| Home | E-Submission | Sitemap | Contact Us |  

The Korean Journal of Pathology 1998;32(7): 494-503.
Image Standardization and Determination of Gray Level Threshold in the Assessment of the Myocardial Fibrosis by the Computerized Image Analysis.
Nam Young Lee, Young Sik Park, Jin Haeng Chung, Jeong Wook Seo
1Department of Pathology, Seoul National University College of Medicine, Seoul, Korea.
2Heart Research Institute, Medical Research Center, Seoul National University, Seoul, Korea.
The computerized image analysis is a useful tool for the quantitative assessment of histopathologic findings. In contrast to the usual microscopic examination by pathologists, the computerization should be accompanied with the standardization process of the image. We developed an algorithm to standardize images and to determine the optimal gray level threshold, using a myocardial fibrosis model. Sirius red staining was more convenient for the image analysis than Masson's trichrome staining because of a better contrast with the surrounding structures. To get an optimal measurement, light intensity was standardized at each of the fibrosis, myocardium and background. In this study, the most promising method to determine the degree of fibrosis was that of revising the background without tissue to a gray level of 200, obtaining a green component of the color image, revising the myocardial fiber to 163, and defining a partial ratio as fibrosis index when the gray level threshold was 120. These threshold levels and parameters were determined after drawing the binarization index curves according to the change of the gray level threshold and by the morphological examination of the actual binarization figures overlaid to the original color image. Through these processes we could get a consistent result on the myocardial fibrosis and we expect a similar principle applies when we analyze color images in the histopathologic quantitation by computerized image analysis.
Key Words: Computerized image analysis; Quality assurance; Myocardial fibrosis
PDF Links  PDF Links
Full text via DOI  Full text via DOI
Download Citation  Download Citation
CrossRef TDM  CrossRef TDM
Related article