In-vivo human liver tissue differentiation

A pattern-recognition algorithm and the required instrumentation to apply it for in vivo human liver tissue differentiation are discussed. The algorithm has been tested successfully, with a confidence interval of 68.27%, on 25 subjects with no history of liver diseases and 15 subjects with different types of abnormalities. Differentiation between normal and abnormal liver tissue is accomplished by calculating the Euclidean distance between a reference vector and a pattern vector. The elements of the reference vector are the precalculated values of the average attenuation and backscattering coefficients of normal liver tissue at each frequency interval in the range from 1.5 to 4.5 MHz. Elements of the pattern vector are the average values of the two coefficients for the liver tissue under consideration at each frequency interval. This distance is a measure of the probability that the liver under consideration is normal. An empirical threshold is selected such that if the distance is less than the threshold, then the liver is declared normal otherwise it is abnormal. The instrumentation implemented is a high speed microprocessor-based data acquisition and analysis system. The system digitizes the backscattered ultrasound signal and stores the digitized data in a microcomputer where it is analyzed