Image Analysis and Statistics: A Quantitative Approach to Immunocytochemistry and Related Techniques

Abstract

Historically, the interpretation of immunocytochemical (ICC) or histological data has been principally qualitative, by which the chromogenic end product was described in terms of a localization pattern. Typically, a subjective assessment is made of the specific cell type(s) involved, the location and intensity of the staining within the cell (e.g., cytoplasmic, membrane associated, nuclear), and if appropriate, the position of the cell(s). More recently, this chromogenic technology has been extended to in situ hybridization (ISH) with the introduction of biotinylated (Cook et al. 1988) and digoxigenin-labeled probes (Farquharson et al. 1990). This has undoubtedly had a considerable effect on the speed at which data can be generated but, compared to isotopic methods quantification, is more problematic. The ability to visualize protein and mRNA expression has been invaluable in advancing our understanding of normal and disease conditions (Hayward et al. 1994, Issazadeh et al. 1995, McGeer et al. 1992, Tokuda et al. 1991). However, the subjective approach used to analyze the data has meant that the amount and type of information obtained depend on the individual analyst and his or her experience. Each analyst may observe, describe, and classify the same feature differently; for example, in Figure 9.1, if plaque 1 is diffuse, plaque 2 primitive, and plaque 3 classic, how would plaque 4 be classified? This use of subjective assessments for determining localization patterns can lead to considerable variations in data analysis and, therefore, in the quality of interpretation.KeywordsAmyloid PlaqueNest ExperimentGray Level HistogramPixel Gray LevelColor Image AnalysisThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.