Hiroshi Isono1, Shinnosuke Hirata2, Tadashi Yamaguchi3, Hiroyuki Hachiya2. 1. Graduate School of Science and Technology, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro, Tokyo, 152-8552, Japan. isono@us.ctrl.titech.ac.jp. 2. Graduate School of Science and Technology, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro, Tokyo, 152-8552, Japan. 3. Research Center for Frontier Medical Engineering, Chiba University, Chiba, 263-8522, Japan.
Abstract
PURPOSE: Chronic liver disease requires careful follow-up during long-term treatment, and development of a quantitative diagnosis method for liver fibrosis based on an ultrasonic imaging system is highly desired. METHODS: Texture analysis using a co-occurrence matrix was applied to both clinical and simulated ultrasonic images of fibrotic livers. A sequence of matrices was generated for pixel-pair distance, r, and texture feature contrast was chosen to examine the response to r in combination with statistical analysis of echo amplitude distribution using a multi-Rayleigh model. RESULTS: The contrast converged with a larger value and fluctuated more significantly in response to r as fibrosis progressed in both the clinical and simulated ultrasonic images. The convergent value rapidly increased at the early stage of fibrosis, and the fluctuation became larger at the advanced stage of fibrosis. Analysis using simulated ultrasonic images with a known fibrous tissue structure theoretically clarified the relationship between contrast behavior and fibrosis progression. CONCLUSION: It was revealed that contrast convergent value and contrast fluctuation provided information on the fibrous tissue structure, and they are expected to be used for quantitative diagnosis of the degree of liver fibrosis.
PURPOSE:Chronic liver disease requires careful follow-up during long-term treatment, and development of a quantitative diagnosis method for liver fibrosis based on an ultrasonic imaging system is highly desired. METHODS: Texture analysis using a co-occurrence matrix was applied to both clinical and simulated ultrasonic images of fibrotic livers. A sequence of matrices was generated for pixel-pair distance, r, and texture feature contrast was chosen to examine the response to r in combination with statistical analysis of echo amplitude distribution using a multi-Rayleigh model. RESULTS: The contrast converged with a larger value and fluctuated more significantly in response to r as fibrosis progressed in both the clinical and simulated ultrasonic images. The convergent value rapidly increased at the early stage of fibrosis, and the fluctuation became larger at the advanced stage of fibrosis. Analysis using simulated ultrasonic images with a known fibrous tissue structure theoretically clarified the relationship between contrast behavior and fibrosis progression. CONCLUSION: It was revealed that contrast convergent value and contrast fluctuation provided information on the fibrous tissue structure, and they are expected to be used for quantitative diagnosis of the degree of liver fibrosis.