Beschreibung:
A quantitative diagnostic method for liver fibrosis using ultrasound B-mode image is highly required. To quantitatively evaluate liver fibrosis using ultrasound B-mode image, we have been focusing on the echo amplitude distribution properties. A probability density function (PDF) of echo envelope from a normal liver can be approximated by a Rayleigh distribution; however, the PDF of echo envelope from a fibrotic liver deviates from the Rayleigh distribution. To approximate the PDF of liver fibrosis, we have been proposing the multi-Rayleigh model which is a combination of Rayleigh distributions with different variances. By using the multi-Rayleigh model, the ultrasound B-mode image of liver fibrosis could be converted to the quantitative fibrotic probability image. In this study, we applied the fibrotic probability imaging method to clinical data which is classified into F0 (normal liver), F1 (early-stage hepatitis), F2 (moderate hepatitis), and F3 (late hepatitis) in accordance with a new Inuyama classification. First, a region of interest (ROI) was automatically selected based on the amplitude distribution property. Then, removal method of nonspeckle signals was used to improve the accuracy of the multi-Rayleigh model approximation. Finally, the ultrasound B-mode image in ROI was converted to the fibrotic probability image based on the multi-Rayleigh model.