Sung-Hyun Yoon

Learn More
PURPOSE To evaluate the risk factors related to the development of new fractures in adjacent vertebrae after vertebroplasty. MATERIAL AND METHODS The study was conducted on 106 patients in whom 212 vertebroplasties were performed during a period of 3 years. Evaluations of the five vertebrae superior and inferior to the treated vertebra were performed.(More)
Hydroxyapatite (HA) is commonly used as bone substitute in clinical practices. However, only few studies have compared the relationship between the mixture ratio of bone graft in the actual clinical field and fusion rate according to bone graft volume. The study aimed to analyze the fusion rate according to the mixture ratio and the amount of bone graft in(More)
Few studies on speaker verification have directly used a deep neural network (DNN) as a classifier. It is difficult to directly apply a DNN as a discriminative model to speaker-verification tasks because the training data for each speaker are very limited. Therefore, a b-vector has been proposed to solve the problem. However, the DNN with the b-vectors(More)
We propose a method to improve speaker verification performance when a test utterance is very short. In some situations with short test utterances, performance of ivector/probabilistic linear discriminant analysis systems degrades. The proposed method transforms short-utterance feature vectors to adequate vectors using a deep neural network, which(More)
PURPOSE We attempted to identify changes in back muscle atrophy occurring in multilevel minimally invasive transforaminal interbody fusion (MITLIF) and the impact of these changes on clinical outcomes. METHODS This study was conducted on 92 patients who underwent unilateral MITLIF between 2006 and 2013, had been tracked with a follow-up for at least 1(More)
We propose an expanded end-to-end DNN architecture for speaker verification based on b-vectors as well as d-vectors. We embedded the components of a speaker verification system such as modeling frame-level features, extracting utterance-level features, dimensionality reduction of utterancelevel features, and trial-level scoring in an expanded end-toend DNN(More)
  • 1