Clin Oral Invest (2013) 17:1847–1853

DOI 10.1007/s00784-012-0882-2

ORIGINAL ARTICLE

Relationship between bone mineral density and trabecular

bone pattern in postmenopausal osteoporotic
Brazilian women
Matheus Lima Oliveira &
Esio Fortaleza Nascimento Chaves Pedrosa &
Adriana Dibo Cruz & Francsico Haiter-Neto &
Francisco Jose Albuquerque Paula &
Plauto Christopher Aranha Watanabe

Received: 27 August 2012 / Accepted: 19 November 2012 / Published online: 14 December 2012
# Springer-Verlag Berlin Heidelberg 2012

Abstract Results FD and PI values were significantly different

Objectives The aim of this study is to evaluate the relative (p<0.05) between normal and osteoporotic patients for the
efficacy of the mandibular trabecular bone in detecting right and left mandibular angle and for the left mandibular
osteoporotic-associated bone changes in Brazilian women. body. There was no significant difference (p>0.05) between
Materials and methods Seventy-three digital panoramic the right and left mandibular angle for FD and PI, but a
radiographs of healthy and osteoporotic postmenopausal significant difference (p<0.05) was observed on the man-
women were used. Regions of interest were selected in the dibular body for PI values. FD and PI values presented
mandibular angle and body of the panoramic images, which positive correlation.
were analyzed for fractal dimension (FD) and pixel intensity Conclusions The mandibular trabecular bone was effective
(PI). Statistical analysis evaluated the distribution pattern, in detecting osteoporotic changes, with the exception of the
compared normal with osteoporotic patients and the right right mandibular body.
with the left mandible, and assessed the relationship be- Clinical relevance Mandibular trabecular bone shows a
tween FD and PI. The significance level was set at 0.05. promising attempt of accurate detection of osteoporotic
changes in some regions of the jaws. Fractal analysis and
pixel intensity had strong positive correlation.
M. L. Oliveira (*) : E. F. N. C. Pedrosa : F. Haiter-Neto
Department of Oral Diagnosis, Piracicaba Dental School, Keywords Osteoporosis . Densitometry . Panoramic
State University of Campinas–UNICAMP, P.O. Box 52, radiography . Radiographic image enhancement
13414-903, Piracicaba, São Paulo, Brazil
e-mail: [email protected]

A. D. Cruz Introduction
Department of Specific Formation, Area of Radiology,
Dental School of the Fluminense Federal University, Osteoporosis is a skeletal disorder characterized by low
Polo of Nova Friburgo, 22, Dr Silvio Henrique Braune Street,
Nova Friburgo 28625-650 Rio de Janeiro, Brazil
bone mass and microarchitectural deterioration of bone tis-
sue. It results from diminished organic bone matrix (mostly
F. J. A. Paula collagen protein and mucopolysaccharides) rather than from
Department of Internal Medicine, School of Medicine bone calcium. This disorder leads to enhanced bone fragil-
of Ribeirao Preto, University of Sao Paulo, 3900,
ity, with a consequent increase in fracture risk. It is estimat-
Bandeirantes Ave, Monte Alegre,
Ribeirao Preto 14040-900 São Paulo, Brazil ed that 200 million people worldwide have osteoporosis [1].
According to the Brazilian Institute of Geography and Sta-
P. C. A. Watanabe tistics, the total Brazilian population is expected to be at 260
Department of Morphology, Stomatology and Physiology,
million by 2050, of which about 30 % will be older than
Ribeirao Preto Dental School, University of Sao Paulo,
3900, Bandeirantes Ave, Monte Alegre, 65 years of age [2]. Considering that old age is a risk factor
Ribeirao Preto 14040-900 São Paulo, Brazil for osteoporosis, an elevation in the number of those
1848 Clin Oral Invest (2013) 17:1847–1853

affected with this disease is expected. Caucasian females are taken at 65 kVp, 6 mA, and 7.4 s. The eligibility criteria
at highest risk of developing osteoporosis. A higher quality for selecting images were females at postmenopausal stage
of life is correlated with a lower risk of osteoporosis. The with a bone densitometry report of the lower spine and hips
major complication of the disease is the increased risk of classified as either normal (BMD T score >−1.0) or osteo-
fractures, which may spontaneously occur in routine activ- porotic (BMD T score <−2.5), in accordance with the World
ities or because of low-intensity trauma [3]. The overall rate Health Organization classifications. Within the group of
of mortality is approximately 20 % in the first 12 months selected images, any images of females with systemic dis-
after a hip fracture, and the costs to the public health system eases that could influence bone mineral density, such as
are significant [4, 5]. In Brazil, there currently are 28 million secondary osteoporosis, poorly controlled thyrotoxicosis,
toothless people. Although the strongest risk factors of tooth primary hyperparathyroidism, malabsorption, liver disease,
loss are dental caries and periodontal disease [6], this may or alcoholism, were not included in the study. The subjects
also be related to bone loss [7]. of each image ranged in age from 45 to 70 years. Exclusion
Dual energy X-ray absorptiometry (DXA) is the most criteria were incomplete bone density measurements, inad-
typical and accurate method to measure bone mineral den- equate radiographic material, presence of condensing ostei-
sity (BMD) [8]. Starting in 1960, the radiographic image of tis, osteosclerosis, or local destructive lesions of the
the mandibular bone has been evaluated for its efficacy in mandible. Standardized regions of interest (ROI) of 230×
the diagnosis of osteoporosis. Since then, several studies 130 pixels were selected on the right and left mandibular
have been done on radiographic findings of osteoporosis body and angle. The selection of the ROIs on the mandib-
in the oral cavity [9–13]. Since its introduction into the ular body was below the apex of the canine, right anteriorly
general practice of dentistry, panoramic radiography has to the mental foramen. On the mandibular angle, the ROIs
become a popular and valuable diagnostic tool [9]. Digital were below the mandibular canal, posteriorly to the molar
radiographs are an increasingly popular option in the clinic. region, so as to prevent interference of masticatory stress in
Such images are composed of pixels with a specific numer- the study. Since patients have different anatomical features,
ical value for each one. Two important methods of evaluat- keeping the same image resolution and ROI size was of
ing digital images are fractal dimension (FD) and pixel great importance (Fig. 1).
intensity (PI) analyses. FD is an objective measurement of FD and PI were measured in all ROIs using ImageJ
the complexity of the bone tissue architecture, describing 1.44o, a public domain software developed by the
shapes and structural patterns [14]. PI is a grayscale mea- National Institutes of Health (Bethesda, MD, USA).
sure, ranging from 0 (black) to 256 (white) in an 8-bit digital To calculate FD of the trabecular bone structure, the
image [10]. ROIs were submitted to a sequence of procedures as
Because the panoramic radiography is an exam more follow: blurring, subtraction, addition, binarization, ero-
common and affordable than DXA, and it shows the entire sion, dilatation, inversion, skeletonization, and superim-
maxillomandibular region on a single film, its application in position (Fig. 2) [15]. As the first step of this sequence,
the early detection of low bone mass would bring the sig- the blurring of the image attenuates high frequency
nificant benefit of early treatment to those afflicted with signal (Gaussian Blur). The subtraction of the blurred
osteoporosis. The aim of this paper was to evaluate the image from the original ROI and the addition of 128 to
relative efficacy of mandibular trabecular bone in detecting each pixel location result in an image in which most
osteoporotic-associated bone changes in panoramic radio- variation in brightness is due to the presence of trabec-
graphs of postmenopausal Brazilian women. ulae versus marrow spaces. The binarization stores each
pixel as a single bit (black or white). The erosion
followed by dilatation avoids the generation of indeter-
Materials and methods minate values for structures occurring along the edges

The experimental protocol was approved by the Human

Research Ethics Committee (protocol # 2007.1.1086.58.6).
Seventy-three digital panoramic images at 265 dpi were
selected from a clinical digital archive of patients referred
for dental treatment. The images were acquired using the
Veraviewepocs 2D (Morita, Kyoto, Japan) unit, a digital X-
ray system, by the same radiology technician. A chin rest
positions the anterior teeth in the focal layer, and a head
stabilizer centers the patient in the unit. The direct automatic
exposure function was disabled and all exposures were Fig. 1 ROI locations in the mandible
Clin Oral Invest (2013) 17:1847–1853 1849

Fig. 2 Resulting images after

digital processing: 1 original
ROI, 2 blurring, 3 subtraction, 4
addition, 5 binarization, 6
erosion, 7 dilatation,
8 inversion, 9 skeletonization,
10 superimposition

of the image. The inversion makes the trabeculae black, between FD and PI values. The significance level was set at
and the skeletonization leaves only the central line of 5 % (α00.05).
pixels to remain. Lastly, the superimposition of the
skeletonized image on the original ROI visually dem-
onstrates that the outline image corresponds to the Results
original trabeculae. FD was calculated for both binary
and skeletonized images by the box-counting method Within the 73 selected panoramic images, 38 (52 %) images
(Minkowski–Bouligand dimension). All data were were of normal patients and 35 (48 %) were of osteoporotic
recorded on a Microsoft Excel for Mac 2011 spread- patients. The normal patients presented T score of −1.0 or
sheet. The Shapiro–Wilk test evaluated the distribution higher at the lower spine and hips. The osteoporotic patients
pattern of the data. Based on the type of data distribu- presented at least two sites with T score of −2.5 or lower.
tion, Student’s t test, Wilcoxon signed-rank, and Mann– FD values were significantly different (p<0.05) between
Whitney U tests compared the normal with the osteo- normal and osteoporotic groups for the right and left man-
porotic group, and the right with the left mandible. dibular angle and left mandibular body. On the right side,
Pearson’s correlation was used to determine the relationship FD values of the mandibular body did not differ
1850 Clin Oral Invest (2013) 17:1847–1853

significantly (p>0.05) between normal and osteoporotic Table 2 Means (± standard deviation) for pixel intensity values of
different mandibular sites based on patient groups
groups. Right and left mandibular angle and body did not
differ significantly (p > 0.05) for both patient groups Normal Osteoporotic
(Table 1).
PI values were significantly different (p<0.05) between Mandibular angle
normal and osteoporotic groups for the right and left man- Right 7.06 (1.74) aA 6.25 (1.74) aB
dibular angle and left mandibular body. On the right side, PI Left 6.81 (1.88) aA 5.96 (2.00) aB
values of the mandibular body did not differ significantly Mandibular body
(p>0.05) between normal and osteoporotic groups. Right Right 6.75 (1.28) aA 6.19 (1.69) aA
and left mandibular angle did not differ significantly Left 7.21 (0.95) bA 6.59 (1.43) bB
(p>0.05) from each other. A significant difference between
Right and left mandible with different lowercase letters differed by
right and left mandibular body was observed (p<0.05) in paired t test. Normal and osteoporotic groups with different capital
both patient groups (Table 2). FD values were positively and letters differed by unpaired t test (p<0.05)
strongly correlated with PI values in all regions and groups
(r from 0.84 to 0.92) (Fig. 3). use. Besides, this technique can offer a dose advantage over
large numbers of intraoral radiographs [17].
FD is a value that indicates how completely a fractal
appears to fill space. Several authors have reported that
Discussion structures with high FD have greater structural complexity.
Lower FD values are indicative of simpler structures [14].
The diagnosis of osteoporosis is made by measuring the FD measurements of healthy bone taken from computed
BMD. DXA is the most widely used technique for bone tomography images range from 1.70 to 1.80 [18]. In the
measurements, since it is considered to be able to provide an present study, the mean values of FD for normal and oste-
accurate estimation of bone mineral density in adults. There oporotic groups taken from digital panoramic radiographs
are several bone sites in the body to perform DXA scans. were 1.40 and 1.36, respectively. Such values indicate lower
The spine or femur must be one of them in the evaluation of complexity of the mandibular trabecular bone pattern of the
bone density [16]. In the present study, evaluated sites were osteoporotic group in the sample studied.
the lumbar spine and hips. Yasar and Akgünlü studied 48 postmenopausal women
Assessment of trabecular bone pattern is an important and found no significant difference in FD between healthy
factor in the analysis of mandibular bone quality for surgical and osteoporotic patients in their study [11]. However, they
planning, implant placement, orthodontic movements, and calculated FD from direct digital periapical radiographs,
prosthetic support. The mandible is a single bone, which which present much higher spatial resolution than panoram-
belongs to the axial skeleton. The effect of masticatory ic radiographs. FD can only be reliably compared when
forces on mandibular trabeculae is still inconsistent. Pano- using radiographs at the same spatial resolution [19]. Law
ramic radiography has been widely used in screening and et al. made use of fractal analysis to evaluate digitized dental
epidemiological studies because of the convenience of their radiographs for signs of osteoporosis [20]. The osteoporotic
group presented higher FD than the normal group, which is
Table 1 Means (± standard deviation) for fractal dimension values of in disagreement with the results found in our study. These
different mandibular sites based on patient groups divergent findings may be related to the age range of the
sample and the acquisition mode. Law et al. used women
Normal Osteoporotic
from 22 to 85 years old as control group [20]. Our study
Mandibular anglea included only postmenopausal women with ages ranging
Right 1.41 (0.08) aA 1.36 (0.10) aB from 45 to 70. The age range affects the standardization of
Left 1.40 (0.07) aA 1.35 (0.11) aB FD values, since hormonal changes are dissimilar between
Mandibular bodyb young and elderly patients. Besides, the present study used
Right 1.40 (0.08) aA 1.38 (0.08) aA direct digital panoramic images at 256 dpi. There is a
Left 1.42 (0.05) aA 1.37 (0.08) aB scarcity of literature concerning fractal analysis with regard
to the reliability of digitized images.
a
Right and left mandible with different lowercase letters differed by In 2003, a group of researchers from five European
paired t test. Normal and osteoporotic groups with different capital
centers started a project known as OSTEODENT. The aim
letters differ by unpaired t test (p<0.05)
b of this project was to identify the most valid and effective
Right and left mandible with different lowercase letters differed by
Wilcoxon test. Normal and osteoporotic groups with different capital radiographic index, or combination of radiographic and
letters differ by the Mann–Whitney test (p<0.05) clinical indices, for the diagnosis of osteoporosis in female
Clin Oral Invest (2013) 17:1847–1853 1851

Fig. 3 Relationship between FD and PI values in normal (a) and osteoporotic (b) mandibular angle, and normal (c) and osteoporotic (d)
mandibular body

patients in the age range 45–70 years who have undergone not skeletonized, making the analysis more susceptible to
radiography for usual dental reasons. Over the years, they interferences. Image skeletonization emphasizes the geo-
have shown that the Osteoporosis Index of Risk (OSIRIS)— metrical and topological properties of shape by reducing
age, weight, current estrogen therapy, and history of low foreground regions of a binary image to a skeletal remnant.
trauma fracture—has higher diagnostic validity than the Additionally, Van der Stelt and Geraets have stated that any
measurement of the mandibular cortical width (MCW) on step of the digital image processing can influence the final
panoramic radiographs [21]. However, patients with MCW radiographic image [28].
thinner than 3 mm should be referred for osteoporosis in- Feltrin et al. studied the lumbar vertebral cancellous bone
vestigation [22]. More recently, in 2010, the OSTEODENT architecture and stated that FD of trabecular bone should be
index, which is the result of the OSIRIS, was showed to considered as a supplement to BMD evaluation in the as-
have value in prediction of hip fracture risk [23]. Even sessment of osteoporosis [29]. Tosoni et al. found no signif-
before the OSTEODENT project, the MCW showed to be icant correlation between FD values of three groups
directly associated with skeletal bone mass [24], including (normal, osteopenic, and osteoporotic) of postmenopausal
young adult men [25]. Nevertheless, Ferreira Leite et al. women, differing from our results [10]. This may possibly
demonstrated that there was no apparent association be- be due to the size and location of ROIs and the inclusion
tween vertebral fragility fractures and the appearance of criteria. Tosoni et al. selected three ROIs of different sizes
mandibular cortex in elderly men [26]. In a study, which on the right side of the mandible in panoramic radiographs
correlated FD from periapical radiographs and panoramic of healthy, osteopenic, and osteoporotic women [10]. In this
radiographs with MCW and morphology, Bollen et al. stated study, two groups were evaluated: normal and osteoporotic.
that FD values increase as BMD decreases [27]. However, The non-inclusion of osteopenic patients aimed to eliminate
in the image processing for fractal analysis, the images were subtle differences between the studied groups, focusing in
1852 Clin Oral Invest (2013) 17:1847–1853

the osteoporosis detection. All ROIs had 230×130 pixels Africans, and Amerindians. Miscegenation reduces the av-
and were located in the right and left mandibular angle and erage genetic distance between different population clusters.
body, away from masticatory stress. This results in biological variation, with polygenic human
In digital imaging, a pixel is a single point. It is the phenotypes, and makes the study of individual phenotypic
smallest unit of picture that can be represented or controlled. differences (genetic traits) even more difficult [36]. Limita-
The intensity of each pixel is variable. In a radiographic tion in the selection of the ROIs on panoramic radiographs
image (grayscale), the value of each pixel carries only should be considered due to anatomic variation among
intensity information, varying from black at the weakest patients. Although complex and not widespread, both the
intensity to white at the strongest [30]. Hedstrom et al. found FD and PI are methods of evaluating digital images avail-
a positive correlation between PI in digital periapical radio- able for the dental practitioner, since some of the imaging
graphs of the mandible and the BMD of the heel [31]. software applications are of free download. “Dentists have a
Unlike Law et al. [20], who analyzed scan lines to determine role to play in the detection and referral of patients at high
PI, we analyzed scan areas. However, both studies showed risk of osteoporosis” [22].
that the mean PI value for the osteoporotic group was lower In conclusion, according to the results found in this
than for the normal group. Moreover, in Southard and research and within the limitations of the methodology
Southard’s study, a comparison of digitized radiographic used, the mandibular trabecular bone was effective in
alveolar features between 20- and 70-year-old women detecting osteoporotic changes in postmenopausal Bra-
allowed the affirmation that a measurable reduction in the zilian women, with the exception of the right side of
magnitude of the PI occurs with age [32]. Tosoni et al. the mandibular body. The methods of fractal analysis
evaluated the relative efficacy of PI analysis in detecting and pixel intensity had similar ability in the analysis of
osteoporotic-associated bone density changes [10]. As in the mandibular trabecular bone with strong positive
our study, the normal group had significantly higher PI than correlation. Further studies about fractal analysis are
the osteoporotic group. Such results could be expected necessary to better characterize osteoporotic-associated
because of the relatively high penetrability of the X-ray factors, since it is a contradictory issue.
photons into osteoporotic bone. This produces a more ra-
diolucent image with consequent lower PI. It could also be
Acknowledgments The authors wish to gratefully acknowledge the
observed that the PI in the right and left mandibular body help from Mr Christopher Wenell for reviewing the manuscript.
were significantly different both for normal and osteoporotic
groups, which may suggest anatomical variations or sensi- Conflict of interest The authors declare that they have no conflict of
tivity of this analysis in detecting unilateral chewing. Pos- interest.
sible positioning errors should be considered, since the
panoramic radiography technique presents a relatively nar-
row image layer in the anterior region [33]. Although this
study had an unbiased random selection of individuals, the References
authors believe that the sample size might not have been
enough to avoid sampling process error. The use of a larger 1. Cooper C (1999) Epidemiology of osteoporosis. Oteoporosis Int 9:
sample could have eliminated such difference. S2–S8
Examining structural parameters in the assessment of 2. Oliveira JC, Albuquerque FRPC, Lins IB (2004) Projeção da
população do Brasil por sexo e idade para o período 1980–2050.
bone quality, Matsubara et al. suggested that PI is a useful
IBGE 2004. http://www.ibge.gov.br/home/estatistica/populacao/
tool in the evaluation of bone quality in radiographic images projecao_da_populacao/metodologia.pdf. Accessed 25 Oct 2012
[34]. Their methodology made use of bone specimens, and 3. Jonasson G, Alstad T, Vahedi F, Bosaeus I, Lissner L, Hakeberg
therefore, there was no soft tissue interference. The presence M (2009) Trabecular pattern in the mandible as bone fracture
predictor. Oral Surg Oral Med Oral Pathol Oral Radiol Endod
of soft tissue increases the X-ray attenuation, leading to a
108:e42–e51
decrease of the signal received by the image receptor, 4. Morales-Torres J, Gutierrez-Urena S (2004) The burden of osteo-
and consequent higher PI values. However, similarly to porosis in Latin America. Osteoporos Int 15:625–632
Matsubara et al. [34], our results were not expected to be 5. Lee S, Lee JW, Jeong JW, Yoo DS, Kim S (2008) A preliminary
study on discrimination of osteoporotic fractured group from non-
influenced by soft tissues, since all of the images were made
fractured group using support vector machine. Conf Proc IEEE
binary and skeletonized as previously described [15]. More- Eng Med Biol Soc 2008:474–477
over, race is an important determinant of BMD, which for 6. Persson GR, Berglund J, Persson RE, Renvert S (2011) Prediction
instance is consistently higher in Black women [35]. The of hip and hand fractures in older persons with or without a
diagnosis of periodontitis. Bone 48:552–556
present study assessed the Brazilian population, which is
7. Krall EA, Garcia DW, Dawson-Hughes B (1996) Increased risk of
well known for its extensive miscegenation, as a result of tooth loss is related to bone loss at the whole body, hip, and spine.
five centuries of interethnic crosses among Europeans, Calc Tiss Int 59:433–437
Clin Oral Invest (2013) 17:1847–1853 1853

8. Reginster JY, Burlet N (2006) Osteoporosis: a still increasing in detecting osteoporosis: the OSTEODENT study. Br Dent J 204:
prevalence. Bone 38:S4–S9 E16, discussion560-1
9. Taguchi A, Tsuda M, Ohtsuka M, Kodama I, Sanada M, Nakamoto T, 23. Horner K, Allen P, Graham J, Jacobs R, Boonen S, Pavitt S,
Inagaki K, Noguchi T, Kudo Y, Suei Y, Tanimoto K, Bollen AM Nackaerts O et al (2010) The relationship between the OSTEO-
(2006) Use of dental panoramic radiographs in indentifying younger DENT index and hip fracture risk assessment using FRAX. Oral
postmenopausal women with osteoporosis. Osteoporos Int 7:387–394 Surg Oral Med Oral Pathol Oral Radiol Endod 110:243–249
10. Tosoni GM, Lurie AG, Cowan AE, Burleson JA (2006) Pixel 24. Horner K, Devlin H, Harvey L (2006) Detecting patients with low
intensity and fractal analyses: detecting osteoporosis in perimeno- skeletal bone mass. J Dent 30:171–175
pausal and postmenopausal women by using digital panoramic 25. Bollen AM, Taguchi A, Hujoel PP, Hollender LG (2001)
images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod Fractal dimension on dental radiographs. Dentomaxillofac
102:235–241 Radiol 30:270–275
11. Yasar F, Akgunlu F (2006) The differences in panoramic mandib- 26. Ferreira Leite A, de Souza Figueiredo PT, Ramos Barra F, Santos
ular indices and fractal dimension between patients with and with- de Melo N, de Paula AP (2011) Relationships between mandibular
out spinal osteoporosis. Dentomaxillofac Radiol 35:1–9 cortical indexes, bone mineral density, and osteoporotic fractures
12. Taguchi A, Ohtsuka M, Tsuda M, Nakamoto T, Kodama I, Inagaki K in Brazilian men over 60 years old. Oral Surg Oral Med Oral
et al (2007) Risk of vertebral osteoporosis in post-menopausal women Pathol Oral Radiol Endod 112:648–656
with alterations of the mandible. Dentomaxilofac Radiol 36:143–148 27. Taguchi A, Sugino N, Miki M, Kozai Y, Mochizuki N, Osanai H et
13. Aranha Watanabe PC, Moreira Lopes De Faria L, Mardegan Issa al (2011) Detecting young Japanese adults with undetected low
JP, Caldeira Monteiro SA, Tiossi R (2009) Morphodigital evalua- skeletal bone density using panoramic radiographs. Dentomaxillo-
tion of the trabecular bone pattern in the mandible using digitized fac Radiol 40:154–159
panoramic and periapical radiographs. Minerva Stomatol 58:73–80 28. Van der Stelt PF, Geraets GW (1991) Computer-aided interpreta-
14. Sánchez I, Uzcátegui G (2011) Fractals in dentistry. J Dent tion and quantification of angular periodontal bone defects on
39:273–292 dental radiographs. IEEE Trans Biomed Eng 38:334–338
15. White SC (2002) Oral radiographic predictors of osteoporosis. 29. Feltrin GP, Macchi V, Saccavini C, Tosi E, Dus C, Fassina A et al
Dentomaxillofac Radiol 31:84–92 (2001) Fractal analysis of lumbar vertebral cancellous bone archi-
16. Watanabe PC, Dias FC, Issa JP, Monteiro SA, de Paula FJ, Tiossi tecture. Clin Anat 14:414–417
R et al (2010) Elongated styloid process and atheroma in pano- 30. Chen SK, Hollender L (1995) Digitizing of radiographs with a
ramic radiography and its relationship with systemic osteoporosis flatbed scanner. J Dent 23:205–208
and osteopenia. Osteoporos Int 21:831–836 31. Hedström L, Baigi A, Bergh H (2010) The relation between
17. Ludlow JB, Davies-Ludlow LE, White SC (2008) Patient risk bone mineral density in the heel and pixel intensity in the
related to common dental radiographic examinations: the impact mandibular jaw bone among elderly women. Dentomaxillofac
of 2007 International Commission on Radiological Protection Radiol 39:409–413
recommendations regarding dose calculation. J Am Dent Assoc 32. Southard KA, Southard TE (1992) Comparison of digitized radio-
139:1237–1243 graphic alveolar features between 20- and 70-year-old women. A
18. Richardson ML, Gillespy T (2003) Fractal analysis of trabecular preliminary study. Oral Surg Oral Med Oral Pathol 74:111–117
bone. Life sciences publications. http://www.fractal.org/Life-Sci 33. Ladeira DB, Cruz AD, Almeida SM, Bóscolo FN (2010) Evalua-
ence-Technology/Publications/Fractal-Analysis-of-Trabecu tion of the panoramic image formation in different anatomic posi-
lar-Bone.htm. Accessed 25 Oct 2012 tions. Braz Dent J 21:458–462
19. Veenland JF, Grashius JL, van der Meer F, Beckers AL, Gelsema 34. Matsubara M, Morita S, Shinomiya K, Kawamata R, Nakamura K,
ES (1996) Estimation of fractal dimension in radiographs. Med Kashima I (2003) Structuring parameters for assessment of bone
Phys 23:585–594 quality using a morphological filter and star volume analysis:
20. Law AN, Bollen AM, Chen SK (1996) Detecting osteoporosis structuring property in the cancellous bone of the human femoral
using dental radiographs: a comparison of four methods. J Am head. J Bone Miner Metab 21:48–56
Dent Assoc 127:1734–1742 35. Katzmarzyk PT, Barreira TV, Harrington DM, Staiano AE,
21. Karayianni K, Horner K, Mitsea A, Berkas L, Mastoris M, Jacobs Heymsfield SB, Gimble JM (2012) Relationship between abdom-
R, Lindh C et al (2007) Accuracy in osteoporosis diagnosis of a inal fat and bone mineral density in white and African American
combination of mandibular cortical width measurement on dental adults. Bone 50:576–579
panoramic radiographs and a clinical risk index (OSIRIS): the 36. Ribeiro RR, Guerra-Junior G, de Azevedo Barros-Filho A (2009)
OSTEODENT project. Bone 40:223–229 Bone mass in schoolchildren in Brazil: the effect of racial misce-
22. Devlin H, Allen P, Graham J, Jacobs R, Nicopoulou-Karayianni K, genation, pubertal stage, and socioeconomic differences. J Bone
Lindh C, Marjanovic E et al (2008) The role of the dental surgeon Miner Metab 27:494–501