×

Microscopic image classification based on descriptor analysis. (English) Zbl 0973.92019

Alt, Wolfgang (ed.) et al., Dynamics of cell and tissue motion. Workshop, Bonn-Röttgen, Germany, March 1995. Basel: Birkhäuser. Mathematics and Biosciences in Interaction. 47-54 (1997).
From the introduction: The direct form of cell image acquired with a computer-based microscopic system reflects the distribution of luminosity (and/or, in some cases, of colour), but does not embrace the original spatial relations of the objects under examination. The fact that the image pixels are interrelated by an artificial structure of the pixel table has two major inconvenient consequences: on the one hand, a huge computer memory is required, because all parts of images, even unimportant such as background, have to be stored with high resolution to avoid loss of significant information and, on the other hand, it is difficult to compare the images from the view point of specific features of the objects, e.g., shape. Therefore, it is necessary to transform the primary, directly acquired image into a standard form of image description closely interrelated with chosen aspects of original images. If the class of examined objects is sufficiently narrow it is often possible to define multiple image aspects which are not important in a given case and to discard them in the process of image description. This approach enables one to develop methods which considerably decrease computer memory required to store images and makes image comparison faster and easier. The method of a two- and three-dimensional cell image analysis presented in this paper is based on this concept. It has been applied in the study concerning the adhesion of L1210 cells (lymphocytic leukemia, mouse) as examined with the use of a fluorescent microscope.
For the entire collection see [Zbl 0877.00017].

MSC:

92C55 Biomedical imaging and signal processing
92-08 Computational methods for problems pertaining to biology
92B99 Mathematical biology in general
PDFBibTeX XMLCite