Texture Feature Analysis for Ultrasound Image of Skeleton Muscle

碩士 === 國立臺灣海洋大學 === 電機工程學系 === 97 === In clinical, although experienced physicians may make accurate diagnosis when using ultrasonography of musculoskeletal system, for beginners it is rather difficult to recognize subtle change of gray level in the involved muscle. This thesis attempts to provide a...

Full description

Bibliographic Details
Main Authors: Hung-Chieh Shy, 施宏潔
Other Authors: Jung-Hua Wang
Format: Others
Language:zh-TW
Published: 2009
Online Access:http://ndltd.ncl.edu.tw/handle/94179900863026496967
Description
Summary:碩士 === 國立臺灣海洋大學 === 電機工程學系 === 97 === In clinical, although experienced physicians may make accurate diagnosis when using ultrasonography of musculoskeletal system, for beginners it is rather difficult to recognize subtle change of gray level in the involved muscle. This thesis attempts to provide an objective tool which can help improve diagnosis accuracy by extracting texture features present in a pair of ultrasonic images. Two texture features ImpSRHGE and ImpLRHGE are derived from gray level run-length matrices to characterize the B-mode muscle ultrasonography. Then, the aforementioned texture features and the contrast value derived from a spatial gray level dependence matrix are used together to evaluate the degree of muscle fibrosis through distinguishing texture feature variation between the image pair (i.e., normal and abnormal muscles). Our empirical results on ten patients have indicated that subjective diagnosis variation among different physicians may potentially be alleviated.