Dentomaxillofacial Radiology (2005) 34, 261267 q 2005 The British Institute of Radiology http:/ /dmfr.birjournals.

org

RESEARCH

Fractal dimension and lacunarity analysis of dental radiographs

nlu F Yas ar* and F Akgu
Selcuk Universitesi Dis Hekimligi Fakultesi, Oral Diagnoz ve Radyoloji Anabilim Dali, Konya, Turkey

Objective: As the occlusal forces transmitted to the jaw bones during mastication might be different in dentate and edentulous regions, there might be different radiographical trabecular bone texture in these regions. Image analysis procedures are promising techniques which are used to detect structural changes of bone texture on radiographs. In this study, the differences of fractal dimension (FD) and lacunarity measurements of radiographical trabecular bone between dentate and edentulous regions were investigated. Methods: Direct digital radiographs of premolar-molar region were taken from 51 patients who were included in our study. Two rectangular regions of interest (ROIs) with the same dimensions (37 119 pixels) were created on these radiographs; one in the edentulous region and the other one in the dentate region. The ROIs were segmented as black and white areas. Box-counting fractal dimension and lacunarity of these regions were calculated. Results: Paired samples t-test and Pearson correlation coefcients were calculated. It was found that there were differences between dentate and edentulous regions for FD and lacunarity (P 0.000). There is a negative correlation between FD and lacunarity (2 0.643, P , 0.01), positive correlation between dentate and edentulous regions and FD (0.819, P , 0.01), and a negative correlation between lacunarity and dentate and edentulous regions (2 0.541, P , 0.01). Conclusions: The differences of occlusal forces generated in dentate and edentulous regions during mastication cause some alterations in trabecular bone structure, and fractal dimension and lacunarity can reveal these alterations quantitatively. Dentomaxillofacial Radiology (2005) 34, 261267. doi: 10.1259/dmfr/85149245 Keywords: image analysis; dental radiographs; fractal dimension; lacunarity Introduction The most common method of assessing bone strength is to monitor loss of bone mass by bone mineral densitometry (BMD) which is a non-invasive quantitative technique. BMD accounts for 80% of bone strength and is still the main evaluation index. Currently, bone strength is thought to depend not only on bone mass but also on bone microarchitecture (structure and morphology) and therefore, it is thought necessary to evaluate bone quality as well as bone mass.1 Bone micro-architecture is assessed using histomorphometric procedures.2 Standard histomorphometric procedures have some drawbacks and one of them is that the invasive nature of bone biopsy technique does not allow longitudinal studies on individuals.3 Conventional radiographs offer a quick, non-invasive, available, and inexpensive method to assess skeletal tissue, but they are
sun Yas *Correspondence to: Dr Fu ar, Selcuk Universitesi Dis Hekimligi Fakultesi, Oral Diagnoz ve Radyoloji Anabilim Dali, 42075 Konya, Turkey; E-mail: drfyasar@hotmail.com Received 22 October 2004; revised 16 February 2005; accepted 15 March 2005

generally considered insensitive for alveolar bone lesions, since a 30 50% mineral loss is required for their delineation.4,5 Following the development of digital imaging systems, researchers attempted to extract more information from images with the help of digital image processing and analysis techniques. The importance of the internal structure of bone lies in the fact that living bones adjust their internal structure to the mechanical forces to which they are subjected.6 If an object consists of an inhomogeneous material, then its response to external mechanical loading is determined by its external macroscopic shape and by the orientation of its microstructure.6 The trabecular architecture of cancellous bone seems to be optimally structured for its load-bearing function, suggesting that its formation is governed by the mechanical forces. It is known that bone structure also adapts later in life, if external forces change.7 Muscle attachments to the bone surface generate functional tensions that strengthen the bone and prevent mineral loss. Bone metabolism in the alveolar process changes markedly

262

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

after extraction of the teeth and functional stress also decreases in those regions where the muscles are attached.8 Researchers have differentiated patients with periodontitis from gingivitis9 and individuals with osteoporosis and those without osteoporosis10 12 by using dental radiographs and image analysing procedures. Fractal dimension (FD) can be viewed as a measure of irregularity of many physical processes.13 The range of application of FD extends from pure and applied mathematics, through physics and chemistry to biology and medicine.14 The heart contains many branching systems with a fractal structure, including the coronary arteries,15,16 the conducting system17 and the chordae tendinea.17 The branching pattern of the bronchial tree is more accurately described in fractal terms than by an exponential process of division,18 and the pulmonary arterial system is likewise accurately modelled by fractal analysis.19 The self similar structure of the intestinal absorptive surface, from mucosal folds down to microvilli, has been described.6 Higher FD values indicate more complex textures but different fractal sets may share the same FD and have strikingly different textures.20 For discriminating these textures, FD alone would be useless and Mandelbrot introduced the term lacunarity to describe the characteristics of fractals of the same dimension with different texture appearances.20 Lacunarity is related to the distribution of gap sizes: low lacunarity geometric objects are homogeneous because all gap sizes are the same, whereas high lacunarity objects are heterogeneous. Lacunarity can be considered a scale dependent measure of heterogeneity of texture.21 Both FD and lacunarity are second order statistical measures which quantify the relationships between neighbouring pixels, and are thus inherently more able to characterize the trabeculation pattern than rst order measures such as trabecular separation, intercept trabecular thickness or intercept trabecular separation.22 Because the occlusal forces generated during mastication are transmitted to the jaw bones by means of teeth, it was suspected that radiographic trabecular bone texture characteristics might be different in dentate and edentulous regions. Our null hypothesis was that the texture characteristics of trabecular bone will be the same in dentate and edentulous regions, so there would not be a signicant difference in the FD and lacunarity values of these regions. Materials and methods Direct digital images of the mandibular posterior region which were taken during the routine examinations of 51 patients, aged 40 64 years, were chosen for the study. None of the patients had known systemic diseases that would affect bone metabolism, cancers with bone metastasis or signicant renal impairment. They were not using specic drugs or hormones (corticosteroids, excess thyroid hormone) which are known to have adverse effects on bone metabolism. They were not smoking and consuming alcoholic beverages. There were neither radiolucent nor radiopaque lesions where the ROIs were located.
Dentomaxillofacial Radiology

The patients had rst and second premolars and had an edentulous region in the molar area without any xed or removable prosthetic appliances (Figure 1). The direct digital periapical images were taken at 65 kV, 10 mA and 2.5 mm aluminium equivalent ltration with CCX Digital Periapical X-ray machine (Trophy Radiologie, France) and Dimaxis (1.52) software (PlanMeca-Helsinki, Finland). The exposure time was 0.08 s. The reason for choosing the mandibular premolar-molar region was that this region is relatively less prone to projection errors than the other areas of maxilla and mandible. ImageJ (IJ) 1.28 program23 was used for all image processing and analysing. First 16-bit direct digital images were converted to 8-bit images because only 8-bit images can be segmented with ImageJ. Two regions of interest (ROIs), whose dimensions were 37 119 pixels (Figure 2a), were created in these images, one between the premolars and the other one in the adjacent edentulous region. The ROIs extended in the corono-apical direction. When creating ROIs, great care was taken not to include lamina dura, periodontal ligament space or root structure.10 Because it has been stated that beyond 2.5 mm there is no effect of bacterial plaque on alveolar process bone,24 the ROIs were created apically as far as possible from the crestal bone. Digital images were segmented to binary image in a similar way described by White and Rudolph.25 The ROIs were duplicated and blurred by a Gaussian lter with a diameter of 35 pixels (Figure 2b). This step removed all ne-scale and medium-scale structure and retained only large variations in density. The resulting heavily blurred image was then subtracted from the original, and 128 was added to the result at each pixel location (Figure 2c). This generates an image with a mean value of 128, regardless of the initial intensity of the image. The aim of this operation was to reect individual variations in the image such as trabeculae and marrow spaces. The image was then made binary, thresholding on a brightness value of 128 and inverted, thus segmenting the image into components that radiographically approximated the trabeculae and marrow spaces (Figure 2d). With this process, the regions which represent trabecular bone were set to black and marrow spaces were set to white. Later the images were eroded and dilated once and the outlines of the images were obtained. (Figure 2e).

Figure 1 Original radiograph

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

263

biological cells and textures. It works on only black pixels on a white background, or white pixels on a black background, so images must be thresholded prior to analysis to ensure that only the pixels of interest are assessed.23 As stated by Plotnick et al,26 Lacunarity, L(r), can be dened in terms of the local rst and second moments (i.e. local mean and variance) measured for different neighbourhood sizes, r, about every pixel in an image, i.e.; Lr 1 {varr =mean2 r } 1

Where mean(r) and var(r) are the mean and variance, respectively, for neighbourhood size, r, using all neighbourhoods of that size within the image. Higher lacunarity values indicate more translational invariance (higher contagion), i.e. wider range of sizes of structures within an image.26 SPSS 10.0 (SPSS Inc., Chicago, IL) was used for the statistical analysis. Dentate regions were classied as Group 1 and edentulous regions were classied as Group 2. Paired samples t-test was computed using P-value as 0.05 with 95% condence interval. The relationships between FD and lacunarity in dentate and edentulous regions were assessed using Pearson correlation coefcient. Results Table 1 shows the means and the standard deviations of the variables and the results of paired samples t-test. The mean FD and lacunarity values for dentate regions were 1.3623 and 0.4088 and for edentulous regions 1.6512 and 0.3372, respectively. According to the paired samples t-test results, it was found that there were statistically signicant differences between dentate and edentulous regions for FD and lacunarity (P 0.000). The ndings of this study did not support the null hypothesis that there were no signicant differences in the FD and lacunarity values of dentate and edentulous sites of the radiographical trabecular bone in the premolar and molar area of the human mandible. There was a negative correlation between FD and lacunarity (2 0.643, P , 0.01) and positive correlation between the groups (dentate vs edentulous) and FD (0.819, P , 0.01). FD showed an increase in edentulous regions. There was a negative correlation between lacunarity and the groups ( 2 0.541, P , 0.01), that is lacunarity decreased in edentulous regions. According to the results of this study it can be said that edentulous regions have more complex and homogeneous trabecular structure than dentate regions. Lower lacunarity values in edentulous

Figure 2 Transformation of the region of interest (ROI) to binary and outlined image prior to fractal dimension (FD) calculation. (a) Original ROI; (b) Gaussian lter applied image; (c) ltered image subtracted from the original image; (d) threshold image; (e) transformation of the binary into an outline image from which the FD and lacunarity were calculated

Fractal dimension and lacunarity FD was calculated with ImageJ in box counting method and lacunarity was calculated with ImageJ plugin named FracLac. It is a plug-in that analyses digital images for the FD, generalized dimension spectrum, lacunarity and other morphometrics. It is suitable for analysing images of
Table 1

The means and standard deviations of the variables and paired samples t-test results P , 0.05 Group N 51 51 51 51 Mean 1.3623 1.6512 0.4088 0.3372 Standard deviation 0.1176 0.0837 0.0640 0.0472 Standard error of the mean 0.0165 0.0117 0.0090 0.0066 t 2 14.306 6.460 P-value (2-tailed) 0.000 0.000 1 2 1 2

Fractal dimension Lacunarity

Group 1: Dentate regions; Group 2: Edentulous regions

Dentomaxillofacial Radiology

264

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

regions represent uniform sized trabecular structures which are regularly arranged within the ROI. Discussion The mechanical properties of cancellous bone depend on its architecture and the tissue properties of the mineralized matrix. The architecture is continuously adapted to external loads.27 For years, bone density has been studied as a major determinant in quantifying bone strength. This can be attributed to the fact that structure constitutes a feature that is, compared with density, difcult to quantify non-invasively.28 Bones have an important role in maintaining posture and motor function29 and bone grows in response to the loads applied to it. The density of bone in a particular location depends on the magnitude of the applied loads.30 Structure and density are two concepts that are physically closely intertwined and therefore strongly correlated. Each structure feature will show a correlation with density.28 The main building blocks of bone are calcium carbonate, calcium phosphate, collagen and water. These minerals give bone its stiffness and strength. Van der Linden et al27 investigated the adaptation of cancellous bone to external loads using computer models. They reported that an increase in bone tissue stiffness would lead to an equal increase in global stiffness if the architecture was not adapted. If the architecture is adapted based on the local deformations of the bone tissue, increased bone tissue stiffness will induce bone resorption: higher tissue stiffness results in smaller deformations in the tissue. The cells will decide that less bone is needed to carry the loads and adapt the bone architecture. Bone tissue will be resorbed until the deformation of the tissue is normalized. This will partly compensate for the increased bone tissue stiffness. They assumed that the bone adaptation is performed according to the mechanism described above and simulated this process in three-dimensional (3D) models of human cancellous bone. They reported that, during the simulation, extra bone was added in highly loaded regions and bone was removed in only slightly loaded regions.27 In general, trabeculae can be lost by two mechanisms. The rst is biological, in which some trabeculae become so thin from age-related reduction in bone formation rates that normal osteoclastic resorption breaks through the entire trabecula. This interrupts the trabecular network, effectively removing a surface upon which osteoblasts can subsequently replace the resorbed one. The second mechanism is by mechanical overload, in which a trabecula is fractured due to high local stress, resulting again in an interruption of the trabecular network and the same sequellae as with the rst mechanism.31 Like the entire skeletal system, the alveolar cortical and cancellous bone undergoes constant remodelling processes, in which bone resorption exceeds bone formation with advancing age.32 Following tooth loss, the maxillary and mandibular alveolar processes additionally undergo extensive bone resorption, mainly as a result of vertical resorption.8,33 In addition, with progressive ridge atrophy, muscular
Dentomaxillofacial Radiology

function decreases in order to protect the bony structures of the ridges, especially in the mandible.34 Individuals with partial or total edentulism, who also display impaired masticatory function,35,36 show a reduced amount of bone in the residual alveolar process and mandibular body.37,38 Kingsmill and Boyde39 examined the tissue level mineralization density distribution in dry mandibles. The effect of dental status was also investigated in their study to see the affects of function on bone turnover. Some of the results of their study were summarized as follows: (1) (2) (3) The pattern of lower mineralization density in the mandible seems to match the pattern of bone loss of teeth. The regions of highest mineralization density mirror the sites thought to experience the highest strains. The mandible undergoes alterations in its net mineralization level on becoming partially dentate, and these probably counter the reduction in torsional rigidity following loss of the posterior teeth.

The diameter of bone trabeculae ranges from approximately 50 mm to 200 mm and to analyse trabecular bone structure high resolution imaging is mandatory.40 There is detailed information about the resolution of digital imaging systems in an article of Versteeg et al.41 The spatial resolution of the system must equal half of the spatial frequency of the smallest details to be detected.41 Commercial computed tomography (CT) scanners have limited spatial resolution (, 0.5 mm) and are unable to properly resolve trabecular bone structure. Modied peripheral CT systems with improved spatial resolution (, 0.20 mm), experimental high resolution (100 150 mm) and microcomputed tomography with spatial resolutions of around 15 mm have all been used in imaging of trabecular bone structure in peripheral skeleton. However, the high dose required to obtain images of the central skeleton at this resolution could not be justied for in vivo evaluation of bone structure.42 These imaging modalities are also beyond the practice of general dental practitioners. Most recent charge-coupled device (CCD) digital imaging systems are in the range of 10 23 lp mm21 43,44 and it is reported that they offer pixel sizes of 19.5 mm.44 Besides this, intraoral radiographs are less expensive than the other imaging modalities and they serve as a readily available source for the dentists. Plain lm radiographs represent a two-dimensional image of 3D trabecular structures. Although plain lm radiographs appear to offer the potential to assess trabecular structure non-invasively, there is at present virtually no understanding as to the relationship between 3D architecture and the associated two-dimensional radiographic pattern. Luo et al45 and Pothuaud et al46 studied on this matter and reported that plain radiographs contain architectural information directly related to the underlying 3D structure. Projection radiographs are necessarily degraded by superimposition of cortical bone on the trabecular structure,42 however, it is reported that the buccal and cortical plates of the mandible and maxilla do not cast a

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

265

discernible image on periapical radiographs.47 In one study it was concluded that the removal of trabecular bone alone did not change the appearance of the bone pattern radiographically; they determined that the pattern interpreted as trabecular bone is actually the result of interaction of the radiation with the bone pattern on the endosteal surface of the cortex.48 Two other studies evaluated the visibility of trabecular structures in oral radiographs and both of the studies concluded that trabecular bone is a substantive contributor to the image of bone striae seen on periapical radiographs.49,50 Texture in images consists of aggregates of many small elements of patterns. Traditional methods of texture analysis can be broadly classied into two main categories: statistical and structural.51 Use of autocorrelation function, modelling texture by random elds, and co-occurence matrix are commonly used techniques in statistical models. Structural models rely on nding an elementary pattern (called a texture primitive or texel) that is replicated to generate the texture pattern.52 The basic concepts of fractal geometry hold that we do not live in a Euclidean world of points, straight lines, rectangles and cubes, except as largely created by man. Natural objects are often rough and are not well described by the ideal constructs of Euclidean geometry. An important characteristic of fractal geometry is the property of self-similarity. Fractal images are similar, in a statistical sense, at all levels of magnication or scale. As a fractal image is viewed at higher and higher magnications, the amount of detail is constant. This is equivalent to stating that any measured length is proportional to the resolvable length raised to a power. In principle, a theoretical or mathematically generated fractal is self-similar over an innite range of scales. Natural fractal images, however, have a limited range of selfsimilarity.53 Lacunarity was originally developed to describe a property of fractals but is not predicated on self-similarity and can be used to describe the spatial distribution of real data sets. This is a distinct advantage over FD, which has been commonly used as a compact descriptor of texture in medical images that often exhibit little self-similarity. Lacunarity plots explicitly characterize the spatial organization of an image, including the average size of any structural sub-unit(s) within an image, making them potentially useful in representing the trabecular thinning and perforation of trabecular bone associated with bone loss. The lacunarity algorithm is simple to implement, depending only on local means and variances calculated for different window sizes throughout the image. Lacunarity plots are robust to the presence of noise and blurring within the imaging system so that corrections for these effects need not be implemented for most studies. They can be generated in near real-time, following image acquisition and binarization.42 Dong reported that detailed textural information is lost in the process of converting a grey-scale image into a binary image and because of this reason, it would be very difcult to analyse the directional properties of textures in a binary image.13 In this study, FD and lacunarity were calculated in binary images and this may be a limitation for the study. The image processing and analysis program

which was used in this study performs operations on binary images. Our aim was to show the differences in trabecular bone architecture in dentate and edentulous regions. As the same image processing and analysis operations were applied for both ROIs, the amount of the information which was lost in binarization should be same for dentate and edentulous regions. In this study tooth roots and anatomical structures were not included in the ROIs and although many of the ROIs were parallel to the long axis of tooth root, some of them had minor deviations. The sizes of the ROIs were kept constant (37 119 pixels). In one study it was reported that the fractal dimensions derived from digitized dental radiographs were not affected by variations in exposure or small variations in alignment and imply an absolute region of interest placement may not be necessary.54 In another study the same investigators reported that the region of interest size and shape may affect the results of fractal analysis of alveolar bone.55 The authors especially stated that care must be exercised in placing the ROIs on images, so that the structures included are the same in all images. And if one is interested in the FD of trabecular bone, one should not include tooth structure in the ROI that denes calculations. From these results, we can conclude that ROI size affects FD calculations more than ROI placement as long as root structures and other anatomical structures are not included in the ROIs. When evaluating the trabecular bone in longitudinal studies, absolute ROI placement may be required. In the dentate mandible, the trabecular bone is surrounded by a thick cortical shell on one side and the cribriform plate on the other side where the periodontal ligament is attached to the teeth.56 Periodontal ligament distributes and cushions the stress generated during chewing.57 Because trabecular bone does not resist the functional loads transmitted by the periodontal ligament, the trabeculae in the dentate mandible appear to be randomly arranged.58 The results of our study are consistent with these results. In this study it was found that dentate regions have lower FD and higher lacunarity. Higher lacunarity values represent a wider range of sizes of bone structures within the ROI. Trabeculae and marrow spaces had more variable shape and arrangement in dentate regions. Wilding et al measured bone-to-marrow proportion and FD for the bone adjacent to the root of a tooth and for an edentulous region in two partially edentate mandibles.59 The mean bone-to-marrow proportion was found as 38.90% for dentate region and 21.92% for edentulous region. The mean FD was 2.501 for dentate region and 2.408 for edentulous region, that is, FD was higher in dentate regions than edentulous regions. Their results are conicting with the results of this study. FD was higher in edentulous regions in this study. FD was calculated in boxcounting method in this study and it is one of the length related methods. Wilding et al59 calculated FD in blanket method and it is a mass related method. They used conventional radiographs and studied on dry mandibles. Direct digital images were used in this study and it was an in vivo study. The structure of the trabecular bone has been
Dentomaxillofacial Radiology

266

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

analysed in relation to osteoporosis with FD in many studies and it has provided conicting results. While some authors stated that FD decreases in osteoporosis,60,61 others reported an increase.62,63 Geraets and van der Stelt64 have stated that most reports based on radiographs in vivo found an association between osteoporosis and increased values of FD. In one study FD analysis of weight-bearing bone of rats was investigated during skeletal unloading. 65 They reported an increase in the FD in skeletal unloading. Their results are similar to the results of this study. Occlusal loads might be less in edentulous regions than dentate regions and this resulted with increased FD and decreased lacunarity. In conclusion, dentate and edentulous regions have different trabecular bone textures; and edentulous regions have more complex and homogeneous trabecular structure than dentate regions. Here homogeneous structure denes
References
1. Tanaka T, Sakurai T, Kashima I. Structuring of parameters for assessing vertebral bone strength by star volume analysis using a morphological lter. J Bone Miner Metab 2001; 19: 150 158. 2. Partt AM. The stereologic basis of bone histomorphometry. Theory of quantitative microscopy and reconstruction of the third dimension. Bone histomorphometry. Techniques and interpretation. Boca Raton, FL: CRC Press, 1983: pp. 53 87. egsegger P, Koller B, Mu ller R. A microtomographic system for 3. Ru the nondestructive evaluation of bone architecture. Calcif Tissue Int 1996; 58: 24 29. 4. Southard KA, Southard TE. Comparison of digitized radiographic alveolar features between 20- and 70-year-old women. A preliminary study. Oral Surg Oral Med Oral Pathol 1992; 74: 111 117. 5. Okano T, Mera T, Ohki M, Ishikawa I, Yamada N. Digital subtraction of radiographs in evaluating alveolar bone changes after initial periodontal therapy. Oral Surg Oral Med Oral Pathol 1990; 69: 258 262. 6. Geraets WG. Comparison of two methods for measuring orientation. Bone 1998; 23: 383 388. 7. Huiskes R. If bone is the answer, then what is the question. J Anat 2000; 197: 145 156. 8. Klemetti E, Kolmakow S. Morphology of the mandibular cortex on panoramic radiographs as an indicator of bone quality. Dentomaxillofac Radiol 1997; 26: 22 25. 9. Shrout MK, Roberson B, Potter BJ, Mailhot JM, Hildebolt CF. A comparison of two patient populations using fractal analysis. J Periodontol 1998; 69: 9 13. 10. Southard TE, Southard KA, Jacobsen JR, Hillis SL, Najim CA. Fractal dimension in radiographic analysis of alveolar process bone. Oral Surg Oral Med Oral Pathol Endod 1996; 82: 569 576. 11. Bollen AM, Taguchi A, Hujoel PP, Hollender LG. Fractal dimension on dental radiographs. Dentomaxillofac Radiol 2001; 30: 270 275. 12. Southard TE, Southard KA, Lee A. Alveolar process fractal dimension and postcranial bone density. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001; 91: 486 491. 13. Dong P. Test of a new lacunarity estimation method for image texture analysis. Int J Remote Sensing 2000; 21: 3369 3373. 14. Smith TG, Lange GD, Marks WB. Fractal methods and cellular morphology dimensions, lacunarity and multifractals. J Neurosci Methods 1996; 69: 123 136. 15. Bassingthwaighte JB, van Beek JHGM. Lightning and the heart: fractal behavior in cardiac function. Proc IEEEE 1988; 76: 693 699. 16. Goldberger AL, Rigney DR, West BJ. Chaos and fractals in human physiology. Sci Am 1990; 46: 42 49. 17. Goldberger AL, West BJ. Fractals in physiology and medicine. Yale J Biol Med 1987; 60: 421 435.

more uniform sized and regularly arranged trabecular architecture. FD and lacunarity can discriminate the texture differences of trabecular bone in dentate and edentulous regions quantitatively. The differences of trabecular bone texture in dentate and edentulous regions may be depending on the differences of occlusal loads, as well as the differences in trabecular bone formation during healing process after the extraction of teeth and the anatomical regional differences. Further research is necessary to reveal the effects of these factors. The conclusion of this study is valid only for radiographic appearance of the trabecular bone, not for the histological properties of trabecular bone architecture. Based on the conclusions of this study and the other studies which are mentioned above, we can state that dentists have the possibility of evaluating the quality of trabecular bone architecture with the use of direct digital radiographs and image analysing techniques.

18. West BJ, Goldberger AL. Physiology in fractal dimensions. Am Sci 1987; 75: 354 365. 19. Lefevre J. Teleonomical optimization of a fractal model of the pulmonary arterial bed. J Theor Biol 1983; 102: 225 248. 20. Mandelbrot BB. The fractal geometry of nature, 3rd edn. New York, NY: WH Freeman & Co., 1983: pp 14 19, 310 318. 21. Gefen Y, Aharony A, Mandelbrot BB. Phase transitions on fractals: 3. Innitely ramied lattices. J Phys A: Math Gen 1984; 17: 1277 1289. 22. Vesterby A, Mosekilde L, Gundersen HJ, Melsen F, Mosekilde L, Holme K, et al. Biologically meaningful determinants of the in vitro strength of lumbar vertebrae. Bone 1991; 12: 219 224. 23. http://rsb.info.nih.gov/ij/ [Accessed 27 April 2005]. 24. Carranza FA. Bone loss and patterns of bone destruction. In: Carranza FA, Newman MG, (editors). Clinical periodontology, (8th edn). Philadelphia: W.B. Saunders 1996: pp. 297 313. 25. 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: 628 635. 26. Plotnick RE, Gardner RH, ONeill RV. Lacunarity indices as measures of landscape texture. Landscape Ecology 1993; 8: 201 211. 27. Van der Linden JC, Days JS, Verhaar JAN, Weinans H. Altered tissue properties induce changes in cancellous bone architecture in aging and disease. J Biomech 2004; 37: 367 374. 28. Veenland JF, Grashuis JL, Weinans H, Ding M, Vrooman HA. Suitability of texture features to assess changes in trabecular bone architecture. Patt Recog Lett 2002; 23: 395 403. 29. Kumasaka S, Kyohara S, Takahashi T, Asai H, Kashima I. Morphologically extracted trabecular skeleton superimposed upon digital radiograph structure. J Bone Miner Metab 2000; 18: 208 211. 30. Gibson LJ. The mechanical behaviour of cancellous bone. J Biomech 1985; 18: 317 328. 31. Keaveny TM, Yeh OC. Architecture and trabecular bone- toward an improved understanding of the biomechanical effects of age, sex and osteoporosis. J Musculoskel Neuron Interact 2002; 2: 205 208. 32. Nedelman CI, Bernick S. The signicance of age changes in human alveolar mucosa and bone. J Prosthet Dent 1978; 39: 495 501. 33. Ulm C, Kneissel M, Schedle A, Solar P, Matejka M, Schneider B, et al. Characteristic features of trabecular bone in edentulous maxillae. Clin Oral Implants Res 1999; 10: 459 467. 34. Tallgren A. The continuing reduction of the residual alveolar ridges in complete denture wearers: a mixed longitudinal study covering 25 years. J Prosthet Dent 1972; 27: 120 132.

Dentomaxillofacial Radiology

Fractal dimension and lacunarity analysis F Yas ar and F Akgu nlu

267

35. Helkimo E, Carlsson G, Helkimo M. Bite force and state of dentition. Acta Odontol Scand 1977; 35: 297 303. 36. Slagter A, Bosman F, van der Glas H, van Der Bilt A. Human jaw elevator muscle activity and food comminution in the dentate and edentulous state. Arch Oral Biol 1993; 38: 195 205. 37. Taguchi A, Tanimoto K, Suei Y, Wada T. Tooth loss and mandibular osteopenia. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995; 79: 127 132. 38. Klemetti E. A review of residual ridge resorption and bone density. J Prosthet Dent 1996; 75: 512 514. 39. Kingsmill VJ, Boyde A. Mineralisation density of human mandibular bone: quantitative backscattered electron image analysis. J Anat 1998; 192: 245 256. 40. Link TM, Majumdar S, Grampp S, Guglielmi G, Van Kuijk C, Imhof G, et al. Imaging of trabecular bone structure in osteoporosis. Eur Radiol 1999; 9: 1781 1788. 41. Versteeg CH, Sanderink GCH, van Der Stelt PF. Efcacy of digital intraoral radiography in clinical dentistry. J Dent 1997; 25: 215224. 42. Dougherty G. A comparison of the texture of computed tomography and projection radiography images of vertebral trabecular bone using fractal signature and lacunarity. Med Eng Phys 2001; 23: 313 321. 43. Farman AG, Farman TT. A status report on digital imaging for dentistry. Oral Radiol 2004; 20: 9 14. 44. Sanderink GC. Intraoral and extraoral digital imaging: an overview of factors relevant to detector design. Nucl Inst Meth A 2003; 509: 256261. 45. Luo G, Kinney JH, Kaufman JJ, Haupt D, Chiabrera A, Siffert RS. Relationship between plain radiographic patterns and three dimensional trabecular architecture in the human calcaneus. Osteoporos Int 1999; 9: 339 345. 46. Pothuaud L, Benhamou CL, Porion P, Lespessailles E, Harba R, Levitz P. Fractal dimension of trabecular bone projection texture is related to three dimensional microarchitecture. J Bone Miner Res 2000; 15: 691 699. 47. White SC, Pharoah MJ. Normal radiographic anatomy. Oral radiology principles and interpretation, (4 edn). St. Louis, MO: Mosby-Year Book, 2000; p 175. 48. Cavalcanti MGP, Ruprecht A, Johnson W, Southard T, Jacobsen J. Radiologic interpretation of bone striae: an experimental study in vitro. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1999; 88: 353 357. 49. Couture RA, Whiting BR, Hildebolt CF, Dixon DA. Visibility of trabecular structures in oral radiographs. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2003; 96: 764 771. 50. Colosi D, Potluri A, Islam S, Geha H, Lurie A. Brief communication: bone trabeculae are visible on periapical images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2003; 96: 772 773.

51. Haralick RM. Statistical and structural approaches to texture. Proc IEEE 1979; 67: 786 809. 52. Samarabandu J, Acharya R, Allen K. Analysis of bone x-rays using morphological fractals. IEEE Trans Medical Imaging 1993; 12: 466 470. 53. Smith TG, Marks WB, Lange GD, Sheriff Jr WH, Neale EA. A fractal analysis of cell images. J Neurosci Methods 1989; 27: 173 180. 54. 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. 55. Shrout MK, Hildebolt CF, Potter BJ. The effect of varying the region of interest on calculations of fractal index. Dentomaxillofac Radiol 1997; 26: 295 298. 56. Misch CE, Qu Z, Bidez WM. Mechanical properties of trabecular bone in the human mandible: implications for dental implant treatment planning and surgical placement. J Oral Maxillofac Surg 1999; 57: 700 706. 57. Pietrzak A, Curnier A, Botsis J, Scherrer S, Wiskott A, Belser U. A nonlinear elastic model of the periodontal ligament and its numerical calibration for the study of tooth mobility. Compt Methods Biomech Biomed Engin 2002; 5: 91 100. 58. Ralph WJ, Thomas CDL. Tooth support in the human mandible. J Oral Rehabilitation 1988; 15: 499 503. 59. Wilding RJC, Slabbert JCG, Kathree H, Owen CP, Crombie K, Delport P. The use of fractal analysis to reveal remodeling in human alveolar bone following the placement of dental implants. Archs Oral Biol 1995; 40: 61 72. 60. Link TM, Majumdar S, Lin JC, Newitt D, Augat P, Ouyang X, et al. A comparative study of trabecular bone properties in the spine and femur using high resolution MRI and CT. J Bone Miner Res 1998; 13: 122 132. 61. Fazzalari NL, Parkinson IH. Fractal properties of cancellous bone of the iliac crest in vertebral crush factors. Bone 1998; 23: 53 57. 62. 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. 63. Pothuaud L, Lespesailles E, Harba R, Jennane R, Royant V, Eynard E, et al. Fractal analysis of trabecular bone texture on radiographs: discriminant value in postmenopausal osteoporosis. Osteoporosis Int 1998; 8: 618 625. 64. Geraets WG, van der Stelt PF. Fractal properties of bone. Dentomaxillofac Radiol 2000; 29: 144 153. 65. Pornprasertsuk S, Ludlow JB, Webber RL, Tyndall DA, Sanhueza AI, Yamauchi M. Fractal dimension analysis of weight-bearing bones of rats during skeletal unloading. Bone 2001; 29: 180 184.

Dentomaxillofacial Radiology