Automated System for Analysis of Microscopic Liver Images for Detection of Hepatic Failure

Abstract

Liver viruses, alcoholism and exposure to toxins are risk factors causes liver damage that leads towards hepatic failure. Progressive damage to the liver indicates the presence of abnormal hepatocytes cells. Automatic screening of hepatic disorder through microscopic images are in place and considered in image processing. An automated system is presented for the detection of hepatic failure in digital liver images. Activation of FoxO3 in mice resulted in metabolic abnormalities which leads to progressive hepatic failure. Hepatic histological analysis of normal and transgenic mice is accomplished to study the behavior of hepatocytes after different time activation of FoxO3 proteins. Feature extraction modules extract a number of color and texture based features followed by classification which classifies liver images as normal and transgenic. Strongest activity is observed at transgenic mice liver by luciferase imaging in vivo after injecting luciferin. Microscopy analysis of animals indicate the presence of abnormal hepatocytes in mice liver. The proposed system is tested on liver image dataset collected from university of Ulm. Multiple classifiers like Knn, SVM, Naïve Bayes are utilized to compare the results for better accuracy measurement.