Abstract

Info

Title:: Quantitative detection of liver parenchymal lesion characteristics after the combination with fast Fourier transform

Author(s):: NIU Guangli1; LIU Xiang1; SONG Jialin2; TANG Xian1; 1. School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China 2. Department of Ultrasound Diagnosis and Treatment, Changzheng Hospital, the Second Military Medical University of Chinese Peoples Liberation Army, Shanghai 200003, China

Keywords:: Keywords: liver parenchyma liver cirrhosis Fourier transform quantitative detection texture feature

Abstract:: Abstract: Objective To explore the feasibility of the quantitative detection of liver parenchymal lesions in high-frequency ultrasound images. Methods The two-dimensional high-frequency ultrasound images of the liver were collected from 47 patients with mild, moderate and severe hepatitis B cirrhosis (mild, moderate and severe cirrhosis groups) and 20 healthy volunteers (normal control group). By combining fast Fourier transform with image morphological processing techniques, the morphological features of liver parenchyma were obtained automatically, and the lesion characteristics were quantified. Finally, the pathological features were classified in detail by scoring strategy. Results There were statistical differences in the data obtained by scoring strategy among mild, moderate, severe cirrhosis groups and normal control group (P<0.001). The multiple comparisons revealed that the quantified data of cirrhosis groups were significantly different from those of normal control group, and that the differences between groups were statistical significant (P<0.05), except for the comparison between severe cirrhosis group and mild or moderate cirrhosis group. Conclusion The texture features of liver parenchyma in accordance with doctors view can be extracted and quantified by fast Fourier transform, which provides data support for the further diagnosis and grading of liver cirrhosis.