Oral Radiol (2009) 25:129–134 DOI 10.1007/s11282-009-0024-y
ORIGINAL ARTICLE
Converting conventional radiographic examination data of trabecular bone pattern values into density measurement values using intraoral digital images Menik Priaminiarti Æ Budi Utomo Æ R. Susworo Æ Hanna Bachtiar Iskandar
Received: 25 November 2008 / Accepted: 31 July 2009 / Published online: 9 September 2009 Ó Japanese Society for Oral and Maxillofacial Radiology and Springer 2009
Abstract Objectives To determine the conversion value of grayscale density measurements from intraoral conventional radiographic examinations of the edentulous maxilla and mandible using intraoral digital radiography. Methods Periapical radiography examinations with a pararelling technique, both conventional and digital, were performed on 18 male and 34 female patients with edentulous maxillas and mandibles. The trabecular bone pattern of 42 maxillary and 61 mandibular regions of interest (ROIs) was classified into five grades. Grayscale density measurements were made within a marked area of the ROI in the image of the periapical digital radiograph in the same corresponding trabecular region. To obtain conversion values, including the effects of age, gender, and region of the jaw, an analysis was made to develop regression equations. Results The kappa value for intra- and interobserver differences was 0.71–0.85. The strength of the radiographic conventional value to predict the grayscale density measurement of digital radiography was gained from the regression analysis, with R2 = 0.75–0.8. The regression B. Utomo Department of Population and Biostatistics, Faculty of Public Health, University of Indonesia, Jakarta, Indonesia R. Susworo Department of Radiotherapy, Faculty of Medicine, University of Indonesia, Jakarta, Indonesia M. Priaminiarti (&) H. B. Iskandar Department of Dental and Maxillofacial Radiology, Faculty of Dentistry, University of Indonesia, Jl.Salemba Raya No.4, Jakarta 10430, Indonesia e-mail:
[email protected];
[email protected]
equation for the maxilla and the mandible were separated, and the age, gender, and region of the jaws were included. Conclusions Conventional intraoral radiographic values of the trabecular bone pattern can be converted to values of grayscale density measurements from intraoral digital radiography. The regression equation for the conversion was obtained by including the effects of age, gender, and region of the jaw. Keywords Conversion value Density measurement Intraoral conventional radiography Digital radiography Regression equation
Introduction Conventional radiographic examinations have several advantages and uses, one of the most important being costeffectiveness [1]. In Indonesia, the availability of modern imaging modalities is still very limited, and radiograph examinations applying conventional methods, particularly intraoral periapical radiography, are regularly used by most dentists. Maxillary and mandibular periapical radiographs provide images with superior resolution and sharpness; they also possess significant amounts of information for the evaluation of bone quality, including bone density, the amount of cortical bone, and the amount and pattern of trabeculations in trabecular bone [1, 2]. The trabecular bone pattern is one aspect representing bone quality, and is important in the diagnosis and treatment planning in many dental procedures, such as the treatment of implants, periodontal diseases, and systemic diseases with oral and maxillofacial manifestation, such as osteoporosis [1–3]. Digital radiography, however, requires less radiation [2, 4].
123
130
This study evaluated the correlation between conventional and digital radiographic examinations. The density measurements in digital images were obtained in a numerical format, in the form of 256 grayscale values for each pixel [5, 6]. This quantization process corresponds to the radiographic image, and represents the measurements of photon intensities in each pixel [5]. The quantization process of the radiographic image in digital radiography can increase the utility of radiographic information for diagnostic procedures [7]. Efforts are needed to increase the utility of diagnostic information obtained from conventional radiographic images, so that it approaches that of digital images. Because of this research, despite the limited availability of digital radiography facilities in Indonesia, better diagnostic information on bone density can be obtained from conventional radiographic equipment. While visual examination of bone density can provide categorical grading classifications, digitization can provide numerical density measurements. Such numeric values have higher diagnostic utility and can be used in further research. For example, such data can be applied to estimate normal jaw bone densities for a certain age of women in Indonesia, which later can be associated with osteoporosis or with long-term evaluations of dental implant treatments.
Methods Patients This research passed the Ethical Clearance Examination conducted by the Board of Ethical Clearance of the Faculty of Dentistry, University of Indonesia. All patients provided written informed consent. The subjects were patients with edentulous regions who came to the clinic of the Department of Prosthodontics, Faculty of Dentistry, University of Indonesia, and intended to undergo prosthodontic treatment. The subjects were evaluated and considered to be normal and not to have any condition that might influence bone state, such as periodontal disease or any other systemic or bone disease. In total, 103 images of maxillary and mandibular intraoral periapical radiographs, by both digital and conventional radiography, were obtained from 52 patients (18 men and 34 women; age range 20–70 years). The 103 radiographs consisted of 42 maxillary and 61 mandibular images from both conventional and digital examinations.
Oral Radiol (2009) 25:129–134
(Belray, Tokyo, Japan), exposed at 70 kVp, 15 mAs, 0.27 s for the anterior region, and 0.33 s for the posterior region. Intraoral radiographic film (Kodak InsightÒ, size IP-21, F Speed; Kodak, Rochester, NY) was used, combined with a paralleling cone indicator device (PCID; HanshinÒ CID-3; Hanshin Technical Laboratory, Nishinomiya, Japan) for performing the intraoral paralleling technique. For image reproducibility purposes in comparing the conventional and digital radiographs of the same area in the edentulous region, a polyether bite registration device (RamitecTM, 3M ESPE; Seefeld, Germany) was placed inside the bite plate in the PCID (Fig. 1). Exposed films were then processed manually using developer and fixer (KM-DOL X and KM-FIX; Konica Minolta, Tokyo, Japan). The time used for manual developing was 4 min at 20°C; the volume of the developing tank was 5 L. All films were processed on the same day, following the manufacturer’s instructions. Evaluation of trabecular bone pattern All radiographs were examined by two dental radiologists with 10–20 years experience in interpreting periapical radiographs. The radiographs were viewed on a masked view box, so that ambient light around the radiographs was removed. The reliability in determining the classification of trabecular bone pattern was assessed by reading all the images randomly using the light box at 59 magnification. The observers read all 103 radiographs three times at 1-week intervals from 0900 to 1000 hours, and the identity of all radiographs was unknown to the reviewers. The assessment of the radiographs was determined on the basis of trabecular bone patterns in the region of interest (ROI)
Intraoral periapical conventional radiographic examination Intraoral periapical conventional radiographs were obtained using a Belmont Belray Long Cone model 096
123
Fig. 1 Polyether bite registration material (RamitecTM 3M ESPE) was placed inside the bite plate of PCID
Oral Radiol (2009) 25:129–134
131
and was classified into five grades. The area of the ROI was constant at 8 9 10 mm. The ROI was predetermined in the area of edentulous alveolar bone (Fig. 2). The observers used the written criteria for classification and reference images set out by Taguchi [8, 9]. The classification grades were grade 1, very sparse; grade 2, sparse; grade 3, normal; grade 4, dense; and grade 5, very dense. Digital radiographic examination Intraoral periapical digital images were acquired using new phosphor storage plates (PSP; DigoraÒ; Soredex, Helsinki, Finland), standard size, number 2. The exposures were performed using a Belmont Belray Long Cone model 096 (Belray). The exposure factors were 70 kVp, 15 mAs, and 0.18 s for the anterior region, and 0.27 s for the posterior region, in the same area of the edentulous region, using the same PCID with bite registration as used in the conventional radiographic examination. Each plate was scanned immediately following the exposure using a DigoraÒ Soredex reader. Figure 3 shows how the images were evaluated using the Digora software (Digora for Windows 1.51) and the mean image density was obtained using a software tool for density measurements within the marked area in the ROI in the same corresponding trabecular region as in the conventional radiographs. Analysis Objectivity in determining the classification of trabecular bone patterns between the two observers was assessed by evaluating the intra- and interobserver agreement using the kappa index. The interpretation of the kappa index value was based on the 6-point scale of Altman: \0.20, poor;
Fig. 3 The mean image density was obtained using a software tool for density measurements (gray scale) within the marked area in the same corresponding region of interest with the previous conventional radiographic examination
0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, good; and 0.81–1.00, very good. In addition to making a correlation, a conversion from the grading classification of the radiological findings based on the visual examination of bone density to the digital value of the density measurement from intraoral digital radiography was made. Then, a regression analysis was performed to assess the correlation between the value from the conventional radiographic examination and that from the digital image in determining the trabecular bone pattern. The analysis to obtain the regression equation for predicting the digital value was conducted by including other factors (jaw area, gender, age) that were believed to have an effect in predicting digital values based on conventional radiograph values. An analysis of variance (ANOVA) was performed to evaluate the consistency and reliability of the analysis, and also to analyze the variance for the difference of the mean of the absolute residual value from the result of the regression analysis.
Results
Fig. 2 Region of interest (ROI) in a conventional periapical radiograph, as seen inside the rectangular region in the edentulous area
The subjects in this study were patients with tooth loss who came to the prosthodontics clinic to obtain dental prostheses. Of all the patients seen over a year, 52 were willing to join the study. The age of the subjects ranged from 20 to 70 years; women (65.4%) outnumbered men (34.6%); and the largest age group was between 20 and 39 years of age (44.2%). Data for all subjects (52 patients and 103 pairs of conventional and digital radiographs of the edentulous area)
123
132
Oral Radiol (2009) 25:129–134
are given in Table 1. Table 1 shows that a larger number of the radiographs were from the lower jaw area and in the 20- to 39-year-old group. The results regarding agreement between the two observers in interpreting the radiographs are presented in Tables 2 and 3. Table 2 gives the intraobserver agreement, with the kappa value ranging from 0.71 to 0.75, and Table 3 shows that the kappa value for the interobserver agreement ranged from 0.70 to 0.85. In performing regression analyses to assess the correlation between the values from conventional radiographic examinations and the values from digital images in determining the trabecular bone pattern, a simple regression analysis was carried out and the result showed that the value of R2 was higher when the analysis was conducted separately for the maxilla and mandible. To evaluate the consistency and reliability of the analysis, the difference of the mean of the absolute residual value was analyzed and an ANOVA was performed. The results showed that in all interpretations, the P value was [0.05. In addition to making a correlation, a conversion from the grading classification of radiological findings based on visual examinations of bone density into the digital values of density measurements from intraoral digital radiographs was made. Age, gender, and jaw area were considered to be factors that affected bone density and were included in the regression analysis. The regression analysis results are shown in Table 4. Based on this analysis, a regression equation was obtained. The following regression equations were obtained for the maxilla and mandible. In the equations, 92 indicates the region of the jaw (anterior = 0; posterior = 1), 93 denotes the gender (female = 0; male = 1), and 94 is the age in years. Maxilla Digital value of the trabecular bone pattern ¼ 41:21 þ 35:56 ðconventional radiographic examination valueÞ þ 1:36 2 þ ð12:29Þ 3 þ ð0:32Þ 4
Mandible Digital value of the trabecular bone pattern ¼ 54:40 þ 25:84 ðconventional radiographic examination valueÞ þ ð5:12Þ 2 þ 3:55 3 þ ð0:13Þ 4 Discussion To obtain reliable values from conventional radiographs for classifying trabecular bone patterns, interpretation was performed by two observers who conducted the assessment randomly at three separate times with a 1-week interval between readings. Tables 2 and 3 show the intra- and interobserver agreement. The kappa value was 0.71–0.75 for the intra-observer agreement and 0.71–0.85 for the interobserver agreement. According to Altman, kappa values of 0.61–0.80 and 0.81–1.00 show good and very good agreement, respectively [10]. Thus, the intra- and interobserver agreement was high, and these results were used for the regression analysis to assess the correlation between the values from conventional radiographic examinations and the values from digital images in determining the trabecular bone pattern. The value of R2 indicates the coefficient of determination, which can explain the predicting power of the regression equation: the higher the value of R2, the higher the predicting power that can be obtained [10]. The result of the simple regression analysis was higher when the analysis was conducted separately for the maxilla and mandible. This can be explained based on the fact that, anatomically, the quality of the maxilla and mandible images is different [1, 2, 11]. The consistency and reliability of the regression analysis was also assessed, and an ANOVA of the mean of the absolute residual value was performed. The results from all interpretations showed that the P values were [0.05, indicating no significant difference between the interpretations in correlating conventional radiograph values with digital values. This result also indicates that conventional
Table 1 Distribution of the number of radiographs based on jaw area and age group Jaw area
Age group
Total
20–39 years old Number
40–49 years old %
Number
%
C50 years old %
Number
%
Maxilla anterior
8
7.8
3
2.9
12
11.6
23
22.3
Maxilla posterior
7
6.8
5
4.9
7
6.8
19
18.5
Mandible anterior
3
25.2
3
2.9
5
4.9
11
10.7
Mandible posterior Total
23 41
22.3 39.8
16 27
15.5 26.2
11 35
10.7 34.0
50 103
48.5 100.0
123
Oral Radiol (2009) 25:129–134
133
Table 2 Intra-observer agreement of two observers in determining the classification of trabecular bone pattern of 103 edentulous regions in the maxilla and mandible on three separate occasions Observer A
Observer B
First interpretation
Second interpretation
Third interpretation
First interpretation
Second interpretation
Third interpretation
First interpretation
–
0.72
0.71
–
0.75
0.73
Second interpretation
0.72
–
0.75
0.75
–
0.74
Third interpretation
0.71
0.75
–
0.73
0.74
–
Table 3 Interobserver agreement of two observers in determining the classification of the trabecular bone pattern of 103 edentulous regions in the maxilla and mandible Observer A
Observer B First interpretation
Second interpretation
Third interpretation
0.85
0.78
0.73
Second interpretation
0.76
0.80
0.80
Third interpretation
0.74
0.70
0.71
First interpretation
Table 4 Converting the conventional regression value of the trabecular bone pattern to the digital value by entering influencing factors R2 Maxilla Mandible Maxilla and mandible CRE
0.75
0.66
CRE ? ja
0.75
0.67
0.64 0.65
CRE ? ja ? gen
0.77
0.67
0.66
CRE ? ja ? gen ? age 0.80
0.67
0.66
CRE Conventional radiographic examination, ja jaw area (region of the jaw), gen gender
radiograph values have a predicting value that correlates with the digital values. The regression equation for predicting the digital value of the trabecular bone pattern was obtained by performing a regression analysis of conventional radiographic examination data and the values from digital images, which included other factors (jaw area, gender, age) that were considered to influence the prediction values. By doing so, the results of the analysis showed a higher R2 value, indicating the influence of these factors in the prediction power of the equation. The inclusion of these factors was based on the knowledge that bone condition is influenced by region of the jaw, gender, and age [12]. Moreover, the bone condition changes and has a tendency to resorb along with increasing age [13, 14]. From an anatomical point of view,
differences exist between the maxilla and mandible. For example, differences in the density and ridge dimensions of the maxilla and mandible can be observed after tooth loss [12]. Hormonal condition could affect the metabolism of bone, which in turn results in different bone conditions between males and females [15, 16]. Thus, gender is another factor that should be included. Based on these results, the regression equation for digital value prediction from conventional radiograph values in determining the trabecular bone pattern was obtained separately for the maxilla and mandible. Both analyses included jaw area, gender, and age as variables in the equations. The results of this study are expected to be of value for research in Indonesia because of the limited digital radiography facilities and other advance imaging modalities. Thus, in the future, performing further research using conventional radiographic equipment with better diagnostic information regarding bone density will be possible.
Conclusions Conventional intraoral radiographic values of the trabecular bone pattern can be converted to the grayscale density measurement values of intraoral digital radiography. The regression equation for the conversion was obtained separately for the maxilla and mandible by including region of the jaw, gender, and age as factors. With this regression equation to predict the digital image value in determining
123
134
the trabecular bone pattern from conventional radiographic examinations, the limitations of the human eye’s capacity for differentiating the gradation in the density spectrum from the conventional radiographic image can be resolved. Diagnostic information from conventional radiographic images can be more accurately compared to data obtained from grading conventional radiographs by eye.
Oral Radiol (2009) 25:129–134
8.
9.
10. 11.
References 1. White SC, Pharoah MJ. Oral radiology: principles and interpretation. 5th ed. St. Louis: Mosby; 2004. pp. 71–120, 225–77, 314– 29, 677–91. 2. Whaithes E. Essentials of dental radiography and radiology. 3rd ed. London: Churchill Livingstone; 2007. p. 289–97. 3. Mupparapu M, Sinner SR. Implant imaging for the dentist. J Can Dent Assoc. 2004;70(1):32. 4. Berkhout WER, Sanderink GCH, van der Stelt PF. Does digital radiography increase the number of intraoral radiographs? A questionnaire study of Dutch dental practices. Dentomaxillofac Radiol. 2003;32:124–7. 5. van der Stelt PF. Filmless imaging: the uses of digital radiography in dental practice. JADA. 2005;136:1379–87. 6. Parks ET, Williamson GF. Digital radiography: an overview. J Contemp Dent Pract. 2002;3(4):23–39. 7. Yoshiura K, Welander U, David WM, Li G, Shi X-Q, Nakayama E, et al. Comparison of the psychophysical properties of various
123
12. 13.
14.
15.
16.
intraoral film and digital systems by means of the perceptibility curve test. Dentomaxillofac Radiol. 2004;33:98–102. Taguchi A, Tanimoto K, Akagawa Y, Wada T, Rohlin M. Trabecular bone pattern of the mandible: comparison of panoramic and CT. Dentomaxillofac Radiol. 1997;26:85–9. Taguchi A, Tanimoto K, Suei Y, Otani K, Wadamoto M, Akagawa Y, et al. Observer agreement in the assessment of mandibular trabecular bone pattern from panoramic radiographs. Dentomaxillofac Radiol. 1997;26:90–4. Altman DG. Practical statistic for medical research. 1st ed. London: Chapman and Hall; 1991. p. 403–9. Ulm C, Tepper G. Structure of atrophic alveolar bone. In: Georg W, editor. Implants in qualitatively compromised bone. London: Quintessence; 2004. p. 29–31. Misch CE. Contemporary implant dentistry. 2nd ed. St. Louis: Mosby; 1999. p. 73–118. Nanci A, Whitson SW, Blanco P. Bone. In: Nanci A, editor. Ten Cate’s oral histology, development, structure, and function. 6th ed. St. Louis: Mosby; 2003. p. 111–44. Neukam FW, Kloss FR. Compromised jawbone quantity and its influence on oral implant placement. In: Zarb G, Lekholm U, Albrektsson T, Tenenbaum H, editors. Aging, osteoporosis and dental implants. Chicago: Quintessence; 2002. p. 67–84. Mundy GR. Bone resorbing cells. In: Favus MJ, editor. Primer on metabolic bone diseases and disorders of mineral metabolism. 2nd ed. New York: Raven Press; 1993. p. 25–32. Prentice A, Schoenmakers I, Laskey MA, de Bono S, Ginty F, Goldberg GR. Nutrition and bone growth and development. Proc Nutr Soc. 2006;65(4):348–60.
