Nele Stoppie Veerle Pattijn Tim Van Cleynenbreugel Martine Wevers Jos Vander Sloten Ignace Naert
Authors’ affiliations: Nele Stoppie, Ignace Naert, Department of Prosthetic Dentistry/BIOMAT Research Group, School of Dentistry, Oral Pathology and Maxillofacial Surgery, Katholieke Universiteit Leuven, Leuven, Belgium Veerle Pattijn, Tim Van Cleynenbreugel, Jos Vander Sloten, Division of Biomechanics and Engineering Design, Department of Mechanical Engineering, Katholieke Universiteit Leuven, Leuven, Belgium Martine Wevers, Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Leuven, Belgium Correspondence to: Ignace Naert Departmenyt of Prosthetic Dentistry/BIOMAT Research Group Kapucijnenvoer 33 3000 Leuven Belgium Tel.: þ 32 16 33 24 38 Fax: þ 32 16 33 23 09 e-mail:
[email protected]
Structural and radiological parameters for the characterization of jawbone
Key words: compression, CT scan, DXA, hounsfield, jawbone, micro-CT, ultrasonography Abstract Objectives: The aim of this study was to determine the Hounsfield values of selected bone sites on a computed tomography (CT) scan of the jaw and to investigate the relationship between this radiological parameter and structural parameters. Materials and methods: A selection of 24 bone samples out of eight cadaver human jaws was made. The following parameters were measured: Hounsfield value in the jaw (HU1) determined by a first CT scan, Hounsfield value of the excised bone specimen (HU2) by a second CT scan, bone mineral density (BMD) by a dual-energy X-ray absorptiometry scan, bone volume (BV/TV) by the microfocus CT scan, first peak transmission time (TUS-1) and first zero crossing transmission time (TUS-2) by an ultrasound measurement and Young’s modulus (EMECH) by a compression test. Results: Thirteen specimens were composed of a mix of trabecular and a small amount of cortical bone, while another 11 specimens were composed of trabecular bone only. A good correlation was found between the HU value of the specimen (HU2) and BMD (r ¼ 0.99), BV/TV (r ¼ 0.97), TUS-1 (r ¼ 0.83), TUS-2 (r ¼ 0.87) and EMECH (r ¼ 0.83). For the pure trabecular bone specimens, the HU value of the excised bone specimen (HU2) was highly correlated (r ¼ 0.95) with the HU value of the total jaw scan (HU1). For mixed trabecular– cortical bone specimens, this relationship was weak (r ¼ 0.57). Conclusion: With the current CT scan technology, predictions of the mechanical properties of trabecular jaw bone based on Hounsfield values were only valid for jaws with a thin layer of cortical bone. For jaws with a thicker cortical layer, the prediction of the mechanical properties decreased significantly.
Date: Accepted 12 May 2005 To cite this article: Stoppie N, Pattijn V, Van Cleynenbreugel T, Wevers M, Vander Sloten J, Naert I. Structural and radiological parameters for the characterization of jawbone. Clin. Oral Impl. Res. 17, 2006; 124–133 doi: 10.1111/j.1600-0501.2005.01204.x
Copyright r Blackwell Munksgaard 2006
124
The success of dental implants has been highly correlated with the jaw bone quantity and quality (Hutton et al. 1995; Esposito et al. 1998). The jawbone volume has been traditionally evaluated using intraoral and panoramic films besides computed tomography (CT) scans. Schwarz et al. (1987a, 1987b) introduced the concept of CT for pre-operative assessment of dental implant candidates. Cross-sectional, axial and panoramic views are provided, giving detailed information on the anatomic
structures in the three dimensions. Besides quantity, bone quality has been proven to have a strong influence on implant success (Engquist et al. 1988; Jaffin & Berman 1991; Jemt et al. 1992). Fanuscu & Chang (2004) described the microstructure of trabecular bone in the mandible and maxilla of one cadaver. They found a wide variation of bone density irrespective of the amount of cancellous bone. In contrast to a precise determination of bone volume (BV), bone quality is usually assessed by a
Stoppie et al . Characterization of jawbone
crude grading method. For pre-operative clinical assessment, Lekhom & Zarb (1985) suggested a bone classification based on macrostructure where the morphology and distribution of cortical and trabecular bone determined the quality. Lindh et al. (1996) found that it was impossible to evaluate the accuracy of the proposed classification because of a wide variation of interobserver (between 49% and 64%) as well as of intraobserver agreement (between 75% and 86%). They proposed a new classification with reference images to assess the trabecular pattern in periapical radiographs before implant treatment. Norton & Gamble (2001) concluded that there is a need for an objective quantitative classification of bone quality, which can be applied pre-operatively and is not operator dependent. They proposed a new classification based on the Hounsfield units of the bone on the CT scan and related it to the existing classification of Lekholm & Zarb (1985). Other suggestions for classification of bone quality were based on the drilling resistance of bone (Misch et al. 1993; Friberg et al. 1995) and could not be considered as a pre-operative diagnostic test anymore. A new pre-operative classification of bone quality is mandatory. As CT is becoming a standard procedure in the preoperative planning of dental implants, the Hounsfield units of the selected implant sites are available and a classification like the one proposed by Norton and Gamble in 2001 seems most suitable. The question remains as to how these Hounsfield values relate to the structural and mechanical properties of the jawbone. The aim of this study was to determine the Hounsfield values of selected jawbone sites and to investigate the relationship of these Hounsfield values with the bone mineral density (BMD) determined by the dualenergy X-ray absorptiometry (DXA) scan, the BV based on the microfocus CT scan, the transmission time of the ultrasound measurement and the Young’s modulus measured with a compression test.
Materials and methods Six cadaver human heads were randomly selected at the department of Anatomy
Fig. 1. Example of defining implant sites in upper jaw with Surgicase V 3.1.0.5 software (Materialise).
K.U. Leuven. No information on age and gender was available. They were formalin embalmed and stored in a freezer at 201. The experiments complied with the general ethical practice at the University of Leuven (Belgium). CT scan
CT scans of the six heads were made with a Siemens Somatom Zoom Plus4 scanner (Siemens AG, Erlangen, Germany). The following parameters were used: 120 kV, 90 mA s and 0.5 mm slice thickness. These CT images were loaded in the Surgicase V 3.1.0.5 software (Materialise, Haasrode, Belgium), and a 3D model of each jaw was reconstructed. Bone cylinders of 8 mm length and 6 mm diameter were virtually defined in the trabecular bone of the 3D models (Fig. 1). A selection of 24 bone samples out of eight fully edentulous jaws (five maxillas and three mandibles) was made. Surgical drill guides based on
stereolithographical models of the jaws were made for trepanation of the bone samples on the pre-selected sites (Materialise). The bone samples were retrieved with a trephine bur (6 mm inner diameter) for the full height of the jaw, under perfuse cooling, perpendicular to the plane of occlusion. To remove the upper and lower cortex, the samples were frozen with nitrogen into a cupper rod of 8 mm width. A robot (Sta¨ubli RX 130, 6 degrees of freedom, Sta¨ubli Unimation, Faverges, France) was used to mill the ends of the bone samples in order to obtain an exact height of 8 mm. The cylindrical bone samples were stored in saline solution in a freezer at –201C. A second CT scan was made of the bone samples in order to determine the influence of the hard and soft tissues present in the first CT scan. The samples were placed in a Perspex cylinder filled with saline solution and scanned again with the Siemens Somatom Zoom Plus4 scanner. The
125 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
Stoppie et al . Characterization of jawbone
parameter settings for this scan were 120 kV, 200 mA s and 0.5 mm slice thickness. The mean Hounsfield value for each planned (HU1) and corresponding excised (HU2) bone sample was determined, respectively, on the first CT scan of the jaw and on the second CT scan of the excised bone sample. The Surgicase V 3.1.0.5 software (Materialise) was used for the planned bone samples, and the Mimics software (Materialise) was used for the excised bone samples. DXA
DXA (DXA fan beam system, QDR 4500a, Hologic Inc., Waltham, MA, USA) mea-
surements were made of each excised bone sample, upright and lying down, twice for each position. The BMD, measured on the projection of the bone sample, was calculated as the quotient of the bone mineral content and the area and was obtained in g/ cm2. The BMD was expressed as the mean of the four measurements. Microfocus CT (Micro-CT)
The excised bone samples were scanned using a microfocus CT system (Skyscan 1072, Skyscan, Aartselaar, Belgium) with the following parameters: 50 kV and 300 mA, resulting in a resolution of 13.67 mm (Fig. 2). The micro-CT images were segmented for bone with a fixed
Fig. 2. Radiograph and micro-computed tomography slice of an excised jawbone sample.
threshold for all bone samples. The volume trabecular and cortical bone as a fraction of the volume of tissue (BV/TV) was calculated. The 3D anisotropy of each bone sample was quantified by the mean intercept length (MIL) (Pattijn et al. 2001). Therefore, a circular region of interest was drawn in three central orthogonal planes and a grid of parallel test lines, under an angle y with the horizontal reference direction, was superimposed onto it. The MIL (y) was then calculated as the total length of the parallel test lines divided by the total number of intersections between the bone and non-bone components along the test lines. These calculations were repeated for each grid under an angle y starting from 01 until 1801 with a 51 increment. For each of the three orthogonal planes, the MIL (y) was plotted as a radius at the angle of measurement y. These data were approximated by an ellipse using a least-square method and the ellipse was also drawn on the polar plot (Fig. 3). The major axis a, the small axis b, the rotation angle d between the major axis and a horizontal reference axis and the quadratic error d of this ellipse were calculated. The quadratic error was defined as the sum of the quadrates of the difference between the calculated MIL value and the radius from the approximated ellipse for each angle of measurement y.
50
50
50
100
100
100
150
150
150
200
200
200
250
250
250
300
300
300
350
350
350
50
100 120
150 200 250 300 350 90 0.5 60 0.4 0.3
150
100 120
50
100 120
0
180
330
210
300 270
0
210
330
300 270
30
0.2 0.1
180
240
150 200 250 300 350 90 0.5 60 0.4 0.3
150
30
0.1
0.1
240
150 200 250 300 350 90 0.5 60 0.4 0.3 0.2
150
30
0.2
50
0
180
330
210
300
240 270
Fig. 3. Three central orthogonal planes of a bone sample with indication of the circular region used to calculate the mean intercept length (MIL) (y), the corresponding polar plot of the MIL (y) and the elliptical approximation.
126 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
Stoppie et al . Characterization of jawbone
An ultrasonic pulse was sent through the bone sample by means of the following setup: a pulse generator Agilent 33250A 80 MHz (Agilent Technologies, Palo Alto, CA, USA), a sensor for acoustic emission VS375-M Vallen (Vallen-Systeme, Munich, Germany) serving as transmitter, and a contact ultrasonic transducer 0.5 MHz V101 Panametrics (Panametrics-NDT, Waltham, MA, USA) serving as a receiver and an oscilloscope LeCroy 9310AM 400 MHz (LeCroy Corporation, Chestnut Ridge, NY, USA). The response was recorded and two measurements were performed. The time for the first peak (TUS-1) and the time for the first zero crossing (TUS2) were recorded (Nicholson et al. 1996). The time marker was placed at the point of the first maximum of the received signal and at the point where the received waveform first crossed the time axis (Fig. 4). Mechanical compression
The specimens were press-fit into aligned brass end caps using cyanoacrylate glue. A pre-load of 3 N was applied on the specimens before testing to ensure contact between the compression plates and the specimens. The specimens were subjected to 10–15 cycles of non-destructive testing in order to pre-condition the bone and to reach a steady state before the destructive test was performed. The conditioning cycles were strain controlled with the maximum set at 0.6% strain and a compression speed of 0.2 mm/min. The strain was reset to zero after each cycle and 10–15 cycles were applied to reach the steady state, defined as when the end point of the curve coincided with the starting point. At this moment, a destructive test was applied with the same compression speed of 0.2 mm/min. Young’s modulus (EMECH) was calculated as the slope of the stress– strain curve of the elastic region of the destructive test. Statistics
For the parameters HU1, HU2, BMD, TUS1, TUS-2 and EMECH, the correlation coefficients were calculated by correcting for the influence of the head from where the samples were taken. The observations were not independent as they were clustered within a head. For all 15 pairs of variables, it has been verified whether
the 24 mixed and the 11 trabecular bone specimens, respectively. The correlation matrix after correction for head differences of the measured parameters HU1, HU2, BMD, BV/TV, TUS-1, TUS-2 and EMECH is shown in Table 3 for the 24 specimens and in Table 5 for the selection of 11 trabecular bone specimens. The 95% confidence intervals are shown in Tables 4 and 6. We note that the lower limits and the upper limits of the 95% confidence intervals in Table 6 were truncated to 0 and 1 when applicable. These asymptotic confidence intervals are rather wide reflecting the small sample size. When considering the total of 24 specimens, a weak correlation was found between the Hounsfield value of the bone specimen inside the jaw (HU1) and all the other parameters. These correlations increased when the Hounsfield value of the second CT scan (HU2) was taken into account. A high correlation was observed between HU2 and BMD (r ¼ 0.99), HU2 and BV/TV (r ¼ 0.97), HU2 and TUS-1 (r ¼ 0.83), HU2 and TUS-2 (r ¼ 0.87) and HU2 and EMECH (r ¼ 0.83). High cor-
linearity was a reasonable assumption. These correlation coefficients and their correspondent 95% asymptotic confidence intervals were obtained by fitting a bivariate model using procedure Mixed in s SAS 8.2. Linear multiple regression analyses were performed to investigate the predictive value of density-related and structural parameters for the mechanical properties of bone, also correcting for the clustering (considering head as a random effect) whenever necessary. The level of significance was set at 0.05 for all statistical tests.
Results Despite the planning of the bone specimens on the first CT scan, the micro-CT images showed that for a total of 24 specimens, 13 specimens were a mix of trabecular and cortical bone, while 11 specimens were composed of trabecular bone only. Tables 1 and 2 depict the mean, SD and the range of all the measured parameters for
Tus-1
150
Ultrasonography
100 50 mV
Ultrasonography
Pulse Response
0 −50
0
−100
500
1000
1500
2000
2500
Tus-2
−150 Time (µs) Fig. 4. Ultrasonography, the time for the first peak of the received signal (TUS-1) and the time for the first zero crossing (TUS-2) was recorded.
Table 1. Summary of the properties of all 24 trabecular and mixed trabecular/cortical bone specimens Variable
Mean
SD
Min
Max
HU1 HU2 BMD (g/cm2) BV/TV (%) TUS-1 (ms) TUS-2 (ms) EMECH (MPa)
361.35 614.4 0.26 34.12 4.58 5.17 374.51
279.84 270.77 0.09 15.25 0.53 0.5 256.63
60.93 179.19 0.11 11.55 3.78 4.43 22.08
973.83 1273.38 0.49 73.08 5.62 6 953.59
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
127 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
Stoppie et al . Characterization of jawbone
Table 2. Summary of the properties of the 11 pure trabecular bone specimens Variable
Mean
SD
Min
Max
HU1 HU2 BMD (g/cm2) BV/TV (%) TUS-1 (ms) TUS-2 (ms) EMECH (MPa)
417.47 538.06 0.23 29.17 4.79 5.34 342.26
283.42 262.65 0.089 12.76 0.54 0.48 273.94
60.93 179.19 0.11 12.22 4 4.55 51.54
831.34 944.15 0.36 47.64 5.62 6 766.47
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
Table 3. Correlation matrix after correction for head differences for all 24 trabecular and mixed trabecular/cortical bone specimens Variable
HU1
HU2
BMD
BV/TV
TUS-1
TUS-2
EMECH
HU1 HU2 BMD BV/TV TUS-1 TUS-2 EMECH
1
0.57 1
0.57 0.99 1
0.56 0.97 0.95 1
0.55 0.83 0.81 0.81 1
0.59 0.87 0.86 0.86 0.97 1
0.63 0.83 0.81 0.77 0.72 0.77 1
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
Table 4. Ninety-five percent asymptotic confidence intervals (CI) of the correlation coefficients of Table 3 Variable
Estimate
SE
Lower 95% CI
Upper 95% CI
HU1–HU2 HU1–BMD HU1–BV/TV HU1–TUS-1 HU1–TUS-2 HU1–EMECH HU2–BMD HU2–BV/TV HU2–TUS-1 HU2–TUS-2 HU2–EMECH BMD–BV/TV BMD–TUS-1 BMD–TUS-2 BMD–EMECH BV/TV–TUS-1 BV/TV–TUS-2 BV/TV–EMECH TUS-1–TUS-2 TUS-1–EMECH TUS-2–EMECH
0.57 0.57 0.56 0.55 0.59 0.63 0.99 0.97 0.83 0.87 0.83 0.95 0.81 0.86 0.81 0.81 0.86 0.77 0.97 0.72 0.77
0.16 0.16 0.16 0.16 0.15 0.14 0.01 0.01 0.071 0.06 0.07 0.02 0.08 0.06 0.08 0.08 0.06 0.10 0.01 0.11 0.09
0.26 0.27 0.24 0.87 0.89 0.35 0.98 0.94 0.97 0.98 0.69 0.91 0.97 0.98 0.64 0.97 0.98 0.58 0.94 0.94 0.96
0.88 0.88 0.88 0.23 0.29 0.91 0.99 0.99 0.69 0.76 0.97 0.99 0.65 0.74 0.97 0.65 0.73 0.96 0.99 0.5 0.59
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
also observed for HU2 and BV/TV in relation to the other parameters BMD, TUS-1, TUS-2 and EMECH. Using multiple regression analyses, the following relationships were determined for the 24 samples (Table 7, left) and for the 11 samples (Table 8, left). The quality of the regression line was better for the 11 samples (adjusted R2 between 0.66 and 0.9) than for the 24 samples (adjusted R2 between 0.14 and 0.71). For the 24 specimens, regressor HU2 showed better results than regressor HU1 (Table 7). For each excised bone sample, Table 9 gives the value of the major axis a, the minor axis b, the angle d between the major axis and a horizontal axis and the quadratic error d of the ellipse fitted on the MIL (y) curve for each of the three orthogonal planes. For most of the bone samples, the quadratic error d was small, hence, a good elliptical approximation was found for the MIL (y) in the analysed planes. These structural parameters were used to improve the prediction of TUS and EMECH, which were now in the multiple regression analysis based only on a density-related parameter (HU1 and HU2) till now. The parameter aXY gives the length of the major axis of the ellipse fitted on the MIL (y) curve in the XY plane. This value is higher when the trabeculae have a more linear pattern. The parameter dMAX is the absolute value of the rotation angle d of the plane XZ or YZ in which the ratio a/b is maximal. This parameter is a measure for the orientation of the trabeculae in relation to the cylindrical axis. Relationships were determined for the 24 samples (Table 7, right) and for the 11 trabecular bone samples (Table 8, right). No or a small improvement of the prediction of TUS and EMECH was observed for the 24 specimens and for the 11 trabecular specimens when including the structural parameters.
Discussion relations were observed between BV/TV and BMD (r ¼ 0.95), BV/TV and TUS-1 (r ¼ 0.81) and BV/TV and TUS-2 (r ¼ 0.86) as well. When considering the selection of 11 trabecular bone specimens, the correlations with the Hounsfield value of the bone
128 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
specimen inside the jaw (HU1) clearly improved. High correlations were observed between HU1 and HU2 (r ¼ 0.95), HU1 and BMD (r ¼ 0.96), HU1 and BV/TV (r ¼ 0.98), HU1 and TUS-1 (r ¼ 0.96), HU1 and TUS-2 (r ¼ 0.97) and HU1 and EMECH (r ¼ 0.83). High correlations were
Despite the planning only 11 bone specimens were composed of trabecular bone only. A height of 8 mm was required for the compression test and in order to reach this length, a number of bone specimens had to be planned reaching the cortical– trabecular border. After trepanation, the
Stoppie et al . Characterization of jawbone
micro-CT images revealed that a thin border of cortical bone was still present at the top and bottom of the specimen. The mixed specimens in this study were composed of approximately 90% of trabecular bone and only 10% of cortical bone. For specimens of only cortical bone, the literature shows no or low correlations between
Hounsfield values and mechanical properties (Snyder & Schneider 1991; Rho et al. 1995). For this reason, the 11 trabecular bone specimens were considered separately during data analysis. The CT image is constructed from absorption characteristics of the subject and displayed as differences in the optical den-
Table 5. Correlation matrix after correction for head differences for the 11 pure trabecular bone specimens Variable
HU1
HU2
BMD
BV/TV
TUS-1
TUS-2
EMECH
HU1 HU2 BMD BV/TV TUS-1 TUS-2 EMECH
1
0.95 1
0.96 0.99 1
0.98 0.98 0.98 1
0.96 0.92 0.92 0.93 1
0.97 0.94 0.94 0.96 0.98 1
0.83 0.93 0.94 0.89 0.75 0.84 1
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
Table 6. Ninety-five percent asymptotic confidence intervals (CI) of the correlation coefficients of Table 5 Variable
Estimate
SE
Lower 95% CI
Upper 95% CI
HU1–HU2 HU1–BMD HU1–BV/TV HU1–TUS-1 HU1–TUS-2 HU1–EMECH HU2–BMD HU2–BV/TV HU2–TUS-1 HU2–TUS-2 HU2–EMECH BMD–BV/TV BMD–TUS-1 BMD–TUS-2 BMD–EMECH BV/TV–TUS-1 BV/TV–TUS-2 BV/TV–EMECH TUS-1–TUS-2 TUS-1–TMECH TUS-2–TMECH
0.95 0.96 0.98 0.96 0.97 0.83 0.99 0.98 0.92 0.94 0.93 0.98 0.92 0.94 0.94 0.93 0.96 0.89 0.98 0.75 0.84
0.03 0.03 0.01 0.03 0.02 0.11 0.00 0.01 0.05 0.04 0.05 0.01 0.05 0.04 0.04 0.04 0.03 0.07 0.01 0.15 1.78
0.89 0.9 0.95 1 1 0.62 0.99 0.96 1 1 0.84 0.95 1 1 0.86 1 1 0.75 0.95 1 1
1 1 1 0.90 0.94 1 1 1 0.81 0.86 1 1 0.82 0.86 1 0.84 0.91 1 1 0.45 1
sity. The raw CT values are converted into Hounsfield values (H), sometimes referred to as CT numbers, by relating the bone values to water (H ¼ 0) and air values (H ¼ 1000) by means of the following formula (Hounsfield, 1995): m mw H ¼ 1000 mw ma where m, mw and ma are the linear attenuation coefficients of the substance of interest, water and air, respectively. The results prove that the HU value of the bone specimen (HU2) is a good parameter to characterize the local bone density, volume fraction and mechanical properties. This conclusion is based on the good correlation found between the HU value of the specimen (HU2) and BMD, BV/TV, TUS-1, TUS-2 and EMECH for both groups of bone specimens. For mixed specimens, the good correlation between HU2 and the mechanical properties seems contradictory to what has been reported by Snyder & Schneider (1991) and Rho et al. (1995) but can be explained by the small amount of cortical bone present in the specimens in this study. For trabecular bone, this HU value of the second scan (HU2) is highly correlated with the HU value of the total jaw scan (HU1). For mixed specimens (trabecular–cortical bone), this relationship is rather weak. This can be explained by two reasons. The first reason is a possible difference between the planned and the excised specimen because of the trepanation of the specimen. Because of the good correlation between HU1 and HU2 for the 11 trabecular bone specimens, this seems unlikely. The second reason is a corrupted HU value of the specimen in the jaw (HU1). It has been reported that the presence of compact bone in a whole bone scan can cause errors (Hangartner et al. 1987). With increasing
Table 7. Multiple regression analysis of all 24 trabecular and mixed trabecular/cortical bone specimens Response
Regressors: HU
Adjusted R2
Regressors: HU and structural parameters aXY, dMAX
Adjusted R2
TUS-1 TUS-1 TUS-2 TUS-2 EMECH EMECH
TUS-1 ¼ 4870.93–0.81HU1 TUS-1 ¼ 5560.86–1.6HU2 TUS-2 ¼ 5479.65–0.85HU1 TUS-2 ¼ 6130.55–1.56HU2 EMECH ¼ 218.59 þ 0.43HU1 EMECH ¼ 87.75 þ 0.75HU2
0.14 0.64 0.19 0.71 0.19 0.61
TUS-1 ¼ 4838.72–0.72HU1 þ 126.43aXY–1.13dMAX TUS-1 ¼ 5664.76–1.63HU2–109.77aXY–0.22dMAX TUS-2 ¼ 5522.4–0.78HU1 þ 71.52aXY–1.61dMAX TUS-2 ¼ 6216.5–1.56HU2–55.76aXY–0.7dMAX EMECH ¼ 66.71 þ 0.55HU1 þ 240.77aXY þ 1.41dMAX EMECH ¼ 251.76 þ 0.78HU2 þ 138.42aXY þ 0.9dMAX
0.19 0.61 0.25 0.69 0.16 0.59
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.the major axis of the ellipse fitted on the mean intercept length (y) curve in the XY plane (aXY) and the absolute value of the rotation angle d of the plane XZ or YZ in which the ratio a/b is maximal (dMAX).
129 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
Stoppie et al . Characterization of jawbone
Table 8. Multiple regression analysis of 11 pure trabecular bone specimens with the same parameters used in Table 7 Response
Regressors: HU
Adjusted R2
Regressors: HU and structural parameters aXY, dMAX
Adjusted R2
TUS-1 TUS-1 TUS-2 TUS-2 EMECH EMECH
TUS-1 ¼ 5521.1–1.75HU1 TUS-1 ¼ 5803.45–1.89HU2 TUS-2 ¼ 6013.38–1.62HU1 TUS-2 ¼ 6266.31–1.73HU2 EMECH ¼ 7.10 þ 0.8HU1 EMECH ¼ 185.51 þ 0.98HU2
0.84 0.83 0.9 0.88 0.66 0.87
TUS-1 ¼ 3807.74–0.78HU1 þ 1666.29aXY þ 5.45dMAX TUS-1 ¼ 3860.84–0.77HU2 þ 1754.48aXY þ 5.23dMAX TUS-2 ¼ 5088.77–1.05HU1 þ 1021.19aXY þ 1.85dMAX TUS-2 ¼ 5243.96–1.08HU2 þ 1048.09aXY þ 1.46dMAX EMECH ¼ 773.25 þ 0.31HU1–817.43aXY þ 1.52dMAX EMECH ¼ 315.36 þ 1.062HU2 þ 68.33aXY þ 1.04dMAX
0.83 0.83 0.92 0.91 0.64 0.84
HU1, Hounsfield value in the jaw; HU2, Hounsfield value of the excised specimen; BMD, bone mineral density; BV/TV, bone volume; TUS-1, first peak transmission time; TUS-2, first zero crossing transmission time; EMECH, Young’s modulus.
Table 9. The major axis a, the minor axis b, the angle d between the major axis and a horizontal axis and the quadratic error d of the ellipse fitted on the mean intercept length MIL (h) for the three orthogonal planes Bone specimen
MIL XY plane
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
XZ plane b (mm)
d (mm2)
a (mm)
b (mm)
d (deg.)
d (mm2)
a (mm)
b (mm)
d (deg.)
d (mm2)
0.52 0.44 0.38 0.36 0.55 0.47 0.76 1.03 0.84 0.97 0.71 0.63 0.63 0.63 0.59 0.75 0.38 0.34 0.78 0.93 0.32 0.33 0.58 0.67
0.45 0.38 0.38 0.31 0.51 0.44 0.66 0.9 0.7 0.92 0.69 0.61 0.56 0.58 0.52 0.68 0.36 0.29 0.71 0.73 0.3 0.32 0.54 0.64
0.79 0.5 0.42 0.31 0.88 0.62 1.25 3.15 1.74 3.06 1.46 1.13 1.21 0.75 0.72 1.72 0.39 0.33 1.75 1.97 0.24 0.3 0.99 1.52
0.5 0.36 0.44 0.36 0.66 0.32 0.66 1.17 1.76 0.96 0.54 0.68 0.74 0.63 0.54 0.92 0.4 0.35 1.11 0.77 0.27 0.34 0.83 0.92
0.38 0.32 0.4 0.34 0.52 0.31 0.6 0.91 1.5 0.86 0.50 0.55 0.64 0.62 0.44 0.87 0.37 0.32 0.96 0.72 0.25 0.26 0.53 0.65
84 81 70 13 87 47 79 74 2 83 77 19 90 30 0 160 48 73 76 12 83 73 37 87
0.72 0.32 0.52 0.29 1.27 0.29 0.94 3.15 11.73 2.41 0.65 1.03 1.79 0.71 0.76 2.31 0.40 0.30 3.47 1.73 0.18 0.22 1.46 3.39
0.55 0.42 0.37 0.31 0.54 0.39 0.52 1.08 1 0.96 0.68 0.66 0.78 0.42 0.67 0.96 0.36 0.36 0.95 0.84 0.27 0.41 0.52 0.97
0.54 0.38 0.35 0.29 0.47 0.36 0.46 0.91 0.91 0.9 0.59 0.57 0.59 0.38 0.53 0.69 0.32 0.31 0.7 0.8 0.25 0.34 0.49 0.8
16 42 73 8 76 165 106 9 72 35 44 29 73 79 23 69 71 93 97 21 75 88 109 74
0.76 0.43 0.34 0.22 0.74 0.35 0.51 2.61 2.85 2.64 0.99 1.13 1.33 0.26 0.89 2.02 0.32 0.33 2.71 2.43 0.15 0.37 0.74 2.59
cortical wall thickness, the measured density of trabecular bone will increase, an artefact caused by a combination of photon scattering, the effect of the exponential edge gradient and the type of reconstruction algorithm used. In the case of the 11 complete trabecular bone samples, only a very thin cortical layer was present in the jaw and the resulting HU1 corresponded with HU2. Concerning the mixed specimens, a thicker layer of cortical bone surrounded the trabecular bone in the jaw and the resulting HU1 did not reflect the real value. A number of authors tried to make a new classification of bone based on the Hounsfield values of the CT scan of the jaw before
130 |
YZ plane
a (mm)
Clin. Oral Impl. Res. 17, 2006 / 124–133
implant insertion. Norton & Gamble (2001) proposed four ranges of HU values: o0, 0–500, 500–850 and a final group 4850 HU. Misch (1999) defined bone categories from D1 to D5 with corresponding ranges o150, 150–350, 350–850, 850– 1250 and 41250 HU. This study indicates that care should be exercised when applying these classifications. Practically speaking, predictions of the mechanical properties of trabecular bone are only valid for implant sites in full trabecular bone or with a very small amount of cortical bone. For jaws with a thin cortical layer, a clear relationship exists between HU1, HU2 and the mechanical properties of bone. For jaws with a thicker cortical layer, HU1 is not
reliable and the prediction of the mechanical properties decreases significantly. The progress of the medical CT scanner and the imaging software should allow obtaining images of higher quality with a better resolution in the near future. This will allow a more exact calculation of the Hounsfield value in the jaw (HU1). Pattijn et al. (2001) investigated the structural and radiological parameters of trabecular femoral bone. A good correlation was found between HU2 and pQCT density (R2 ¼ 0.95) and HU2 and BV/TV (R2 ¼ 0.95). A lower correlation was found between HU1 and HU2 (R2 ¼ 0.75). Young’s modulus EMECH did not correlate well with the density parameters HU1 and
Stoppie et al . Characterization of jawbone
HU2 (R2 ¼ 0.56 and 0.52, respectively). The inclusion of two structural parameters derived from the MIL improved the prediction of the mechanical properties by 10– 14%. For trabecular bone, our results show a good correlation between HU1 and HU2 (r ¼ 0.95), HU2 and BMD (r ¼ 0.99), HU2 and BV/TV (r ¼ 0.98), HU1 and EMECH (r ¼ 0.83) and HU2 and EMECH (r ¼ 0.93) when comparing with the study of Pattijn et al. (2001). The good correlation between HU1 and HU2 can be explained by the difference in retrieval of the specimens. In this study, a surgical guide based on a stereolithographical model of the jaw was made. No guide was used in the study of Pattijn et al. (2001). The good correlation between the density parameter HU1, HU2 and EMECH can be explained by the difference in compression testing. Pattijn et al. (2001) simulated an impact load; in our study, 10–15 conditioning cycles were applied before a slow destructive compression test was performed. The inclusion of the same structural data derived from the MIL gave no or a small improvement of the prediction of the mechanical properties. This can be because of the fact that good correlations were already observed with the density parameters HU1 and HU2. For the measurement of BMD, a number of in vivo methods are available, including single-photon absorptiometry, QCT, single-energy X-ray absorptiometry and DXA. Currently, DXA is widely accepted as the gold standard method of clinical bone mineral measurements in the prevention of osteoporosis (Patel et al. 2004). Denissen et al. (1999) concluded that DXA is precise for mineral density assessment of small threphined jawbone biopsy specimens. They found a BMD in the range of 0.084–0.1471 g/cm2 for the fresh jawbone specimens and a BMD in the range of 0.121–0.1801 g/cm2 for the histological jawbone biopsy specimens. Choe¨l et al. (2003) measured BMD of 63 mandibular bone specimens by DEXA. The mean BMD for the global bone specimen (trabecular and cortical bones) was 0.604 g/cm2 for the dentate jaws and 0.521 g/cm2 for the edentulous jaws. In this study, the BMD had a mean value of 0.264 g/cm2 (range 0.110–0.490 g/cm2) for the 24 specimens and a mean value of 0.234 g/cm2 (range 0.110–0.360 g/cm2) for the selection of 11 trabecular bone specimens. The difference
in BMD values can be explained by the amount of cortical bone present. The microfocus CT technique was first described by Feldkamp et al. (1989). Nowadays, it is a popular technique for visualizing trabecular bone because of its speed, the full three-dimensional information and its non-destructive character. O’Mahony et al. (2000) determined the BV/TV of seven samples of cancellous bone of a fresh cadaver edentulous mandible and found an average BV fraction of 0.33 (range 0.12– 0.49). In this study, the BV/TV for trabecular bone, corresponded to a mean value of 0.29 (range 0.12–0.47). In the presence of cortical bone the BV/TV will increase significantly. Nkenke et al. (2003) determined the cortical BV per tissue volume and found values between 53% and 98%. In this study, the BV/TV was calculated using the sum of cortical and trabecular BV and for the 24 specimens the range for BV/ TV was 11–73%. Cancellous bone is a tricky material to test mechanically. It is anisotropic, inhomogenous and viscoelastic. Ideally, specimens for compression testing have a length/diameter ratio of 2 (Keaveny et al. 1993). In the jaws, these dimensions cannot be reached, because of the small amount of trabecular bone present. The compression test method with conditioning cycles was used to deal with the viscoelasticity of bone (Linde & Hvid 1999). In this study, the 24 bone samples had a mean Young’s modulus of 374.51 MPa (range 22.08–953.59 MPa) and the selection of 11 trabecular bone specimens had a mean Young’s modulus of 342 MPa (range 51–766 MPa). The studies of Misch et al. (1999) and O’Mahony et al. (2000) indicated lower Young’s moduli. This can be explained by two reasons. Firstly, the use of end caps reduced the structural end phenomenon. This artefact occurs when vertically oriented and unsupported trabeculae in a cut-out specimen slide along the surface of the test platen. Consequently, without the use of end caps, strain inhomogeneity occurs in the specimen, leading to underestimation of stiffness. The study of Linde & Hvid (1989) showed a 19% increase in stiffness with the use of embedded specimens. Secondly, the specimens in this study were embalmed, because no other jaws were available. Fixation in this manner increases collagen
cross-linking and will therefore alter the properties of the bone tissue. The goal of this study justified their use despite the embalming, as a constant decrease or increase of Young’s moduli will not alter the correlation with other parameters. Ultrasound waves are sound waves consisting of frequencies above 20 kHz generated by a mechanical vibration. They are able to provide information about the medium through which they propagate as the waves are altered by the medium. In this study, it was impossible to measure the transmission time accurately ; instead, the time for the first peak of the received signal (TUS-1) was registered as a new method of recording. Also the time for the first zero crossing (TUS-2) was recorded, as described by Nicholson et al. (1996). The correlations between TUS-1 and the other parameters HU1, HU2, BMD and EMECH are comparable with the correlations with TUS-2. The latter always demonstrated slightly higher correlation coefficients, but considering the 95% confidence intervals, only the coefficient between TUS-2 and EMECH was not significantly different from zero. This might be because of the small sample size. The new method of recording TUS-1 appears to be an interesting approach, but more research is needed in this domain. High correlations were observed between TUS-1, TUS-2 and HU2; for the 24 specimens, r equalled 0.83 (TUS-1) and 0.87 (TUS-2). For the 11 trabecular bone specimens, r equalled 0.92 (TUS-1) and 0.94 (TUS-2). The negative correlations can be explained by the presence of air in the bone specimen, as the bone samples were taken out of their container before ultrasonic measurement. Lower HU values indicate more porous bone with more air present in the specimen. This air acts as an isolator and the wave will propagate slower; hence, TUS increases.
Conclusion Predictions of the mechanical properties of trabecular bone are only valid for implant sites in full trabecular bone or with a very small amount of cortical bone. A good correlation was found between the HU value of the specimen (HU2) and BMD, BV/TV, TUS and EMECH. For trabecular bone alone, a good correlation existed be-
131 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
Stoppie et al . Characterization of jawbone
tween the Hounsfield value of the jaw scan (HU1) and the Hounsfield value of the excised bone specimen (HU2). For mixed specimens (cortical and trabecular bone), this relationship was weak. We conclude that for jaws with a thin cortical layer, a clear relationship exists between HU1, HU2 and the mechanical properties of trabecular bone. For jaws with a thicker cortical layer, HU1 is not reliable and the prediction of the mechanical properties decreases significantly.
the data, the Department of Anatomy K.U. Leuven (Prof. Moerman) for the delivery of the cadaver heads and Materialise Belgium for providing the planning software and surgical guides.
Acknowledgements: This study was supported by the Fund for Scientific Research, Flanders (no. 101/8 and 1.5.118.02, Belgium). Silvia Cecere (Biostatistical Center, School of Public Health, Leuven, Belgium) is acknowledged for statistical analysis of
References Choe¨l, L., Duboeuf, F., Bourgeois, D., Briguet, A. & Lissac, M. (2003) Trabecular alveolar bone in the human mandible: a dual-energy X-ray absorptiometry study. Oral Surgery, Oral Medicine, Oral Pathology 95: 364–370. Denissen, H., Eijsink-Smeets, R., van Lingen, A. & van Waas, R. (1999) Assessing mineral density in small trephined jawbone biopsy specimens. Clinical Oral Implants Research 10: 320– 325. Engquist, B., Bergendal, T., Kallis, T. & Linden, U. (1988) A retrospective multicenter evaluation of osseointegrated implant supporting overdentures. International Journal of Oral & Maxillofacial Implants 3: 129–134. Esposito, M., Hirsch, J.M., Lekholm, U. & Thomsen, P. (1998) Biological factors contributing to failures of osseointegrated oral implants. II Etiopathogenesis. European Journal of Oral Sciences 106: 721–764. Fanuscu, M.I. & Chang, T.-L. (2004) Three dimensional morphometric analysis of human cadaver bone: microstructural data from maxilla and mandible. Clinical Oral Implants Research 15: 213–218. Feldkamp, L.A., Goldstein, S.A., Parfitt, A.M., Jesion, G. & Kleerekoper, M. (1989) The direct examination of three-dimensional bone architectures in vitro by computed tomography. Journal of Bone and Mineral Research 4: 3–11. Friberg, B., Sennerby, L., Roos, J. & Lekhom, U. (1995) Identification of bone quality in conjunction with insertion of titanium implants. A pilot study in jaw autopsy specimens. Clinical Oral Implants Research 6: 213–219. Hangartner, T.N., Battista, J.J. & Overton, T.R. (1987) Performance evaluation of density measurements of axial and peripheral bone with X
132 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
ray and gamma ray computed tomography. Physics in Medicine and Biology 32: 1393–1406. Hounsfield, G.N. (1995) Computerized transverse axial scanning (tomography): Part I. Description of system. 1973. British Journal of Radiology 68: H166–H172. Hutton, J.E., Heath, M.R., Chai, J.Y., Harnett, J., Jemt, T., Johns, R.B., McKenna, S., McNamara, D.C., van Steenberghe, D. & Taylor, R. (1995) Factors related to success and failure rates at 3year follow-up in a multicenter study of overdentures supported by Bra¨nemark implants. International Journal of Oral & Maxillofacial Implants 10: 33–42. Jaffin, R.A. & Berman, C.L. (1991) The excessive loss of Bra¨nemark fixtures in type IV bone: a 5year analysis. Journal of Periodontology 62: 2–4. Jemt, T., Book, K., Linden, B. & Urde, G. (1992) Failures and complications in 92 consecutively inserted overdentures supported by Bra¨nemark implants in severely resorbed edentulous maxillae: a study from prosthetic treatment to first annual check-up. International Journal of Oral & Maxillofacial Implants 7: 162–167. Keaveny, T.M., Borchers, R.E., Gibson, L.J. & Hayes, W.C. (1993) Theoretical analysis of the experimental artifact in trabecular bone compressive modulus. Journal of Biomechanics 26: 599. Lekholm, U. & Zarb, G.A. (1985) Patient selection and preparation. In: Bra¨nemark, P.I., Zarb, G.A. & Albrektsson, T., eds. Tissue Integrated Prostheses: Osseointegration in Clinical Dentistry, 199–209. Chicago: Quintessence Publ Co. Linde, F. & Hvid, I. (1989) The effect of constraint on the mechanical behaviour of trabecular bone specimens. Journal of Biomechanics 22: 485–490. Linde, F. & Hvid, I. (1999) Nondestructive mechanical testing of cancellous bone. In: Yuehuei, H.A.
& Draughn, R.A., eds. Mechanical Testing of Bone and the Bone–Implant Interface, 151–157. Boca Raton, FL: CRC Press. Lindh, C., Petersson, A. & Rohlin, M. (1996) Assessment of the trabecular pattern before endosseous implant treatment: diagnostic outcome of periapical radiography in the mandible. Oral Surgery, Oral Medicine, Oral Pathology 82: 335– 343. Misch, C.E. (1993) Density of bone: effect on treatment planning, surgical approach and healing. In: Misch, C.E., ed. Contemporary Implant Dentistry, 469–485. St Louis: Mosby Inc. Misch, C.E. (1999) Misch bone density classification. In: Misch, C.E., ed. Contemporary Implant Dentistry, 113–114. St Louis: Mosby Inc. Misch, C.E., Qu, Z. & Bidez, M.W. (1999) Mechanical properties of trabecular bone in the human mandible: implications for dental implant treatment planning and surgical placement. International Journal of Oral and Maxillofacial Surgery 57: 700–706. Nicholson, P.H.F., Lowet, G., Langton, C.M., Dequeker, J. & Van Der Perre, G. (1996) A comparison of time-domain and frequency-domain approaches to ultrasonic velocity measurement in trabecular bone. Physics in Medicine and Biology 41: 2421–2435. Nkenke, E., Hahn, M., Weinzierl, K., RadespiegelTro¨ger, M., Neukam, F.W. & Engelke, K. (2003) Implant stability and histomorphometry: a correlation study in human cadavers using stepped cylinder implants. Clinical Oral Implants Research 14: 601–609. Norton, M.R. & Gamble, C. (2001) Bone classification: an objective scale of bone density using the computerized tomography scan. Clinical Oral Implants Research 12: 79–84.
Stoppie et al . Characterization of jawbone
O’Mahony, A.M., Williams, J.L., Katz, J.O. & Spencer, P. (2000) Anisotropic elastic properties of cancellous bone from a human edentulous mandible. Clinical Oral Implants Research 11: 415–421. Patel, R., Blake, G.M. & Fogelman, I. (2004) An evaluation of the United Kingdom national osteoporosis society position statement on the use of peripheral dual-energy X-ray absorptiometry. Osteoporosis International 15: 497–504. Pattijn, V., Van Cleynenbreugel, T., Vander Sloten, J., Van Audekercke, R., Van Der Perre, G. & Wevers, M. (2001) Structural and radiological
parameters for the nondestructive characterization of trabecular bone. Annals of Biomedical Engineering 29: 1064–1073. Rho, J.Y., Ho Ba Tho, M.C. & Ashman, R.B. (1995) Relations of mechanical properties to density and CT numbers in human bone. Medical Engineering and Physics 17: 347–355. Schwarz, M.S., Rothman, S.L.G., Rhodes, M.L. & Chafetz, N. (1987a) Computed tomograph: part I. Preoperative assessment of the mandible for endosseous implant surgery. International Journal of Oral & Maxillofacial Implants 2: 137–141.
Schwarz, M.S., Rothman, S.L.G., Rhodes, M.L. & Chafetz, N. (1987b) Computed tomograph: part II. Preoperative assessment of the maxilla for endosseous implant surgery. International Journal of Oral & Maxillofacial Implants 2: 143–148. Snyder, S.M. & Schneider, E. (1991) Estimation of mechanical properties of cortical bone by computed tomography. Journal of Orthopaedic Research 9: 422–431.
133 |
Clin. Oral Impl. Res. 17, 2006 / 124–133
