L. Bjørndal (1), T. Darvann(2), M. Bro-Nielsen(2,3) and A. Thylstrup(1).
1) Department of Cariology and Endodontics, School of Dentistry, University
of Copenhagen
Nørre Alle 20, DK 2200 Copenhagen, Denmark
2) 3D-lab, School of Dentistry, University of Copenhagen, Nørre
Alle 20, DK 2200 Copenhagen, Denmark
3) Department of Mathematical Modelling, Technical University of Denmark
INTRODUCTION AND AIM
The terminology 'reduced odontoblasts' has conventionally been associated
with the concept of a reduced function of the cells or to be a sign of
catabolic reactions. This interpretation is probably a result of methodological
problems specifically related to examination of demineralized sections.
The aim of this study was to apply advanced image processing to undemineralized
tooth sections [Donath, 1987; Bjørndal et al., 1994] in order to
make a quantitative analysis of the odontoblast-predentine region in relation
to caries progression.
MATERIAL AND METHODS
The material consisted of undemineralized 10-15 mm thin tooth sections
prepared from 10 unerupted and erupted third molars wihtout signs of caries.
Three different histological cutting profiles of the coronal dentinal tubules
were defined in the light microscope (x1000): A longitudinal (length of
cutting profiles of the dentinal tubules > 3 mm, Fig. 1a); a longitudinal-transversal
(length of cutting profile between 1mm and 3 mm, Fig. 2) and a transversal
(length of cutting profiles < 1 mm, Fig. 3). The sections were stained
with toluidine blue and pyronin, and by a modification of Brown and Hopp's
method. A total of 63 photomicrographs (x400) equally distributed among
the cutting profiles were scanned by a high-resolution (1000 l/cm) scanner
(Eskofot Eskoscan) in order to produce images for the computerized histomorphometric
analyses. The following conditions were measured: A) The cytoplasm:nucleus
(C:N) ratio of the odontoblast cell; B) The odontoblast cell:dentinal tubule
(OB:DT) ratio, and C) the adjacent predentine area (mm ). In addition,
40 sections were examined from 20 molars with caries lesions and the findings
compared with intra-section control areas. These lesions were classified
with respect to lesion progression, lesion activity and lesion age.
RESULTS
Initial analysis disclosed that the statistical discriminant analysis
(SDA) [Saeboe, 1985; Haslett, 1985] provided more information than methods
based on simple thresholding. Figs. 1a - c present an example of the SDA
procedure [Larsen, 1996] including contextual classification, for computing
the C:N ratio of the odontoblast cells. After an initial user interaction
for description of training classes on one image using the image processing
package Mvox [Bro-Nielsen, 1996] (Fig. 1b), an automatic segmentation of
the images with respect to odontoblast cell nuclei, cytoplasm and background
is computed by SDA. Fig. 1c shows the result of the SDA analysis
followed by some additional cleanup. Standard image processing tools
is applied to this image in order to count the number of pixels occupied
by nuclei and cytoplasm, respectively. The number of dentinal tubules
is found by applying thresholding and morphological operators to the area
within the blue box in Fig. 1d. The predentine area is computed after manual
drawing as indicated by the red outline. In longitudinal profiles
the OB:DT ratio is close to 1:1 (mean: 1.19 +/- 0.20) which is significantly
different (ANOVA, p < 0.01) from the transversal cutting profiles (mean:
1.72 +/- 0.57). Data presented in Table 1 indicates that the C:N ratio
in longitudinal cutting profiles is higher in unerupted teeth (6:1) than
in erupted teeth 3:1. Fig. 4 displays the C:N ratios and the predentine
area in relation to the cutting profiles. The data indicates that the variation
in the predentine area measurements was markedly reduced in longitudinal
cuttings. For this reason the results on odontoblast-predentine reactions
are described on the basis of longitudinal cutting profiles. The OB:DT
ratio was about 1:1 in active and inactive lesions, and the ratio were
not influenced by progression stage (+/- enamel cavitation, Fig. 5). Figs.
6 and 7 show the relation between C:N ratio and predentine area indicating
that cell size is positively associated with predentine area. Fig. 8 gives
the same relation by lesion age.
CONCLUSION
1. The combination of undemineralized thin sections and advanced image
processing made it possible to perform a quantitative analysis of the odontoblast-predentine
reactions to dental caries.
2. The new technique allowed us to demonstrate that 'reduced odontoblasts'
is a problematic terminology as almost all primary odontoblasts were involved
in dentine reactions even in active lesions with enamel cavitation.
REFERENCES