Home About us Editorial board Search Ahead of print Current issue Archives Submit article Instructions Contacts Login 
Home Print this page Email this page Users Online: 197



 
 Table of Contents  
ORIGINAL ARTICLE
Year : 2015  |  Volume : 3  |  Issue : 2  |  Page : 39-43

Fractal dimension of the mandibular trabecular bone measured on digital and digitized images


1 Department of Oral Diagnosis and Surgery, Araraquara Dental School, São Paulo State University - UNESP, Araraquara, SP, Brazil
2 Department of Public Health, Araraquara Dental School, São Paulo State University - UNESP, Araraquara, SP, Brazil

Date of Web Publication22-May-2015

Correspondence Address:
Dr. Matheus Lima Oliveira
Rua Humaitá 1680, 14801-903 - Araraquara - SP
Brazil
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/2321-3841.157519

Rights and Permissions
  Abstract 

Aims: This study compared fractal dimension (FD) values on mandibular trabecular bone in digital and digitized images at different spatial and contrast resolutions. Materials and Methods: 12 radiographs of dried human mandibles were obtained using custom-fabricated hybrid image receptors composed of a periapical radiographic film and a photostimulable phosphor plate (PSP). The film/PSP sets were disassembled, and the PSPs produced images with 600 dots per inch (dpi) and 16 bits. These images were exported as tagged image file format (TIFF), 16 and 8 bits, and 600, 300 and 150 dpi. The films were processed and digitized 3 times on a flatbed scanner, producing TIFF images with 600, 300 and 150 dpi, and 8 bits. On each image, a circular region of interest was selected on the trabecular alveolar bone, away from root apices and FD was calculated by tile counting method. Two-way ANOVA and Tukey's test were conducted to compare the mean values of FD, according to image type and spatial resolution (α = 5%). Results: Spatial resolution was directly and inversely proportional to FD mean values and standard deviation, respectively. Spatial resolution of 150 dpi yielded significant lower mean values of FD than the resolutions of 600 and 300 dpi (P < 0.05). A nonsignificant variability was observed for the image types (P > 0.05). The interaction between type of image and level of spatial resolution was not significant (P > 0.05). Conclusion: Under the tested, conditions, FD values of the mandibular trabecular bone assessed either by digital or digitized images did not change. Furthermore, these values were influenced by lower spatial resolution but not by contrast resolution.

Keywords: Digital dental radiography, fractals, mandible, X-ray film


How to cite this article:
Oliveira ML, Saraiva JA, Scaf G, Monteiro Loffredo LC, Tosoni GM. Fractal dimension of the mandibular trabecular bone measured on digital and digitized images. J Oral Maxillofac Radiol 2015;3:39-43

How to cite this URL:
Oliveira ML, Saraiva JA, Scaf G, Monteiro Loffredo LC, Tosoni GM. Fractal dimension of the mandibular trabecular bone measured on digital and digitized images. J Oral Maxillofac Radiol [serial online] 2015 [cited 2021 Sep 27];3:39-43. Available from: https://www.joomr.org/text.asp?2015/3/2/39/157519


  Introduction Top


Radiographic digital imaging is a dynamic process, in which the operator is allowed to manipulate the images by means of postprocessing methods. Several advantages of digital radiography over film-based imaging modalities have been described in the scientific literature, [1] although analog film has proved to have similar diagnostic accuracy for measurements, [2] density analysis, [3] and caries [4] and external root resorption [5] detection. Financial issues seem to be the main reason why digital systems have not been used worldwide yet and despite the fact that analog imaging methods are still being used and tend to disappear in the future, film-based radiographs will be inevitably present on dental practitioners' archives for years.

When a method of digital analysis is required from a film-based radiograph, it must undergo a digitization process. This process uses an indirect method of digital image acquisition and may result in loss of quality and information for the analysis of the trabecular bone image. [6],[7] The quality of a digital image is determined by factors such as spatial resolution and contrast resolution. The spatial resolution defines the degree of sharpness or detail, and it may be represented by dots per inch (dpi) or by line pairs per millimeter. The contrast resolution or bit depth defines the range of shades of gray in the radiographic image such as 8, and 16 bits that represent 256 and 65536 shades, respectively. Digital images with higher adjustments for better quality presents bigger file size, which is an important factor in the process of communication because smaller sized images are transmitted to distant locations faster and require less space when stored on computers' hard drive.

Fractal analysis is a mathematical method that numerically describes complex shapes and structural patterns as a fractal dimension (FD), and has been used to assess the pattern of the trabecular bone in digital imaging. [8],[9],[10],[11] Although the density is crucial to determine bone strength and fracture risk, the complexity of the trabecular bone is also important in the diagnosis of bone quality. In clinical practice, information about the trabecular bone arrangement is valuable for implant treatment planning and in case of systemic diseases such as osteoporosis. [12] Therefore, many attempts have been made to analyze and predict the structure of the trabecular bone by means of fractal analysis, [13],[14],[15],[16],[17],[18] given that this analysis may show changes not always apparent in the radiographic image. [13],[19] There is evidence in the literature to suggest that different methods of digital image capturing may explain the discrepancy in the results of fractal analysis of the trabecular bone. [9],[14]

At this point, little is known about the effect of digitized images on the calculation of FD in the mandibular trabecular bone, and further, if imaging factors, such as spatial resolution, can change this estimation. Thus, considering the possible need of comparing current digital images with analog-film radiographs from the patients' records, the aim of this study was to compare FD values on mandibular trabecular bone in digital and digitized images at different spatial and contrast resolutions.


  Materials and Methods Top


This experimental protocol was designed according to the local Institutional Research Ethics Committee and carried out after its approval (protocol #09/2007).

A total of 12 hybrid image receptors composed of a periapical radiographic film size 2 (Kodak InSight, Carestream Dental, USA) and a photostimulable phosphor plate (PSP) number 2 (DenOptix, Gendex Dental Systems, Italy) were custom-fabricated. In a darkroom, the lead foil of each film's packet was removed, and the films were sealed with black adhesive tape to prevent light exposure. The films were then placed over the PSPs, forming film/PSP sets (dual image receptor), which were used for all radiographic exposures.

A total of 12 periapical radiographs were obtained from dried human mandibles, in regions with at least one tooth, using the X-ray unit Gendex-765 DC (Des Plaines, USA) at 65 kVp, 0.7 mAs and a focus-receptor distance of 40 cm, and the central X-ray beam focusing at a right angle to the object and receptor. The dual image receptor was positioned with its long axis vertical and parallel to the teeth for all regions, and was fixed with adhesive tape and utility wax. A 10-step aluminum wedge was placed on the upper end of each dual image receptor, and an acrylic plate (15-mm thick) was used as soft tissue simulation.

The dual image receptors were disassembled and the PSPs were processed in the DenOptix scanner (Gendex Dental Systems, Italy), which was adjusted to produce 600-dpi and 16-bit images, using the imaging software VixWin PRO (Gendex Dental Systems). In addition, these images were exported as tagged image file format (TIFF) and at different contrast resolutions (16 and 8 bits) and spatial resolutions (600, 300 and 150 dpi), forming six groups of direct digital images.

The films were automatically processed using the Dent-X 9000 unit (Dent-X, NY, USA) set at 4 min and 30 s. Subsequently, conventional radiographs were digitized on a flatbed scanner (model 1236, Agfa, Belgium), using the dedicated software Agfa FotoSnap 32V3.00.05 as an interface for acquisition and archiving the scanned images. The films were digitized 3 times with different spatial resolutions (600, 300 and 150 dpi) and the only available contrast resolution (8 bits), and stored as TIFF.

Considering that the scanner used in this study only produces 8-bit images, for methodological reasons, the contrast resolution was combined with scanning methods (digitized and digital) and referred to as an image type. Thus, the 12 radiographic exposures produced a total of 108 images, which were divided in 9 groups according to spatial resolution (600, 300 and 150 dpi) and image type (digital 8-bit, digital 16-bit and digitized 8-bit).

All images were analyzed using ImageJ version 1.32j, public domain software developed by the National Institutes of Health (Bethesda, MD, USA). On each image, a circular region of interest (ROI) was randomly selected on the trabecular alveolar bone, away from root apices. The pixel coordinates (location) of each ROI were recorded to allow the same location in other corresponding ROI from another image group [Figure 1].
Figure 1: Region of interest selected in periapical images of the same anatomical location according to image type (a: Digital, 8-bit; b: Digital, 16-bit; c: Digitized, 8-bit)


Click here to view


Before the calculation of FD, the image of each ROI was processed according to the method previously described by White and Rudolph. [20] Briefly, to remove the large-scale variations in image brightness caused by the difference in the thickness of the object and the radiation exposure, the ROI image was blurred with a Gaussian filter (sigma = 35 pixels). This step removes all fine-scale and medium-scale structure and retains only large variations in density (low-pass filtering). The blurred image was then subtracted from the original image, and 128 were added to the result at each pixel location. Thereafter, the image was converted into binary (black and white) with threshold corresponding to the median of the histogram [Figure 2]a. The binary image was eroded and dilated once to reduce noise. The image of the trabeculae was then inverted to make the trabeculae more apparent and then skeletonized, which means, eroded until only the central line of pixels remains [Figure 2]b. This image was then stored in TIFF format. [Figure 2]c shows the superimposition of the skeletonized images on the original ROI to demonstrate that the skeletal structure studied corresponded to the trabeculae of the original image.
Figure 2: Region of interest at different steps of the image processing: (a) Binarization; (b) skeletonization; (c) superimposition


Click here to view


The FDs were calculated by tile counting method. When log (number of tiles occupied) is plotted against log (length of the tile), a straight line is obtained with slope S. Then FD = 1-S with 1< FD <2. [9] Two-way ANOVA and Tukey's test were conducted to compare the mean values of FD, according to image type and spatial resolution. The level of significance was set at 5%.


  Results Top


[Table 1] shows that spatial resolution was directly and inversely proportional to FD mean values and standard deviation (SD), respectively. For instance, lower dpi values yielded lower FD mean values and greater SD. Very subtle difference is observed on the mean values and SD of FD when comparing different image types with the same spatial resolution.
Table 1: Mean values and SD of FD according to image type and spatial resolution (in dpi) (n = 108; 12 per group)


Click here to view


[Table 2] shows a significant variability in relation to spatial resolution (P < 0.0001), which are different resolution levels changed the mean values of FD. On the other hand, it could be observed a nonsignificant variability for the image types (P > 0.05). The interaction between type of image and spatial resolution level (T × R) was not significant (P > 0.05).
Table 2: Two-way ANOVA for FD estimations according to different image types (T) and spatial resolutions (R)


Click here to view


Spatial resolution of 150 dpi yielded significant lower mean values of FD than the resolutions of 300 and 600 dpi (P < 0.05). Indeed, these two resolutions had similar behavior of the mean FD values that were, respectively, 1.7798 and 1.8054.


  Discussion Top


Radiographic digital imaging has shown numerous advantages over film-based radiographic images since its introduction to the dental market. Over the years, dentists have witnessed a steady evolution of intraoral imaging systems, which have had a significant increase of image quality and convenience of use.

The digitization process of film-based radiographic images has demonstrated similar accuracy to digital imaging for several diagnostic tasks, [2],[3],[4],[5] however there is no other study in the scientific literature comparing both scanning methods with regard to FD. Our study showed that the FD did not differ between different image types, providing evidence that the process of digitization of the conventional film does not result in loss of quality and information for FD estimation.

The results of our study differ with those of Pornprasertsuk et al. [21] who found that these two types of imaging receptors could significantly affect the calculation of FD when measuring the FD in bones of rats with film and charge-coupled device sensor. These results cannot be directly compared to ours since we made use of different types of bone, imaging receptors and fractal analysis methods. There are several methods for determining the FD, which may lead to different results. [8] In this study, we used the tile counting method that has also been used by other authors [11],[15],[22],[23] to show changes in the trabecular bone.

Another reason that may partly explain this inconsistency is the fact that, in the study of Pornprasertsuk et al., [21] there was no geometric standardization of X-ray exposures between the two types of image receptors and the information in the literature is controversial regarding geometric projection, which means that it may or may not affect the FD estimation. [8],[24],[25],[26] Differently from that study, and aiming to eliminate or reduce possible interference caused by radiographic exposure factors, we chose a methodology similar to Janhom et al. [27] and conducted simultaneous exposure of the film and the phosphor plate. Thus, the results of this study support the hypothesis that, in longitudinal studies, film-based images kept in archives can be digitized and compared to direct digital images in order to evaluate bone quality.

The final quality of the radiographic image may be affected by inherent factors of the scanning process, [6],[7] such as spatial resolution and contrast resolution. In the present study, our results showed that the scanning process does not change the FD values only when spatial resolution is the same. All images were obtained with spatial resolution of 600, 300 and 150 dpi, which represents a pixel size of approximately 42, 84 and 169 µm, respectively. Images with higher spatial resolution were able to capture more details and/or complexity of the trabecular bone, producing higher FD values. Considering that images with spatial resolution >300 dpi are considered to be more appropriate for interpretation and diagnosis, [6] spatial resolution lower than 300 dpi is not recommended for the calculation of FD in the mandibular trabecular bone. Regarding contrast resolution, we found that this factor did not affect the calculation of FD and studies comparing contrast resolution and FD of the trabecular bone could not be found in the literature. Considering that the file size, in bytes, of a 16-bit image may be twice the size of an 8-bit image, the first should not be recommended for FD analysis.


  Conclusion Top


Fractal dimension values of the mandibular trabecular bone assessed either by digital or digitized images did not change. Furthermore, these values are influenced by lower spatial resolution but not by contrast resolution.

 
  References Top

1.
Farman AG, Levato CM, Gane D, Scarfe WC. In practice: How going digital will affect the dental office. J Am Dent Assoc 2008;139 Suppl:14S-9.  Back to cited text no. 1
    
2.
Ravi V, Lipee P, Rao CV, Lakshmikanthan L. Direct digital radiography versus conventional radiography - Assessment of visibility of file length placed in the root canal: An in vitro study. J Pharm Bioallied Sci 2012;4:S285-9.  Back to cited text no. 2
    
3.
Kamburoglu K, Gülsahi A, Genç Y, Paksoy CS. A comparison of peripheral marginal bone loss at dental implants measured with conventional intraoral film and digitized radiographs. J Oral Implantol 2012;38:211-9.  Back to cited text no. 3
    
4.
Abesi F, Mirshekar A, Moudi E, Seyedmajidi M, Haghanifar S, Haghighat N, et al. Diagnostic accuracy of digital and conventional radiography in the detection of non-cavitated approximal dental caries. Iran J Radiol 2012;9:17-21.  Back to cited text no. 4
    
5.
Mesgarani A, Haghanifar S, Ehsani M, Yaghub SD, Bijani A. Accuracy of conventional and digital radiography in detecting external root resorption. Iran Endod J 2014;9:241-5.  Back to cited text no. 5
    
6.
Janhom A, van Ginkel FC, van Amerongen JP, van der Stelt PF. Scanning resolution and the detection of approximal caries. Dentomaxillofac Radiol 2001;30:166-71.  Back to cited text no. 6
    
7.
Parissis N, Kondylidou-Sidira A, Tsirlis A, Patias P. Conventional radiographs vs digitized radiographs: Image quality assessment. Dentomaxillofac Radiol 2005;34:353-6.  Back to cited text no. 7
    
8.
Ruttimann UE, Webber RL, Hazelrig JB. Fractal dimension from radiographs of peridental alveolar bone. A possible diagnostic indicator of osteoporosis. Oral Surg Oral Med Oral Pathol 1992;74:98-110.  Back to cited text no. 8
    
9.
Geraets WG, van der Stelt PF. Fractal properties of bone. Dentomaxillofac Radiol 2000;29:144-53.  Back to cited text no. 9
    
10.
Bollen AM, Taguchi A, Hujoel PP, Hollender LG. Fractal dimension on dental radiographs. Dentomaxillofac Radiol 2001;30:270-5.  Back to cited text no. 10
    
11.
Shrout MK, Jett S, Mailhot JM, Potter BJ, Borke JL, Hildebolt CF. Digital image analysis of cadaver mandibular trabecular bone patterns. J Periodontol 2003;74:1342-7.  Back to cited text no. 11
    
12.
Wilding RJ, Slabbert JC, Kathree H, Owen CP, Crombie K, Delport P. The use of fractal analysis to reveal remodelling in human alveolar bone following the placement of dental implants. Arch Oral Biol 1995;40:61-72.  Back to cited text no. 12
    
13.
Heo MS, Park KS, Lee SS, Choi SC, Koak JY, Heo SJ, et al. Fractal analysis of mandibular bony healing after orthognathic surgery. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2002;94:763-7.  Back to cited text no. 13
    
14.
Tosoni GM, Lurie AG, Cowan AE, Burleson JA. Pixel intensity and fractal analyses: Detecting osteoporosis in perimenopausal and postmenopausal women by using digital panoramic images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2006;102:235-41.  Back to cited text no. 14
    
15.
Yu YY, Chen H, Lin CH, Chen CM, Oviir T, Chen SK, et al. Fractal dimension analysis of periapical reactive bone in response to root canal treatment. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107:283-8.  Back to cited text no. 15
    
16.
Corpas Ldos S, Jacobs R, Quirynen M, Huang Y, Naert I, Duyck J. Peri-implant bone tissue assessment by comparing the outcome of intra-oral radiograph and cone beam computed tomography analyses to the histological standard. Clin Oral Implants Res 2011;22:492-9.  Back to cited text no. 16
    
17.
Sánchez I, Uzcátegui G. Fractals in dentistry. J Dent 2011;39:273-92.  Back to cited text no. 17
    
18.
Oliveira ML, Pedrosa EF, Cruz AD, Haiter-Neto F, Paula FJ, Watanabe PC. Relationship between bone mineral density and trabecular bone pattern in postmenopausal osteoporotic Brazilian women. Clin Oral Investig 2013;17:1847-53.  Back to cited text no. 18
    
19.
Jett S, Shrout MK, Mailhot JM, Potter BJ, Borke JL. An evaluation of the origin of trabecular bone patterns using visual and digital image analysis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2004;98:598-604.  Back to cited text no. 19
    
20.
White SC, Rudolph DJ. Alterations of the trabecular pattern of the jaws in patients with osteoporosis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1999;88:628-35.  Back to cited text no. 20
    
21.
Pornprasertsuk S, Ludlow JB, Webber RL, Tyndall DA, Yamauchi M. Analysis of fractal dimensions of rat bones from film and digital images. Dentomaxillofac Radiol 2001;30:179-83.  Back to cited text no. 21
    
22.
Demirbas AK, Ergün S, Güneri P, Aktener BO, Boyacioglu H. Mandibular bone changes in sickle cell anemia: Fractal analysis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2008;106:e41-8.  Back to cited text no. 22
    
23.
Updike SX, Nowzari H. Fractal analysis of dental radiographs to detect periodontitis-induced trabecular changes. J Periodontal Res 2008;43:658-64.  Back to cited text no. 23
    
24.
Shrout MK, Potter BJ, Hildebolt CF. The effect of image variations on fractal dimension calculations. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1997;84:96-100.  Back to cited text no. 24
    
25.
Chen SK, Chen CM. The effects of projection geometry and trabecular texture on estimated fractal dimensions in two alveolar bone models. Dentomaxillofac Radiol 1998;27:270-4.  Back to cited text no. 25
    
26.
Jolley L, Majumdar S, Kapila S. Technical factors in fractal analysis of periapical radiographs. Dentomaxillofac Radiol 2006;35:393-7.  Back to cited text no. 26
    
27.
Janhom A, van der Stelt PF, van Ginkel FC. Interaction between noise and file compression and its effect on the recognition of caries in digital imaging. Dentomaxillofac Radiol 2000;29:20-7.  Back to cited text no. 27
    


    Figures

  [Figure 1], [Figure 2]
 
 
    Tables

  [Table 1], [Table 2]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
Abstract
Introduction
Materials and Me...
Results
Discussion
Conclusion
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed3385    
    Printed110    
    Emailed0    
    PDF Downloaded500    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]