An Automatic Image Analysis System for Color Liver Microscopic Images
| Autor: | Guo ,Xiao-Zhen, 郭蕭禎 |
|---|---|
| Rok vydání: | 1995 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 83 Liver pathology is usually an important technique used in the diagnoses of liver diseases. In chronic liver disease liver fibrosis is one of the most important progress and it may lead to cirrhosis. The diagnosis of fibrosis is established by analysis of a liver biopsy. However, to evaluate the severity of fibrosis from biopsy specimen is currently quite subjective, or at best, semi-quantitative by scoring systems. In this thesis, a PC-based automatic analysis system has been developed for evaluating the severity of fibrosis. We apply the computer morphometry to measure the biopsy specimen objectively by segmenting color liver microscopic images. The system will examine the whole liver section with automatic control, including shiftable slide seat and proper image grabbing equipment which are all program controllable. In particular, a novel method to segment color liver images has been proposed and implemented in two different modes, i.e., unsupervised and supervised. The unsupervised method uses the HSI color space, while the supervised one utilizes a new linear color space based on the canonical transformation. Both two modes are implemented with a two-stage fast segmentation algorithm which can effectively separate hepatocyte, fiber and background on every captured liver image. Furthermore, the system also provides quantificational ratio of fibrosis in liver diagnosis. The system seems very promising in clinical utilization. It takes from 10 to 40 minutes to analyze a single specimen which includes all the necessary acquisition, control and image processing procedures. Thirty-one liver biopsy specimens from chronic viral hepatitis B patients were analyzed in our system. The experimental results show that the proposed method has high correlation with conventional diagnoses. That is, the system is deemed a quite reliable and useful tool for objective quantitation of the severity of liver fibrosis. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|