Quantitative morphometry of the orbit in Chinese adults based on a three-dimensional reconstruction method

Abstract

The three-dimensional (3D) reconstruction technique serves as a practical tool in diagnosis, surgical planning, and outcome prediction of plastic and reconstructive surgery. In our study, the morphologic features of the Chinese adult orbit were described by 11 anatomic parameters using a 3D reconstruction technique. Sixty-four Chinese adults were selected randomly from patients who had undergone craniofacial computed tomography (CT) scans to diagnose conditions other than craniofacial or orbital deformations. The morphologic parameters of orbit such as bony orbital volume, orbital foramen area and orbital rim perimeter were measured on 3D models using this technique. Differences between the two orbits and between the two sexes were investigated. The method of measurement showed high reproducibility of results. No difference between the two orbits was found. There were significant differences between men and women in all anatomic parameters other than orbital height. In men and women, respectively, mean bony orbital volume was 26.02 and 23.32 mL, mean orbital foramen area 11.80 and 11.10 cm², mean orbital rim perimeter 12.65 and 12.20 cm, mean orbital height 33.35 and 33.22 mm, mean orbital width 40.02 and 38.00 mm; mean orbital floor length 47.93 and 46.18 mm, mean orbital roof length 52.93 and 50.89 mm, mean medial orbital wall length 46.43 and 44.41 mm, mean lateral orbital wall length 48.38 and 46.91 mm, mean intraorbital distance 27.18 and 25.11 mm, mean extra-orbital distance 98.77 and 93.69 mm. It is concluded that the measurements of these orbital parameters could be obtained from a 3D reconstruction method. The two orbits were symmetric based on orbital volume and other anatomic parameters. Orbital size was significantly smaller in women than in men; orbital height, however, was similar. The findings of the present study allow for quantification of the orbital features of Chinese adults and provide parameters for preoperative planning and prediction of postoperative outcome.

Introduction

The stereo-structure of the orbit is affected by several orbital diseases, e.g. congenital orbital dysplasia, orbital fracture, and intraorbital tumor. Deformation of the orbit results in apparent physical signs such as enophthalmos and exophthalmos, and may also lead to serious disequilibration of bilateral craniofacial development, especially in children. The main goals of plastic surgery for congenital orbital hypoplasia and orbital fracture are to repair the stereo-structure of the orbit and to reestablish the symmetric relationship between the two orbits. Empirical evaluation of orbital deformations is no longer satisfactory for the level of accuracy that can now be achieved in reconstruction surgery, and a quantitative morphometric method is needed. Three-dimensional (3D) reconstruction based on high-resolution spiral computed tomography (CT) scans has been used for more than 10 years to reveal the anatomic location and morphologic features of orbital abnormalities and plays a important role in diagnosis, surgical planning, and outcome prediction (Katsumata et al. 2005; Lopes et al. 2008). Application of the 3D technique in plastic and reconstructive surgery overcomes the limitations of two-dimensional scans, making it possible to observe craniofacial bones from various angles and to calculate the lengths and volumes of some parameters with certain software (Park et al. 2006). This technique has proven to have practicability and high accuracy (Regensburg et al. 2008), but relevant studies have focused only on measurements of the orbital volume and a complete and systematic set of anatomic parameters which describe the orbital features more precisely and quantitatively is lacking. In addition, the data acquired in previous studies are mainly from Caucasian subjects and cannot be applied to a Chinese population. Few reported studies have been performed in Chinese subjects.

Therefore, the purpose of our study was to calculate the orbital parameters of normal Chinese subjects using a 3D-reconstruction method to determine criteria for a morphologic description of the human orbit in Chinese adults.

Subjects and methods

Subjects

Chinese adults (n = 64; 30 men, 34 women) ranging in age from 18 to 50 years were randomly selected from patients examined at the Shanghai 9th People's Hospital who had undergone craniofacial CT scans to diagnose conditions other than craniofacial or orbital deformation. Subjects who had diseases that might affect the eye or orbit, such as thyroid disease, orbital fracture, intraorbital or intraocular tumor, congenital microphthalmia or anophthalmia, and orbitofacial cleft, were excluded from the study. The Medical Ethics Committee did not require informed consent for this study. The research adhered to the tenets of the Declaration of Helsinki.

CT data acquisition

Craniofacial scans of these subjects were obtained on a 64-row multi-slice CT (GE LightSpeed 16, Milwaukee, WI) using high-resolution contiguous sections in an axial plane for 3D reconstruction with the following protocol: tube voltage, 120 kvp; tube current, 100–140 mA; slice thickness, 1.25 mm; field of view, 25 × 25 cm; matrix, 512 × 512. All scan images were stored on a compact disc in DICOM format for offline analysis.

Calculation method

Image analysis was processed with simmed1.0 software (Pamilab, Shanghai Jiantong University, China) in two steps. First, multiplanar (coronal and sagittal) images and 3D-reconstructed images were displayed at the bone window on the basis of the original horizontal sections. Surface reconstruction of whole orbital content (WOC) was generated by 3D semi-automatic segmentation. The CT numbers for WOC segmentation were set from −250 H to +200 H. To ensure accuracy, manual segmentation was used to define the boundary of the orbit. The borders of the WOC were defined by the four orbital walls. The posterior boundary was the junction of medial and lateral walls at the optic foramen (Fig. 1). A simulated surface was determined that covered the orbital foramen with the orbital rim. We used this surface to separate the WOC into two parts, of which the posterior part was defined as the bony orbit. In this way, the area of the orbital foramen and the perimeter of the orbital rim could be calculated automatically (Fig. 2). Anatomic landmarks of the orbit (n = 5; Table 1; Fig. 3A,B) were located on the 3D model of the orbit and the relevant length measurements were calculated (n = 8; Table 1; Fig. 3C) using software tools.

Fig. 1.

Axial (A), coronal (B), and sagittal (C) views of cranio-orbital CT scans during segmentation of the whole orbital content (WOC). Three-dimensional model of the orbit (D) and WOC (E).

Fig. 2.

Curve along the orbital rim (A, B). Simulated surface of the orbital foramen (C,D). Three-dimensional model of the bony orbit (E).

Table 1.

Definition of anatomical landmarks of the orbit and relevant length parameters

Orbital landmarks/Orbital length parameters	Definition
Maxillofrontal (Mf)	Junction between frontomaxillary suture and medial orbital rim
Ectoconchion (Ec)	Junction between the lateral orbital rim and the horizontal line that divides the orbital foramen into two equal parts
Supraorbital point (Os)	Superior junction between the superior orbital rim and the perpendicular bisector line of line Mf-Ec
Infraorbital point (Oi)	Inferior junction between the inferior orbital rim and the perpendicular bisector line of line Mf-Ec
Point on the optic foramen (Of)	Optic foramen
Orbital breadth	Mf-Ec
Orbital height	Os-Oi
Medial orbital wall length	Mf-Of
Lateral orbital wall length	Ec-Of
Orbital roof length	Os-Of
Orbital floor length	Oi-Of
Intraorbital distance (IOD)	Mf-Mf
Extraorbital distance (EOD)	Ec-Ec

Fig. 3.

(A,B) Anatomical landmarks located on the three-dimensional model of the orbit. (C) Length parameters of the orbit: black lines on the left orbit indicate orbital height and breadth; red lines on the right side indicate the lengths of the four orbital walls; blue lines indicate the intraorbital distance (IOD) and extraorbital distance (EOD).

Assessments of reproducibility

In this study, 10 left orbits of male subjects were used to assess the reproducibility of the measurements. For intraobserver variability, one observer, observer A, calculated all of the measurements of these 10 orbits twice on different days. For interobserver variability, another observer, observer B, performed the same measurements independently on the same 10 orbits. Both the intraobserver variability and interobserver variability were analyzed by calculating the intraclass correlation coefficient.

Statistical analysis

The statistical software spss 14.0 for Windows (SPSS Inc, Chicago, IL) was used for statistical analysis. The measured values are presented as mean ± SD. To assess reproducibility, the intraclass correlation coefficients were calculated for all parameters to determine the intraobserver variability (between two measurements made by observer A) and interobserver variability (between the measurements of the same orbits by observers A and B). A two-way mixed-effects model (0 = no agreement, 1 = perfect agreement) was used to present the intraclass correlation coefficients as previously reported (Regensburg et al. 2008). The intraclass correlation coefficient was considered to be adequate if it was > 0.75 (Bland & Altman, 1990). Comparisons between the two orbits in the same subject were evaluated using a paired samples t-test. Comparisons between the two sexes were performed using a grouped t-test. A P value of < 0.05 was considered to be statistically significant.

Results

Data from 64 subjects (128 orbits) were included in this study. The average age of the 64 subjects was 30.0 years (range 18–50). The male group comprised 30 subjects with a mean age of 31.1 ± 10.0 years. The female group comprised 34 subjects with a mean age of 28.9 ± 7.8 years. There was no statistical difference in the age composition of the two groups (t = 0.97, P = 0.34). Repeated measurements of the same orbits by the same observer to test the intraobserver variability revealed that the intraclass correlation coefficients for each parameter were > 0.98, indicating stable accuracy of each measurement. Repeated measurements of the same orbits by two observers to test the interobserver variability revealed intraclass correlation coefficients of > 0.95 for each parameter. This high interobserver correlation showed a low variability in the measurements between different observers (Table 2). The measured values of the two orbits are listed in Table 3. Comparisons of the parameters using a paired sample t-test showed P values all > 0.1 and provided no evidence of laterality for any parameters of the Chinese human orbit. The results of the orbital morphometrics for the males and females are summarized in Table 4. The average measured values of the male orbits were larger than those of the females for all parameters, especially the bony orbital volume (BOV; 26.02 and 23.32 mL, P < 0.001), area of the orbital foramen (11.80 and 11.10 cm², P = 0.001), perimeters of the orbital rim (12.65 and 12.20 cm, P < 0.001), orbital width (40.02 and 38.00 mm, P < 0.001), and EOD (98.77 and 93.69 mm, P ≤ 0.001). There was no difference in orbital height between groups (33.35 and 33.22 mm, P = 0.742).

Table 2.

Intraclass correlation coefficient (ICC) of measurements by Observer A (measurements 1 and 2) and Observer B (measurement 3)

	ICC (1–2)	ICC (1–3)	ICC (2–3)
Bony orbital volume	0.989	0.958	0.983
Orbital foramen area	0.989	0.967	0.978
Orbital rim perimeter	0.992	0.975	0.972
Orbital height	0.996	0.997	0.993
Orbital breadth	0.986	0.972	0.953
Orbital floor length	0.991	0.997	0.987
Orbital roof length	0.994	0.995	0.982
Medial orbital wall length	0.989	0.986	0.986
Lateral orbital wall length	0.989	0.993	0.975
Intraorbital distance	0.984	0.986	0.974
Extraorbital distance	0.997	0.996	0.991

ICC outcome values: 0, no agreement; 1, complete agreement.

Table 3.

Comparison of measured values between the two orbits (n = 64)

	Left (Mean ± SD)	Right (Mean ± SD)	L-R (Mean ± SD)	t	P
Bony orbital volume (mL)	24.59 ± 2.61	24.61 ± 2.76	−0.02 ± 0.97	−0.15	0.881
Orbital foramen area (cm2)	11.43 ± 0.86	11.50 ± 0.87	−0.07 ± 0.33	−1.65	0.105
Orbital rim perimeter (cm)	12.41 ± 0.49	12.39 ± 0.51	0.03 ± 0.19	1.21	0.232
Orbital height (mm)	33.28 ± 1.58	33.45 ± 1.63	−0.17 ± 0.92	−1.44	0.155
Orbital breadth (mm)	38.94 ± 1.88	39.10 ± 1.83	−0.16 ± 0.79	−1.66	0.109
Orbital floor length (mm)	47.00 ± 2.75	46.85 ± 2.81	0.15 ± 0.73	1.63	0.107
Orbital roof length (mm)	51.84 ± 2.77	51.67 ± 2.56	0.18 ± 0.89	1.60	0.115
Medial orbital wall length (mm)	45.36 ± 2.71	45.18 ± 2.90	0.18 ± 1.52	0.94	0.353
Lateral orbital wall length (mm)	47.60 ± 2.47	47.77 ± 2.33	−0.17 ± 1.45	−0.91	0.365

Table 4.

Comparison of measured values between men and women

	Men (Mean ± SD)	Women (Mean ± SD)	t	P
Age (years)	31.13 ± 9.99	28.94 ± 7.77	0.97	0.336
Bony orbital volume (mL)	26.04 ± 2.60	23.32 ± 1.87	4.85	< 0.001
Orbital foramen area (cm2)	11.80 ± 0.78	11.10 ± 0.81	3.53	0.001
Orbital rim perimeter (cm)	12.65 ± 0.45	12.20 ± 0.43	4.07	< 0.001
Orbital height (mm)	33.35 ± 1.44	33.22 ± 1.73	0.33	0.742
Orbital breadth (mm)	40.02 ± 1.63	38.00 ± 1.56	5.06	< 0.001
Orbital floor length (mm)	47.93 ± 2.68	46.18 ± 2.57	2.67	0.010
Orbital roof length (mm)	52.93 ± 2.89	50.89 ± 2.30	3.15	0.003
Medial orbital wall length (mm)	46.43 ± 2.67	44.41 ± 2.41	3.19	0.002
Lateral orbital wall length (mm)	48.38 ± 2.50	46.91 ± 2.25	2.47	0.016
Intraorbital distance (mm)	27.18 ± 2.76	25.11 ± 2.25	3.31	0.002
Extraorbital distance (mm)	98.77 ± 3.91	93.69 ± 4.09	5.06	< 0.001

Discussion

The orbit is a craniofacial structure that can be affected by a large number of congenital, traumatic, neoplastic, vascular, and endocrine disorders. In these cases, the measurement of BOV and a description of the orbital shape may have crucial clinical applications for estimating craniofacial asymmetry, the severity of injury, and possible complications in preoperative planning and in post-operative evaluation. Several methods for assessing BOV have been proposed. The water-filling method is considered to be the gold standard criterion for volume measurement, but can only be used in cadaver skulls (Forbes et al. 1985; Acer et al. 2009). The magnetic resonance imaging scan is favored due to the lack of radiation exposure, especially in cases of children and normal subjects (Chau et al. 2004), but CT scans delineate better the bony structures, and are thus more widely used (Perry et al. 1998; Fan et al. 2007). With CT images printed on films in square frames, a point-counting method can easily be used to measure the surface area of the orbital cavity in each section (Acer et al. 2009). With a known section thickness of the consecutive sections, the BOV can be calculated using a formula for volume estimation of radiologic images. This method is simple but rough. The point-counting method cannot be used to measure linear parameters or curved parameters, however, such as the lengths of the orbital walls and orbital rim perimeter, which need to be measured simultaneously. During the past few years, a CT-based 3D-reconstruction method has demonstrated advantages for direct measurements on a 3D model and has therefore led to a trend to substitute the 3D reconstruction method for two-dimensional measurements (Park et al. 2006). The reliability and accuracy of this method have been verified in a previous study (Regensburg et al. 2008). The test for the reliability of the software used in our study also showed stability of each measurement and high reproducibility of this measurement technique.

The orbital shape is like a cone, in which the apex is defined as the optic foramen, the base as the orbital foramen formed by the plane extending from the supraorbital rim to the infraorbital rim in the superoinferior direction, and the frontozygomatic suture to the anterior lacrimal crest in the lateromedial direction. BOV is a common parameter used to estimate orbital changes or abnormalities, but it is not the appropriate parameter for describing the shape of the orbit. In studies of orbital dysplasia, the shape of the orbit is important to evaluate not only for diagnosis, but also during the course of treatment. Therefore, in the present study, a set of parameters, including BOV, was determined to describe the main characteristics of the orbit. Among these parameters, the linear parameters are mostly based on the anatomic landmarks of the orbit; however, due to large variations in the location among individuals, the superior orbital fissure and the inferior orbital fissure were not considered reliable landmarks in our study. Instead, we placed orbital width in a horizontal position (line Mf-Ec), and the orbital height in a vertical position (line Os-Oi) to standardize the measurements of these two parameters in individuals. The orbital rim, as the margin of the orbital foramen, is a superficial structure that determines the orbitofacial appearance and maintains the symmetry between sides. The shape of the orbital rim is considered to be rounder in females, and more squared in males. A recent study verified the significance of orbital rim shape in the assessment of sexual dimorphism and it is considered a morphometric trait in sex determination (Pretorius et al. 2006). The geometric morphometrics used previously were based on several landmarks on the rim to roughly imitate the curve of the rim. The exact perimeter of the rim was not measured, because it is not a regular curve. In our study, a curve was drawn continuously on the 3D model along the rim and a surface was subsequently simulated to cover the orbital foramen with the margin determined; therefore, the perimeter of the orbital rim and the area of the orbital foramen were calculated automatically. The mean perimeter of the orbital rim and mean area of the orbital foramen were significantly larger in men than in women, which is consistent with previous studies. Moreover, according to the results of our study, nearly all orbital parameters were significantly larger in men than in women (except for orbital height). Therefore, sexual dimorphism in the orbital rim was confirmed, and the whole set of orbital parameters can also be used as a sexually dimorphic trait.

With respect to the laterality of the orbits in individuals, Forbes et al. (1985) reported a minimal difference in orbital volume between the right and left sides. Orbital volume, however, was also the only parameter to show a difference or equality between the two orbits in most of the previous reports. In the present study, there were no differences in the anatomic parameters between the two sides in the same individual. Thus, the symmetry of the two orbits in both volume and shape was validated. No laterality of the orbital features was detected in a normal Chinese adult population.

The average orbital volume of a Chinese adult was 26.02 mL in men and 23.32 mL in women in our study, similar to the results of Ye et al. (2006) in Korean subjects (23.94 ± 3.47 mL) and the results of Regensburg et al. (2008) in Caucasians (25.17 ± 0.06 mL), but larger than those reported by Furuta (2001) in Japanese subjects (23.6 ± 2.0 mL for males, 20.9 ± 1.3 mL for females), by Chau et al. (2004) in Hong Kong Chinese (22.20 ± 1.38 mL for males, 19.81 ± 2.23 mL for females) and by Acer et al. (2009) in Turkish subjects (17.84 mL by water-filling method and 17.05 mL by point-counting method). Our values were smaller than those reported by Deveci et al. (2000) in Caucasians (28.41 ± 2.09 mL). The variation might be due to the ethnic differences, the different measurement methods, as well as to the sample size. Chau used magnetic resonance images, which may differ from CT images in defining the boundary of the bony orbit. The measurements of orbits by M. Deveci M and N. Acer were from dried skulls, which should not be the same as living subjects. The other anatomic parameters presented in our study were not mentioned in previous articles, so there are no data available for comparison. Because the statistical analysis in our study was based on direct measurements of the 3D models, the data acquired should be more reliable than those simulated from two-dimensional scans in previous studies. Moreover, the sample size in our study was much larger than in previous studies, so the results may be more representative.

Conclusions

The present study lists the group of anatomic parameters describing the orbital features. Calculation of these statistics could be manipulated using a 3D-reconstruction method whose reproducibility was validated in our study. Orbital volumes and other anatomic parameters demonstrated bilateral symmetry of the orbits. Orbits were significantly smaller in women than in men, although there was no difference in orbital height. All these data were obtained from normal Chinese adults and may help to develop a database to determine normal orbit values. This technique can be used for quantitative assessment of orbital disease and orbitofacial deformities, both for preoperative planning and for assessing postoperative outcome. As our sample size was limited, racial and regional factors were not considered in the current study. Therefore, the measurements presented can not be generalized for use as national parameters. The present findings are particularly important because there are no similar investigations aimed at Chinese adults in previous studies. Future studies are planned to improve this measurement technique and to application of this method to the investigation of pathologic orbits.