Constant False Alarm Rate Processing Based on Multi-Rayleigh Model for Extraction of Fibrotic Signals in Liver Fibrosis: A Preliminary Study

An extraction of fibrotic signals in fibrotic liver will contribute to diagnosis of liver fibrosis. As the fibrotic signals show higher variance of envelope amplitudes compared to that of norma tissue signals, the extraction of higher amplitude signals can extract a part of fibrotic signals. A constant false alarm rate (CFAR) processing is one of the thresholding methods to extract higher amplitude signals with quantitatively setting a threshold to be that the false alarm rate, that is, the false extraction probability of background signals, becomes constant. In a previous study, a Rayleigh based CFAR processing was proposed to extract the fibrotic signals because a probability density function (PDF) of normal liver tissue can be modeled by a Rayleigh distribution. However, in a progressive liver fibrosis, the PDF greatly deviates from the Rayleigh distribution; therefore, the PDF of background signals cannot be modeled by the Rayleigh distribution. In this study, we examined a multi-Rayleigh based CFAR processing. The multi-Rayleigh model is a PDF model for fibrotic liver and can extract the normal tissue component. Therefore, the threshold value was set to be that the false alarm rate of PDF of normal tissue component in the multi-Rayleigh model becomes constant. The in vivo data analysis showed that the multi-Rayleigh based CFAR processing increased the extracted rate of fibrotic signals for the liver fibrosis without increasing that for the non-fibrotic liver. Thus, the multi-Rayleigh based CFAR processing has a potential to improve the sensitivity of extraction of fibrotic signals, that will contribute to the higher-sensitivity detection of liver fibrosis.