- The sensitivity is very high. Almost all of ballooning hepatocytes is detectable.
- Assists diagnosis of NASH and study for biomarker and therapeutic technology.
Background and Technology
NASH (Nonalcoholic steatohepatitis) is a category of nonalcoholic fatty liver disease and possibly deteriorate into liver cirrhosis and hepatocellular carcinoma. Although diagnosis of NASH has been performed biopsy to find ballooning hepatocytes in a microscopic image, even skilled pathologist sometimes does not notice the hepatocytes and make a wrong diagnosis. No established biomarker to detect NASH has been found. Thus, diagnosis of NASH is very difficult and should be developed further for tens of millions of patients with liver diseases.
We propose new NASH diagnosis strategy with biopsy image analysis. The process is consisted of two technologies. One is Gaussian Mixture Modelling (GMM). GMM is a statistical model used for clustering data information, for example a histogram of intensity or RGB color. In our technology GMM is performed to distinguish cell fiber, cytoplasm or nucleus in biopsy image (see the left side in the figure). The other technology is detection of outline of hepatocyte. Fitting ellipse-shape model to outline of hepatocytes in the image enables to estimate deformation of the cells and detect its ballooning automatically (see the right side). This method established almost perfect diagnosis of NASH with liver biopsy image.
Data and Publications
- Ballooning hepatocytes are automatically detected from biopsy image. Our technology detected almost all of hepatocytes manually detected by mature pathologist.
Professor Eiji MIYOSHI (Osaka University)
We are looking for a company to use this technology as companion diagnosis software of NASH and/or research tool to study the disease and perform clinical trial.
Product No: TP-00736