Transactions on Mass-Data Analysis of Images and Signals (ISSN:1868-6451)


Volume 2- Number 1 - September 2010 - Pages 29-40


Object Oriented Color Image Segmentation: Applications in Digital Pathology and USG Sonoelastography

E. Kaczmarek1, C. Helak-Łapaj1, J. Jagielska1, A.Seraszek1, Ł. Łapaj2, A. Kasprzak3, M. Chmielewski4 and A. Surdacka5

1Dept. of Bioinformatics and Computational Biology, Chair of Pathology, University of Medical Sciences, Przybyszewski str 49, 60355 Poznan, Poland
2Department of Orthopaedics, Orthopaedic Oncology and Traumatology, University of Medical Sciences, Poznan, Poland
3Department of Histology and Embryology, University of Medical Sciences, Poznan, Poland
4Department of Gastroenterology, Human Nutrition and Internal Diseases, University of Medical Sciences, Poznan, Poland
5Department of Conservative Dentistry and Periodontology, University of Medical Sciences, Poznan, Poland


Abstract

Digital biomedical image archives provide fascinating new opportunities for research, education and consultation. Objective interpretation of tissue slides should be performed on the basis of their quantitative measurements. The aim of this paper was to implement efficient image analysis applications for color segmentation of objects of interest and their measurements in histological digital slides. A compatibility of quantitative evaluation of liver injury based on digital slides with clinical diagnosis based on USG sonoelastography images was also studied. In addition, examples of measurements of biomarkers’ expression in gallbladder epithelial cells and gingival blood vessels in chronic periodontitis are presented in this paper. Concluding, efficient algorithms of object oriented color image segmentation will facilitate quantitative studies of whole-slide-images stored in digital pathology archives.



Back to Table of Contents