High Resolution Digital Tissue Image Processing Using Texture Image Databases

摘要

Texture based image databases integrated with effective searching algorithms are useful solutions for many scientific and industrial purposes. Medical image processing of high resolution tissue images is one of the areas, where the cell/tissue classification can rely on such solutions. In this paper we are describing the design, development and usage of a specialized medical texture image database. Our primary aim with this texture database is to provide Digital Imaging and Communication in Medicine (DICOM) compatible texture image dataset for cell, gland and epithelium classification in histology. Our solution includes a Picture Archiving and Communication System (PACS) subsystem, which is mainly provide a communication interface (texture image searching and retrieval) and enables image processing algorithms to work more effectively on high resolution tissue slide images. In this paper we describe how our Local Binary Pattern (LBP) based algorithm benefits texture database usage when solving image processing problems in histology and histopathology.