From the Indiana University School of Medicine, Indianapolis, Indiana.
The authors have no financial or proprietary interest in the materials presented herein.
Address correspondence to Dr. Matthew Hauck, 402 N. Meridian St. #408, Indianapolis, IN 46204.
Exophthalmometry is an important tool in evaluating orbital processes including thyroid-associated ophthalmopathy, orbital trauma, and orbital tumors. Various devices for measuring the axial position of the eye within the orbit exist, but the Hertel exophthalmometer is the most widely used.1–3 The Hertel exophthalmometer is a useful clinical tool, which is convenient and easy to use in evaluating orbital processes.
Orbital imaging via computed tomography (CT) or magnetic resonance imaging is often indicated in the evaluation and management of orbital processes, especially when vision is threatened or surgery is contemplated. Increased availability of these imaging modalities has expanded their utility. Currently, these studies are used primarily to corroborate clinical findings and to delineate deeper processes not apparent on external examination. We questioned if radiographic imaging could be reliably used to quantify globe position.
Previous investigators have considered exophthalmometry using radiographic images; yet overall, the literature is relatively scarce.4,5 We attempt to further define the role and evaluate CT image-based exophthalmometry.
Materials and Methods
Exophthalmometry measurements were obtained digitally using a GE™ PACS radiograph viewing system and with conventional Hertel testing in 53 orbits (44 thyroid-associated ophthalmopathy and 9 post-trauma). Institutional review board approval was obtained. Corresponding clinical charts were reviewed to obtain Hertel readings. Orbits with non-stable disease were excluded to prevent recording differences in readings that represented true changes in the position of the eye within the orbit. Specifically, orbits with acute or subacute processes such as edema, abscess, hemorrhage, or intervening surgery were excluded. Patients with swelling or tenderness at the lateral canthus were excluded for the possibility of error in Hertel as described by Vardizer et al.3
CT measurements were obtained by selecting the axial plane that included the most anterior cut through the eye. This was typically the corneal apex. On this section, a line was drawn connecting the left and right lateral orbital rims. The distance from the anterior cornea to this line, perpendicularly, was recorded with the viewer software (Fig. 1). The parameters were similar to those described by Gibson.4
Figure 1. Technique Used for CT Exophthalmometry
The measurement by imaging software was compared with Hertel exophthalmometry performed by an experienced examiner. Patients were measured in the upright position with the eyes in the primary position in utilizing the narrowest possible base or intercanthal distance. The observers were mutually masked to the measurements obtained with the other method.
The mean difference between Hertel and CT measurements was calculated with a 95% confidence interval (CI) for the mean difference. To measure the agreement between Hertel and CT metrics, an intraclass correlation coefficient (ICC) was calculated. The ICC was assigned a number between 0 and 1, with an ICC of 0 indicating no agreement and an ICC of 1 indicating perfect agreement.
The mean measurement by Hertel and CT were 22.7 and 22.67 mm, respectively. The mean exophthalmometry by Hertel and CT in the thyroid-associated ophthalmopathy group were 23.9 and 23.8 mm, respectively. The mean values of by Hertel and CT in the post-trauma group were 17.1 and 17.1 mm, respectively. 33 of 53 CT measurements were within 1 mm of the value recorded from Hertel testing (62.3%). 46 of 53 CT measurements were within 2 mm of the Hertel values (86.7%). 1 of 53 CT measurements had a difference of 3 mm or more when compared to Hertel values (1.8%) (Table 1).
Table 1: Hertel and CT Measurement Data
The CT metrics, on average, provided exophthalmometry that was 0.03 mm shorter than Hertel (95% CI). The ICC was 0.95 (95% CI) for the two forms of measurement, indicating excellent correlation.6
CT exophthalmometry demonstrated excellent correlation with Hertel exophthalmometry in this series of thyroid-associated ophthalmopathy and orbital trauma patients. We conclude that CT exophthalmometry may offer reliable data that can supplement or proxy (when not available) clinical Hertel measurements.
While CT exophthalmometry will never replace Hertel readings, it may provide additional data. Radiation exposure, cost, and the inconvenience of repeated neuroimaging preclude the use of CT solely for exophthalmometry readings; yet, many clinical scenarios exist where CT images are already present and CT exophthalmometry readings can be useful. Use of CT exophthalmometry might be as simple as its use in whom orbital imaging has either already been obtained or is indicated. When a Hertel exophthalmometer is not available or its use is not feasible, as in patients with acute orbital disease or trauma presenting to the emergency department, the CT reading might provide a reasonable proxy measurement. CT exophthalmometry readings might also be helpful in evaluating new patients by comparing new Hertel readings with older CT exophthalmometry readings. The advent of telemedicine offers additional applications either by reviewing scans collaboratively within one institution or exchanging information remotely across hospital systems.
We acknowledge several potential limiting factors. In addition to radiation exposure, postural effects and imaging planes may limit CT exophthalmometry. Freuh et al. demonstrated postural effects on the Hertel exophthalmeter.7 Supine positioning, which is standard in conventional orbital CT scanners, may similarly introduce variability, especially when comparing with Hertel readings, which were obtained with patients upright.
Imaging planes and globe rotation may also introduce error. If the orbital CT image is off the precise axial plane, a longer anterior-posterior measurement would be expected. Similarly, globe rotation and eyelid closure might introduce inaccuracies. Despite these potential causes for inaccuracy, our data showed good correlation between the two techniques. Further investigation is necessary to elucidate the effects of these parameters.
Data stratification for underlying disease (thyroid-associated ophthalmopathy vs trauma) did not provide subset dissimilarities between CT exophthalmometry and Hertel. Moreover, the intraclass correlation was high at all levels of exophthalmos. However, we acknowledge that a wider range of patients and more subjects would strengthen the evidence. Additionally, prospective data would be useful.
In summary, CT exophthalmometry was comparable with Hertel in this series of thyroid-associated ophthalmopathy and trauma patients. With the increasing availability and use of radiographic imaging in the orbital evaluation, CT exophthalmometry readings may provide additional relevant data useful to the clinician. Orbital MRI may offer similar data and without radiation exposure, but its efficacy in exophthalmometry needs further investigation and is beyond the scope of this paper. The overall role of radiographic imaging in quantifying orbit and globe relationships needs further clarification, but CT scans may provide globe position metrics that compare well with clinical measurements and may prove useful as an adjunctive way to measure globe position within the orbit.
- Musch DC, Frueh BR, Landis JR. The reliability of Hertel exophthalmometry. Observer variation between physician and lay readers. Ophthalmology. 1985;92:1177–1780.
- Sleep TJ, Manners RM. Interinstrument variability in Hertel-type exophthalmometers. Ophthalmic Plast Reconstr Surg. 2002;18:254–257. doi:10.1097/00002341-200207000-00004 [CrossRef]
- Vardizer Y, Berendschot TT, Mourits MP. Effect of exophthalmometer design on its accuracy. Ophthalmic Plast Reconstr Surg. 2005;21:427–430. doi:10.1097/01.iop.0000180066.87572.39 [CrossRef]
- Gibson RD. Measurement of proptosis (exophthalmos) by computerised tomography. Aus Radiol1984;28:9–11. doi:10.1111/j.1440-1673.1984.tb02462.x [CrossRef]
- Segni M, Bartley GB, Garrity JA, et al. Comparability of proptosis measurements by different techniques. A J Ophthalmol. 2002;133:813–818. doi:10.1016/S0002-9394(02)01429-0 [CrossRef]
- Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics. 1977;33:363–374. doi:10.2307/2529786 [CrossRef]
- Frueh BR, Garber F, Grill R, Musch DC. Positional effects on exophthalmometer readings in Graves’ eye disease. Arch Ophthalmol. 1985;103:1355–1356.
Hertel and CT Measurement Data
|Patient Characteristics||Hertel (mm)||CT (mm)||Difference (CT-Hertel)|
|1||39||F||Graves type I||26||26||24.1||24.3||−1.9||−1.7|
|2||65||M||Graves type II||21||16||21.9||18.4||0.9||2.4|
|3||64||F||Graves type II||25||26||24.5||23.3||−0.5||−2.7|
|4||34||M||Graves type I||22||24||23.2||23.2||1.2||−0.8|
|5||71||F||Graves type II||24||28||22.6||26.4||−1.4||−1.6|
|6||48||F||Graves type II||22||20||21.5||19.3||−0.5||−0.7|
|7||54||F||Graves type II||24||25||23.4||25.5||−0.6||0.5|
|8||54||F||Graves type II||19||23||20.3||22.3||1.3||−0.7|
|9||53||F||Graves type II||27||27||28.5||27.6||1.5||0.6|
|10||38||M||Graves type II||26||23||25||22||−1||−1|
|11||26||M||Graves type I||23||22||20.1||21||−2.9||−1|
|13||29||F||Graves type I||21||23||23.1||23.3||2.1||0.3|
|14||26||F||Graves type II||23||23||24.9||23.9||1.9||0.9|
|15||33||F||Graves type II||20||24||19.2||20.8||−0.8||−3.2|
|16||66||M||Graves type II||27||22||25.9||22||−1.1||0|
|18||64||F||Graves type II||20||21||20.3||21.2||0.3||0.2|
|19||61||M||Graves type II||31||32||31.7||32.6||0.7||0.6|
|20||66||M||Graves type II||25||25||25.7||26||0.7||1|
|22||45||F||Graves type II||20||18||21.2||17.7||1.2||−0.3|