Automated Image Analysis in Undetermined Sections of Human Permanent Third Molars



AbstractA computerized histomorphometric analysis was made of Karnovsky-fixed, hydroxethylmethacrylate embedded and toluidine blue/pyronin-stained sections to determine: (1) the two-dimensional size of the coronal odontoblasts given by their cytoplasm:nucleus ratio; (2) the ratio between the number of coronal odontoblasts and dentinal tubules; and (3) the relation between odontoblast size and adjacent predentine. All conditions were measured in relation to three well-defined sectioning profiles of the dentinal tubules. The sections were randomly taken from 10 unerupted and erupted third-molar crowns. Sixty-three photomicrographs (x100), equally distributed among the three sectioning profiles, were scanned in a high-resolution scanner to produce images for the analysis. After initial user interaction for the description of training classes on one image, an automatic segmentation of the images with respect to odontoblast cell nuclei, cytoplasm and background was computed by statistical discriminant analysis. In longitudinal profiles of the dentinal tubules the cytoplasm:nucleus ratio in erupted teeth was 3.1 +or- 0.54, and the mean of the odontoblast cell:dentinal tubule ratio was 1.19 +or- 0.20. Analysis of cytoplasm:nucleus ratio and the adjacent predentine in relation to the chosen sectioning profiles disclosed that there was less variation in the predentine measurements in the longitudinal sections. Thus, in future two-dimensional studies of the odontoblast-predentine region only longitudinal sectioning profiles should be analysed. The use of advanced image processing on undemineralized tooth sections provides a rational foundation for further work on the reactions of the odontoblasts to external injuries including dental caries.
Keywordsautomated image analysis, statistical discriminat analysis, dntine, pulp, odontoblasts
TypeJournal paper [With referee]
JournalArchives of Oral Biology
Year1997    Vol. 42    No. 4    pp. 329-332
BibTeX data [bibtex]
IMM Group(s)Image Analysis & Computer Graphics