Background/Aims Up to date, no standardised reproducible orbital volume measurement method is available. Therefore, this study aimed to investigate the accuracy of a new measurement method, which delineates the boundaries of orbital cavity three-dimensionally (3D).
Methods In order to calculate the orbital volume from axial CT slice images of the patients, using our first described measurement method, the segmentation of the orbital cavity and the bony skull was performed using Amira 3D Analysis Software. The files were then imported into the Blender program. The stereographic skull model was aligned based on the Frankfurt horizontal plane and superposed according to defined anatomical reference points. The anterior sectional plane ran through the most posterior section of the lacrimal fossa and the farthest dorsal point of the anterior latero-orbital margin, which is positioned perpendicular to the Frankfurt horizontal plane. The volume of each orbital cavity was then determined automatically by the Blender program.
Results The 10 consecutive subjects (5 female, 5 male) with mean age of 50.3±21.3 years were considered for analysis in the current study. The first investigator reported a mean orbital volume of 20.24±1.01 cm3 in the first and 20.25±1.03 cm3 in the second evaluation. Furthermore, the intraclass correlation coefficient (ICC) showed an excellent intrarater agreement (ICC=0.997). Additionally, the second investigator detected a mean orbital volume of 20.20±1.08 cm3 in his assessment, in which an excellent inter-rater agreement was found in ICC (ICC=0.994).
Conclusions This method provides a standardised and reproducible 3D approach to the measurement of the orbital volume.
Statistics from Altmetric.com
Measuring orbital volume is difficult due to the complex anatomy of the eye socket.1 However, the ability to measure orbital volume and its changes is helpful for the analysis of orbital fractures and for the evaluation of their reconstruction. Fractures of the orbital cavity can lead to diplopia or hypoglobus, and may cause enophthalmos2–5 which is defined as an eye recessed by more than 2 mm in comparison with the other one and is aesthetically noticeable.2 ,6 ,7 An increase of 1 cm3 in post-traumatic orbital volume results in a enophthalmos of approximately 1 mm (interval: 0.62–1.6 mm).8–13 Orbital volume can be measured and restored to the desired state in order to prevent these symptoms or to treat them with higher accuracy.14
In order to measure orbital volume in 1873, Gayat15 inserted lead pellets into the orbital cavities of ex vivo human skulls. Alternative methods for measurement on corpses using silicone, water, glass beads or sand have also been described.8 ,16–20 In 1985, Forbes et al17 described a method of orbital volume calculation using CT images. Since then, several methods have been described to measure the volume of the orbital cavity for the patients by the development of layered imaging techniques, CT and MRI.6–8 ,16 ,17 ,20–22 In the recently published studies, administration of CT images was recommended as the standard method to determine the volume of the orbital cavity in living patients.2 ,7 ,14 ,23–25 However, the accuracy of measurement with CT imaging compared with conventional methods has been investigated multiple times.20 ,22
With a thickness of only 0.3 mm laterally and 0.5 mm caudally, the bones forming eye socket are very thin.26 As a result, they are only partially shown even on high resolution, 1 mm Multislice CT Scans, which complicates the delineation of the orbital cavity, because the difference in contrast between the orbital soft tissues and the bony orbital wall is too low. To date, mostly coronary or axial CT slice images have been used to determine orbital volume.6 ,7 ,14 ,23 ,27
Due to the lack of evaluation regarding the accuracy of existing orbital volume measurement methods, the confusing anatomical boundaries of the orbital cavity and the variability in measurement limits of each CT slice image, we attempted to apply a new interpersonally/intrapersonally controlled, stereographic orbital volume measurement method, in which the boundaries of the orbital cavity––shown on the CT slice image––are three-dimensional (3D) rather than two-dimensional. Based on these considerations, the aim of this study was to determinate the accuracy, the reproducibility and measurement errors among different measurements and to compare the obtained values from two blinded experienced investigators.
Materials and methods
Validation of the method
The study was approved by the Ethics Committee of the Canton of Bern (Switzerland). An experienced member of staff calculated the orbital volume from axial CT slice images of 10 patients (20 healthy orbital cavities). In order to test the proportion of agreement (intraobserver/interobserver agreement), the primary rater and another independent rater re-evaluated the same CT data sets after 4 weeks.
Inclusion and exclusion criteria
This study included consecutive patients who had received a cranial CT during January and February 2013 at Inselspital (Bern University Hospital, Switzerland). The CT data sets were required to have a minimal continuous slice thickness of 1 mm and had to exhibit a ‘bone window’. Patients with indications of pathology and/or consequences of trauma to the orbital cavities and/or the neighbouring structures, as well as preceding orbital or sinus surgery, were excluded. Additionally, the patients with congenital or acquired craniofacial malformations were not considered for the analysis in the study.
Image generation and analysis
Facial CTs at the Inselspital, Bern University Hospital (Switzerland), were used. The segmentation of the orbital cavity and bony skull was performed with three different computers using Amira 3D Analysis Software for Life Sciences V.5.0 (FEI Company, Oregon, USA). In order to measure the orbital volume, the bony orbit boundary was segmented on each individual image.
The Lasso tool was used for this purpose, using the mouse to avoid the boundaries of the orbital cavity and the anterior soft tissues. This was carried out for each native CT slice. The boundaries in the two reconstructed planes were then inspected and corrected. As a result, uniform results for the sagittal, frontal and transverse planes were achieved (figure 1). Because Kwon et al28 showed that volumes may vary according to the sectional plane, this step was carried out to improve accuracy. If the lateral wall was not visible because it lies in a different plane, the boundaries were searched for on the adjacent images. These limits were then carried over and subsequently inspected in the other two planes.
The specific areas of interest in the orbital foramina, optic canal, nasolacrimal canal and superior and inferior orbital fissures were aligned at their smallest circumference using a straight line. The surfaces thus obtained were then assembled into a 3D model of the orbital cavity including the anterior soft tissues, using Amira software (figure 2). In a separate step, the bony structures of the skull were assembled into a 3D model in Amira, based on the threshold value segmentation.
The obtained data were then imported into the Blender program V.2.7 (Blender Foundation, Amsterdam, The Netherlands). The stereographic skull model was aligned in the space, so that the Frankfurt horizontal plane came to rest on either side of the horizontal plane. ‘The Frankfurt horizontal plane is determined as the upper margin of the Porion (highest point of the bony ear canal opening) and the lowest point of the orbitals’.29
The anterior sectional plane ran through the most posterior section of the lacrimal fossa and the farthest dorsal point of the anterior latero-orbital margin, which is positioned perpendicular to the Frankfurt horizontal plane. Then, the anterior portion of this sectional plane was removed, leaving an area located posterior to it. In order to obtain a clean 3D, posterior boundary, a perpendicular plane was placed through the orbital cavity at the exit point of the optical nerve. The area located posterior to this was also subtracted. The volume of each orbital cavity was then determined automatically by the program Blender (figure 3).
All analyses were conducted using the Statistical Package for the Social Sciences V.21.0 (SPSS, Chicago, Illinois, USA). Descriptive statistics were presented for the patients as mean±SD. All p values relate to two-sided tests with a level of 0.05. A non-parametric Mann Whitney test was used to detect differences between groups. Furthermore, the intraclass correlation coefficient (ICC) (two-way random model) was computed to estimate the inter-rater/intrarater reliability. Value of ICC >0.7 in absolute single measures is considered to be an acceptable agreement.
The 10 consecutive subjects (5 female and 5 male) with mean age of 50.3±21.3 years (22–88 years) were considered for the analysis in the current study. Total amount of orbital volume measured were 20 for each measurement cycle. Mean orbital volume in this study was 20.23±1.02 cm3 (table 1).
The first investigator reported a mean orbital volume of 20.24±1.01 cm3 in the first and 20.25±1.03 cm3 in the second evaluation, which was not significantly different (p=0.989). Furthermore, the ICC showed an excellent intrarater agreement (ICC=0.997, p<0.001).
Additionally, the second investigator detected a mean orbital volume of 20.20±1.08 cm3 in his assessment, which was also not statistically significant compared with the first evaluation (p=0.989), and an excellent inter-rater agreement was found in ICC (ICC=0.994, p<0.001) (table 2).
In recent years, coronary or axial CT slice images have gained increasing significance as a method to determine orbital volume.6 ,7 ,14 ,23 ,27 However, no standardised reproducible orbital volume measurement method is available. In fact, the described evaluation techniques are mostly using own algorithms, the reference points are not clearly defined and the methods not fully described (table 3). For that reason, the measurements are directly related to the investigating institution. As a result, the current study recommended a new stereographic orbital volume measuring method, whose acceptable intrarater/inter-rater agreement was shown by an ICC>0.7. This method is facilitated by the 3D orientation, the standardised reference points and strict segmentation guidelines.
The measurement of orbital volume is clinically relevant since an orbital fracture often leads to an increase in orbital volume. An orbital fracture can lead to functional and aesthetic deformities, such as diplopia or hypoglobus, and may cause enophthalmos.2–5 Orbital volume must be restored correctly in order to prevent, or treat, these symptoms.14 However, the degree of precision and symmetry needed to obtain the best orbital reconstruction is still unknown. In fact, the relation of the volume reconstruction's precision to the post-traumatic orbital soft tissue changes and long-term results is up to date not investigated. The first step to answer these clinical relevant questions is to develop a standardised and reproducible measurement method.
As no standard method of orbital volume measurement exists in the literature,7 ,16 ,17 ,20 ,28 no meta-analysis and no direct comparison among the different measurement methods are possible. Kwon et al showed that—depending on the plane in which the images are evaluated—other reference points to determine the anterior orbital boundary must be used. As a result, volume measurements can vary significantly.28 In many different publications, it is difficult to discern which reference points were used to determine orbital volume6 ,7 ,14 ,16 ,21 ,23 ,27 ,30 (table 3). We consider three points to necessarily achieve the best possible reproducibility of measurements: (1) the orientation of the skull in the 3D space, according to the plane described above; (2) the definition of distinctive landmarks to create measurement limits and (3) an exact algorithm for virtual closure of the orbital foramen.
The most notable limitation of this method is that it will not include the total anatomical soft tissue of the orbital cavity, which has been measured on deceased individuals many times.17 ,20 This anterior loss of volume cannot be avoided without losing accuracy. Furthermore, the stereographic method is time-consuming (about 2 hours per measurement) and also technically complex. However, the clinically relevant range of the bony orbital funnel can be determined accurately. The advantage of using a multiplanar segmentation method is the possibility to regulate the orbital boundaries in all three planes and convert the CT slice images into a 3D model early on in the editing process. This allows to define the anterior and posterior boundaries of the orbital cavity in 3D. Two-dimensional determination, however, is problematic because the anatomical reference points are not visible on all CT slices, and approximations must be made.16 ,21 The 3D skull orientation based on the Frankfurt plane makes this method independent of head alignment during image acquisition.
Finally, this reproducible evaluation method serves as an instrument for clinical studies as well as quality control evaluations after different orbital reconstruction approaches using various implant types. Furthermore, it can be very helpful in the treatment of unilateral orbital fractures. Once a complete set of orbital volume models has been gathered and aligned, the healthy side is mirrored, and a manifold 3D model with the fractured and the mirrored orbit can be created and 3D-printed. The 3D-printed model can be used to prebend an ‘off-the-shel f implant’. As a result, a low-cost, patient-specific implant for orbital reconstruction can be prepared before surgery. The stereographic orbital volume measurement after reconstruction with ‘off-the-shelf implant’ compared with patient-specific implants is the subject of ongoing research.
The present stereographic orbital volume measurement method using Amira and Blender software to analyse CT scans provides accurate and reproducible results. The position of the head is stereographically calibrated, and the orbital volume was defined by using three anatomical reference points. Its implementation in basic research on digital orbital volume measurements, in ophthalmopathy, in preoperative planning of orbital cavities interventions and in quality control following orbital reconstructions is proposed and recommended.
MM and CAW contributed equally.
Contributors MM and CAW contributed to the design, data acquisition, data analysis and interpretation, wrote the first draft and made revisions to the manuscript. SMSJ contributed to the data analysis and interpretation and revision of the manuscript. KK and BS contributed to the implementation of the design, data acquisition, interpretation and revisions of the draft manuscript.
Competing interests None declared.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.