Nanmu Huang1, Yu Zhang2, William Calawerts3, Jack J Jiang3. 1. Department of Applied Marine Physics and Engineering, College of Ocean and Earth Sciences, Xiamen University, Xiamen, Fujian, China. 2. Key Laboratory of Underwater Acoustic Communication and Marine Information Technology of the Ministry of Education, Xiamen University, Xiamen, China; State Key Laboratory of Acoustics, Institute of Acoustics, Chinese Academy of Sciences, Beijing, China. Electronic address: yuzhang@xmu.edu.cn. 3. Division of Otolaryngology - Head and Neck Surgery, Department of Surgery, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin.
Abstract
OBJECTIVE: The present study aims to compare the correlation dimension and second-order entropy at the minimum embedding dimension with the correlation dimension (D2) and second-order entropy (K2) based on their efficiency and accuracy in differentiating between normal and pathologic voices. METHODS: The minimum embedding dimension was estimated with the Cao method. Nonlinear dynamic parameters, such as correlation dimension and second-order entropy, were used to quantitatively analyze the normal and pathologic voice samples. RESULTS: The computing time of the correlation dimension and second-order entropy at the minimum embedding dimension was reduced to approximately one third of that of traditional D2 and K2 calculations, reflecting higher efficiency. The statistical results of linear fitting suggested that the correlation dimension was highly correlated to the correlation dimension at the minimum embedding dimension, and second-order entropy calculation was highly correlated to the second-order entropy at the minimum embedding dimension. Lastly, the results of statistical comparison proved that the correlation dimension at the minimum embedding dimension and second-order entropy at the minimum embedding dimension were able to significantly differentiate between normal and disordered voices (P <0.001). CONCLUSIONS: The results suggest that the correlation dimension and second-order entropy at the minimum embedding dimension are valid analysis tools for the diagnosis of voice disorders. Additionally, the efficiency and accuracy of these parameters yield potential for clinical usage because of lower computation time than current methods.
OBJECTIVE: The present study aims to compare the correlation dimension and second-order entropy at the minimum embedding dimension with the correlation dimension (D2) and second-order entropy (K2) based on their efficiency and accuracy in differentiating between normal and pathologic voices. METHODS: The minimum embedding dimension was estimated with the Cao method. Nonlinear dynamic parameters, such as correlation dimension and second-order entropy, were used to quantitatively analyze the normal and pathologic voice samples. RESULTS: The computing time of the correlation dimension and second-order entropy at the minimum embedding dimension was reduced to approximately one third of that of traditional D2 and K2 calculations, reflecting higher efficiency. The statistical results of linear fitting suggested that the correlation dimension was highly correlated to the correlation dimension at the minimum embedding dimension, and second-order entropy calculation was highly correlated to the second-order entropy at the minimum embedding dimension. Lastly, the results of statistical comparison proved that the correlation dimension at the minimum embedding dimension and second-order entropy at the minimum embedding dimension were able to significantly differentiate between normal and disordered voices (P <0.001). CONCLUSIONS: The results suggest that the correlation dimension and second-order entropy at the minimum embedding dimension are valid analysis tools for the diagnosis of voice disorders. Additionally, the efficiency and accuracy of these parameters yield potential for clinical usage because of lower computation time than current methods.