Dentomaxillofacial Radiology (2009) 38, 17-22
© 2009 British Institute of Radiology
doi: 10.1259/dmfr/99191766
Correlation between histopathological image and radiographic image pattern in fibro-osseous lesions in relation to bone complexity and distribution
M Araki*,1,4,
S Kawashima1,4,
N Matsumoto2,5,
S Nishimura3,5 and
K Komiyama2,5
1Department of Oral and Maxillofacial Radiology, Nihon University School of Dentistry, Tokyo, Japan; 2Department of Pathology, Nihon University School of Dentistry, Tokyo, Japan; 3Department of Oral and Maxillofacial Surgery, Nihon University School of Dentistry, Tokyo, Japan; 4Division of Advanced Dental Treatment, Dental Research Center, Nihon University School of Dentistry, Tokyo, Japan; 5Division of Immunology and Pathobiology, Dental Research Center, Nihon University School of Dentistry, Tokyo, Japan
*Correspondence to: M Araki, Department of Oral and Maxillofacial Radiology, Nihon University School of Dentistry, 1-8-13 Kanda-Surugadai, Chiyoda-ku, Tokyo 101-8310, Japan. E-mail: araki-m{at}dent.nihon-u.ac.jp
Received 4 October 2007;
revised 5 January 2008;
accepted 7 January 2008
 |
Abstract
|
|---|
Objectives: This study examined correlations between radiographic patterns and the shape of osteoid tissue formations, as determined histopathologically.
Methods: 20 cases of fibro-osseous lesions were investigated, comprising 5 radiographic patterns: focal (n = 3), target (n = 6), lucent (n = 4), calcification (n = 3) and multiconfluent (n = 4). Histopathological images in the central area of a full-section specimen were transformed into binary images and then into 8-bit scale images. Bone complexity and density of bone distribution were calculated and compared between patterns.
Results: Bone complexity score was 7384.64 for lucent, 2029.85 for focal, 2713.40 for multiconfluent, 8388.63 for calcification and 1364.27 for target pattern. The results could be broadly separated into two types: small (target, focal and multiconfluent patterns), and large (lucent and calcification patterns). Density of bone distribution was relatively low in all areas for lucent and calcification patterns, and high for focal, multiconfluent and target patterns. No significant differences in bone complexity or density of bone distribution were seen between individual patterns.
Conclusions: Correlations appear to exist between image patterns from radiography and the shape of osteoid tissue on histopathology, but reorganization of the five patterns may be warranted.
Keywords: fibro-osseous lesion; histopathology; binary image; bone complexity
|
Introduction
|
|---|
Differential diagnosis of fibro-osseous lesions (FOLs) in the jaws is often difficult, as the internal condition of lesions changes over time.1–4 In 2003, we demonstrated that FOLs shown on panoramic view are histopathologically classifiable into three groups: tumour, dysplasia and inflammation. FOLs can also be subdivided into five patterns based on radiographic findings: focal (Figure 1a
), target (Figure 1b), lucent (Figure 1c), calcified (Figure 1d) and multiconfluent (Figure 1e).5 We have already reported that the quantity of osteoid tissue formations tends to be associated with radiopaque strength according to the radiographic pattern.6 The osteoid tissue formation ratio (O/F) represents the ratio of pixel numbers between osteoid tissue (O) and fibrous connective tissue (F), and is considered very significant in determining the image pattern in panoramic radiography, separating osteoid tissue from fibrous connective tissue according to pixel numbers using ImageJ software (v1.37; Wayne Rasband, NIH, Bethesda, MD). However, using these data alone cannot clearly explain image patterns on radiography without reference to different factors. Clinicians also need to practise lesion diagnosis based on radiographic results to plan treatment. The present study attempted to use the complexity and distribution of factors to clarify morphological features of lesions and allow comparison with radiographic patterns using binary images.
View larger version (85K):
[in this window]
[in a new window]
|
Figure 1 Radiographs of clinical cases were classified into 5 patterns. (a) Focal pattern, radiopacity with uniformity (case 14); (b) target pattern, central radiopacity within radiolucent area (case 8); (c) lucent pattern, well-defined radiolucent area (case 18); (d) calcification pattern, calcified flecks or scattered within radiolucent area (case 3); (e) multiconfluent pattern, radiopacity with mottled pattern (case 6)
|
|
|
Materials and methods
|
|---|
The present study examined 20 FOLs and analysed both complexity and distribution of osteoid tissue in the 5 patterns of pathological findings, based on radiographic patterns. The 20 cases of FOL had already been diagnosed histopathologically and radiographic patterns determined as focal (n = 3), target (n = 6), lucent (n = 4), calcification (n = 3) or multiconfluent (n = 4).
Preparation of histopathological images
A total of 20 cases of FOLs were selected in the study. Excised specimens were fixed with 10% buffered formalin and decalcified with K-CX solution (FALMA, Tokyo, Japan). Decalcified specimens were cut in the centre of the long axis. Thereafter, the specimens were embedded in paraffin and serial sectioned at 10 µm thickness. Sliced tissues were mounted on a glass slide and stained with haematoxylin and eosin.
Pathological images were taken under x20 magnification at 5 glass slide intervals. The microscope system was built from a microscope (AX-80; Olympus, Japan), charge-coupled device camera (HC-2500; Fujifilm, Tokyo Japan) and software (Photograb-2500 v1.1; Fujifilm, Tokyo, Japan). Approximately 20–45 pathological images were assembled into a whole image of glass slides using Tiling Boutique software (v3 for Windows; Sanyo Electric, Osaka, Japan). Assembled images were saved as tagged image file format (TIFF).
Transformation to binary images
The tiling images of 20 cases were transformed into binary images and then into 8-bit scale images using ImageJ software. When histopathological images were modified to create binary images, the threshold was determined carefully. The score, with regard to the threshold, often differed among specimens due to the degree of bone tissue staining and the duration of decalcification. In this study, we therefore judged the difference between bone tissue and fibrous connective tissue and selected a threshold of bone tissue that is equivalent to 160–250 for each case (Figure 2). In addition, we had a problem concerning how to manage small scattered areas of osteoid tissue that were less than 100 pixels; these were excluded from the bone tissue data in this study.
View larger version (46K):
[in this window]
[in a new window]
|
Figure 2 a) Histopathological examination of a tiling image in the central area of a case (target pattern ) used in this study (magnification x20; haematoxylin and eosin). (b) Binary image divided by the threshold, directly confirming whether bone tissue or fibrous connective tissue is present at the site of the histopathological image
|
|
Complexity and distribution
The number of pixels constituting the areas of osteoid tissue, length of the perimeters and bone density of distributions in osteoid tissues were calculated using the included measurement program. Complexity was used in this study to support an index of shape and was determined using the following formula: complexity = (area)2/(perimeter). Bone complexity shows a higher score with an increasingly complicated shape of the osteoid tissue. Bone complexity also offers a useful means of comparing the conditions of internal structures in different osteoid tissues.
The distribution of osteoid tissue within each lesion was examined, due to the scattering of osteoid tissues forming in each sample. Lesions were divided into 25 equal blocks to estimate the rate of osteoid tissue as a matter of convenience (Figure 3). These blocks comprised 16 marginal areas, 8 middle areas and 1 central area. Distribution was estimated using bone density in each block, expressed as an average for each of the three areas.
View larger version (103K):
[in this window]
[in a new window]
|
Figure 3 Binary image of a target pattern divided into 25 equal areas, comprising 16 marginal areas, 8 middle areas and 1 central area
|
|
Statistical analysis
Relationships to bone complexity and distribution of bone density were compared between patterns using the Kruskal–Wallis test. Relationships between each pattern and the three areas (marginal, middle and central) were further analysed using Welch's test.
|
Results
|
|---|
Table 1 shows the mean bone complexity score from binary images. These results indicate two broad types of lesion: small type, in target, focal and multiconfluent patterns; and large type, in lucent and calcification patterns. The bone complexity score for the multiconfluent pattern was between that for the target and focal patterns. Calcification pattern showed a higher complexity score than lucent pattern. These results showed different tendencies in terms of mean lesion density on intraoral film and O/F ratio from histopathological images.
View this table:
[in this window]
[in a new window]
|
Table 1 Bone complexity and means measured from binary image-transformed bone tissue in histopathological findings
|
|
Table 2 shows the mean score for distribution of bone density. Distribution of bone density was relatively low in all areas for lucent and calcification patterns, and high in all areas for focal, multiconfluent and target patterns.
|
Discussion
|
|---|
FOLs observed clinically on radiography in this study used the new classification, improving on the classification described in 1984 by Eversole et al.4 This new classification consists of focal, target, radiolucent, calcification and multiconfluent patterns (Figure 1
). However, patterns do not objectively represent a radiopaque mass as such, but instead offer characteristics of images interpreted through extensive analysis. Indeed, we are often puzzled by the classification of FOL patterns. Transforming histopathological images into binary images is not easy. In many cases, properly separating bone tissue from fibrous connective tissue was nearly impossible to do according to a fixed threshold, due to the degree of staining in osteoid tissue and the scattered nature of small areas of osteoid tissue. Basically, bone tissue may not be suitable when combined with osteoid tissue on the way to growth in matured bone. Another problem in this method was how to manage small scattered areas of osteoid tissue less than 100 pixels. Measurements of these areas were excluded from analysis to avoid influencing bone tissue data.
The focal pattern represents radiopacity with uniformity on radiography. This pattern is nearly spherical, so that the perimeter is relatively simple. This corresponds to a low level of complexity. Similarly, osteoid tissues in target and multiconfluent patterns display aspects of both mixed large and small osteoid tissue. Conversely, osteoid tissues in lucent and calcification patterns were relatively small. The lucent pattern shows predominant radiolucency, but with many fine areas of osteoid tissues on histopathological examination. These data may also correspond to measurements of bone complexity, as the perimeter is very large, relatively speaking.
Both lucent and calcification patterns showed high scores for bone complexity. Although levels of bone complexity and distribution were similar, histopathological findings differed considerably in other aspects. However, total of quantities of osteoid tissue formation were similar and radiographic images displayed similarities in terms of scattered small osteoid tissue in the early stages.
Data for the five patterns showed no significant differences between patterns. However, differences in bone complexity should be reflected in density of the radiographic image. From the perspective of the density of radiographic images, reclassification of the decided patterns may be warranted. Classification of FOL patterns should probably be simplified.
Distribution of bone density showed the density of osteoid tissue formations in 25 equal areas of each lesion. Analysis of this method allows observation of scattered osteoid tissue within the lesion. Each area is graded for quality of osteoid tissue. Target, multiconfluent and focal patterns appear relatively centralized, and show a higher degree of calcification than other patterns. Each of the 3 areas (16 marginal areas, 8 middle areas and 1 central area) showed significant differences. However, the rate of calcification is easily changed by various causes.
If all portions display high scores, the lesion represents a hard, bone-like mass, while a low score shows a lesion dominated by fibrous connective tissue. Calcification, focal and lucent patterns showed uniformity of scores between the three different areas, and so may be relatively homogeneous lesions on radiography.
Many methods for analysing bone structure have now been described, including bone mineral density and bone trabecular structural changes.7–24 According to Watanabe et al,7 some of these methods can raise some problems over long intervals due to issues such as observation judgment, reproducibility and binary processing. X-ray bone morphometric analysis for the evaluation of bone trabecular structure is also clinically applicable.7 However, the objective of the present study was mainly to identify differences in the inner condition according to radiographic patterns, as we have studied the diagnosis of internal condition in FOL for a long time. Using a histopathological specimen allows definition of the relationship between bone and fibrous connective tissue. This method may include various problems in the methods of determining threshold levels for binary processing. In the present study, full-section images made from the centre area of histopathological specimens were thought to appear as outlines of expression in each pattern on radiography. However, results from these data are not always satisfactory, but enable analysis and application of three-dimensional imaging prospectively. Further use of these images with both bone and fibrous connective tissue will in future allow clear demonstration of the three-dimensional structural image of lesion patterns. Conversely, these data may suggest simpler modifications for how to perform division into the five patterns under radiographic imaging.
|
Conclusion
|
|---|
Bone complexity appears to be intrinsically related to size, number and complicated outline of osteoid tissue, and distribution broadly corresponded to the size of the osteoid tissue formation. Correlations thus appear to exist between image patterns from radiography and the shape of osteoid tissue on histopathological examination. However, the goals of the present study were not completely fulfilled and different factors will need to be examined in future. Analysis regarding three-dimensional conformations and grade of calcification in an increased number of cases is sorely needed.
|
Acknowledgments
|
|---|
This study was supported by grants from the Dental Research Center at Nihon University School of Dentistry for 2005, and the Sato Fund and Nihon University School of Dentistry for 2006.
|
References
|
|---|
- Waldron CA. Fibro-osseous lesion of the jaws. J Oral Maxillofac Surg 1985;43:249–262.[Medline]
- Waldron CA. Fibro-osseous lesion of the jaws. J Oral Maxillofac Surg 1993;51:828–835.[Medline]
- McDonnell D. Dense bone island. A review of 107 patients. Oral Surg Oral Med Oral Pathol 1994;78:611–620.[CrossRef][Medline]
- Eversole LR, Stone CE, Strub D. Focal sclerosing osteomyelitis/focal periapical osteopetrosis: Radiographic pattern. Oral Surg Oral Med Oral Pathol 1984;58:456–460.[Medline]
- Araki M, Hashimoto K, Matsumoto K, Shinoda K, Komiyama K. Classification of radiographic patterns of fibro-osseous lesions in the jaws. Dent Radiol 2003;43: 121–129.(in Japanese)
- Araki M, Hashimoto K, Matsumoto K, Ezima K, Kawashima S, Matsumoto N, et al. Radiographic patterns of fibro-osseous lesions in the jaws – comparison with histopathological image. Dent Radiol 2005;45: 97–104.(in Japanese)
- Watanabe A, Sakurai T, Kumasaka S, Kashima I. Quantitative analysis of bone trabecular structural change using X-ray bone morphometric analysis: a correlation between structural change and structural parameters. Oral Radiology 2004;20:49–56.
- Matsubara M, Morita S, Shinomiya K, Kawamata R, Nakamura K, Kashima I. Structuring parameters for assessment of bone quality using a morphological filter and star volume analysis: structuring property in the cancellous bone of the human femoral head. J Bone Miner Metab 2003;21:48–56.[Medline]
- Fujikawa T, Sakurai T, Kashima I. Effects of a supplement containing vitamin K2 (MK7) and calcium phosphate (salmon bone powder) on the femur trabecular structure in ovariectomized mice: a study using three-dimensional imaging analysis. Oral Radiology 2004;20:37–43.
- Odgaard A. Three-dimensional methods for quantification of cancellous bone architecture. Bone 1997;20:315–328.
- Feldkamp LA, Goldstein SA, Parfitt AM, Jesion G, Kleerekoper M. The direct examination of three-dimensional bone architecture in vitro by computed tomography. J Bone Miner Res 1989;4:3–11.[Medline]
- Hasegawa Y, Kushida K, Yamasaki K, Inoue T. Volumetric bone mineral density using peripheral quantitative computed tomography in Japanese women. Osteoporos Int 1997;7:195–199.[Medline]
- Ryo JY, Ashman RB, Turner CH. Young's modulus of trabecular and cortical bone material: ultrasonic and microtensile measurement. J Biomech 1993;26:111–119.[CrossRef][Medline]
- Tanaka T, Sakurai T, Kashima I. Structuring of parameter for assessing vertebral bone strength by star volume analysis using a morphological filter. J Bone Miner Metab 2001;29:150–158.
- Uchiyama T, Tanizawa T, Kashima I, Endo N, Takahashi HE, Hara T. A morphometric comparison of trabecular structure of human ilium between microcomputed tomography and conventional histomorphometry. Calcif Tissue Int 1997;61:493–498.[Medline]
- Kumasaka S, Kashima I. Initial investigation of mathematical morphology for the digital extraction of the skeletal characteristics of trabecular bone. Dentomaxillofac Radiol 1997;26:161–168.[Abstract]
- Kumasaka S, Matsuki T, Kashima I. Skeletal pattern extraction of bone trabeculae using mathematical morphology. Oral Radiol 1997;13:35–43.
- Majumdar S, Weinstein RS, Prasad RR. Application of fractal geometry techniques to the study of trabecular bone. Med Phys 1993;20:1611–1619.[Medline]
- Caligiuri P, Giger MI, Favus M. Multifractal radiographic analysis of osteoporosis. Med Phys 1994;21:503–508.[CrossRef][Medline]
- Hahn M, Vogei M, Pompesiu-Kempa M, Delling G. Trabecular bone pattern factor: a new parameter for simple quantification of bone microarchitecture. Bone 13:327–330.
- Vogal M, Hahn M, Delling G. Trabecular bone structure in patients with primary hyperparathyroidism. Virchows Arch 1995;426:127–134.[Medline]
- Whitehouse WJ. The quantitative morphology of anisotropic trabecular bone. J Microsc 1974;10:153–168.
- Goldstein SA, Goulet R, McCubberey D. Measurement and significance of three-dimensional architecture to the mechanical integrity of trabecular bone. Calcif Tissue Int 1993;53:127–133.[Medline]
- Ito M, Ohki M, Nakamura T, Hayashi K, Yamada M, Uetani M, et al. Trabecular texture analysis of CT images in the relationship with spinal fracture. Radiology 1995;194:55–59.[Abstract/Free Full Text]
Top
Abstract
Introduction
