Computer morphometry for liver fibrosis using an automatic image analysis system

Liver biopsy is the gold standard for evaluation of liver diseases. The severity of fibrosis is considered the stage of chronic liver disease. However, it is evaluated by subjective description or semiquantitation. We developed an automatic image analysis system, which including a microscope, computer-driven slide-driver, and a software system for image acquisition, processing and data analysis. The automatic image analysis system deals mainly with color image segmentation. The severity of liver fibrosis is reported as the percentage of the entire fibrous area to the whole liver tissue area. Thirty-one liver needle biopsy specimens are used for this study. The results from computer morphometry were compared with that from colorimetric method and Knodell's score. Pearson correlation and Spearman rank test revealed that the Kondell's score and colorimetric method are significantly correlated with the result of computer morphometry. We found that the system is a reliable tool for evaluating the severity of liver fibrosis.