of the signal intensity of the transplanted meniscus allograft to that of the control normal meniscus in the ipsilateral knee was calculated to obtain a standardized signal intensity value. IMSI was assessed in terms of postoperative time and location (anterior v posterior horn). The Lysholm score was used to evaluate knee function. RESULTS The IMSI of transplanted allograft menisci was higher than that for nontransplanted menisci at all 4 postoperative time points (P < .01). The anterior horn allograft IMSI was greater than the posterior horn allograft IMSI at all time points (P < .01). The allograft IMSI increased starting 3 months postoperatively for the anterior horn (F(3,40) = 7.5, P < .01) and 6 months postoperatively for the posterior horn (F(3,40) = 9.2, P < .01). These increases were maintained to the final assessment at 1 year postoperatively. No correlation was found between IMSI and postoperative Lysholm score. CONCLUSIONS Transplanted allograft menisci had higher signal intensities than normal menisci. Signal intensity was higher for the anterior horn than the posterior horn throughout the first postoperative year. Signal intensity increased over time, and this increase was maintained at 1 year postoperatively. However, signal intensity was not related to clinical outcome. LEVEL OF EVIDENCE Level II, development of diagnostic criteria based on analysis of consecutive patients, applying a universally recognized gold standard.

Feature selection method using WF-LASSO for gene expression data analysis

There has been a lot of research that demonstrates the phenomenon of life or the origin of the disease, and classifies or diagnoses the state of the cell. These are usually achieved by the strength of the gene expression under certain circumstances using microarrays, which can observe tens and thousands of gene expression profiles. It is not feasible to use all the attributes because of the huge amount of gene expression data that are involved in microarray experiments. It is not feasible to use all the attributes because a lots of gene expression data are involved in microarray experiments. That is, because microarray data have a small number of samples compared to the number of the attributes, in the analyzing of the data there will be overfitting which requires a high cost due to the high dimensionality of the data. We propose a feature selection method using a technique which combines filter method with wavelet transform, and LASSO regression method based on a statistical regression analysis. We obtain the best classification results by applying, in order, the DWT, the filter method, and then finally LASSO. That is, the feature selection method with the best classification performance was WF-LASSO method. The contribution of this paper is in that it is possible to solve problems by reducing the dimensionality of a high volume of data by using the proposed method, so that the performance of the classification can be improved and a more stable classification model can be constructed.

Acute Biophysical Responses and Psychological Effects of Different Types of Forests in Patients With Metabolic Syndrome

We aimed to examine the biophysical and psychological effects of two different types of forests on women with metabolic syndrome (MetS). Seventy-one middle-aged Korean women participated in a half-day healing program at either Saneum Recreational Forest (a wild forest) or Seoul Forest (a tended forest) depending on their preference. The participants’ biophysical parameters and Profile of Mood States (POMS) were measured before and after the healing program. The two groups differed significantly in acute insulin responses, pulse rate, oxidative stress markers, and stress hormone level, suggesting better homeostasis in the wild forest. The POMS suggested that the mean vigor and confusion scores significantly decreased in the wild forest. Considering the significantly more favorable acute insulin reaction and levels of oxidative stress and the trend toward positive mood state in the wild forest, we recommend that middle-aged patients with MetS participate in healing programs in wild forests.

Feature Selection of Gene Expression Data Using Regression Model

There have been a lot of researches that demonstrate the phenomenon of life or the origin of the disease and classify or diagnose the state of the cell. These are usually achieved by the strength of the gene expression under certain circumstances by the microarray which can observe tens and thousands of gene expression profile. It is not feasible to use all the attributes because a lots of gene expression data are involved in microarray experiments. Therefore, in order to select the significant genes from lots of data, we applied the hybrid method combining filter method with LASSO model. As experimental data set, leukemia data are applied to a number of classifiers such as naïve Bayesian, SVM, Bayesian network, logistic regression and random forest. In the experimental result, we found that the gene selection method using the LASSO outperforms the existing gene selection method.

Value of CT added to ultrasonography for the diagnosis of lymph node metastasis in patients with thyroid cancer

The benefit of CT for the diagnosis of lymph node metastasis in patients with thyroid cancer is still unclear.

Canonical Discriminant Analysis of Multinomial Samples with Applications to Textual Data

We develop the canonical discriminant analysis of the G groups data consisting of n 1, …, n G multinomial samples within each group, on scaled Euclidean space with chi-square distance. Our discriminant analysis produces quantification plots showing both q observation categories and N (= n1+…+ n G ) multinomial sample units, as well as G group centroids. We apply the proposed method to Korean text analysis to extract statistical characteristics of Korean language by genres.