Clinicopathologic Reports, Case Reports, and Small Case Series
December 2004

Normalization of Upper Eyelid Height and Contour After Bony Decompressionin Thyroid-Related Ophthalmopathy: A Digital Image Analysis

Author Affiliations



Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004

Arch Ophthalmol. 2004;122(12):1882-1885. doi:10.1001/archopht.122.12.1882

The treatment of eyelid retraction in Graves disease is one of the mostchallenging aspects of ophthalmic plastic surgery. There are numerous theoriesabout the etiology of upper eyelid retraction, including overaction of thelid retractor in combination with fibrosis of the inferior rectus muscle,fibrosis of the lacrimal gland and adjacent levator aponeurosis, enlargementof the levator fibers, and increased sympathetic tone in the superior tarsalmuscle.14

Several anatomical explanations account for the lateral accentuationof upper eyelid retraction that occurs in Graves disease. In 1980, Grove2 noted fibrosis in the lacrimal gland and adjacentlateral levator aponeurosis, perhaps indicative of more lateral eyelid retractorshortening. In 1991, Lemke5 noted that theforces that affect the upper eyelid are governed by variations in orbitalsize and shape, globe size and position, and the length-tension characteristicsof the eyelid structures. Enhanced lateral upper eyelid retraction occursin part because the eye in primary position must adduct away from the axialprojection of the orbit and exposes more lateral sclera. This retraction isaccentuated by axial proptosis.

To further investigate the influence of the anatomical position of theglobe on eyelid retraction and lateral accentuation of retraction, we analyzeddigital photographs of patients’ eyelids taken before and after bonydecompression surgery with no additional lid recession procedures. In thisfashion, the effect of reduced proptosis on the contour and lateral accentuationof eyelid retraction could be measured in a consistent patient population.


Twenty-one patients (5 men and 16 women; age, 27-60 years) with Gravesdisease were photographed before any orbital or eyelid surgery. A total of38 eyes were included. Patients were placed in a modified slitlamp headrest.A posterior strap held the head in place with the chin and forehead restingfirmly against the appropriate rests. The height of the chin rest was adjustedto place both lateral canthi at the level of a previously marked point onthe headrest; this ensured proper alignment of the head both horizontallyand vertically. A digital camera (Nikon Coolpix 950 digital camera; NikonUSA, New York, NY) mounted at a fixed distance and height was used to photographpatients. Mounted light fixtures were used to obtain consistent lighting;flash was not used to guard against the blinking reflex.

Opposite eyes were occluded to ensure fixation with the photographedeye. Patients were asked to fixate on a mounted pointer adjacent to the cameralens to obtain axial photographs. Preoperative Hertel measurements were takenby one of the authors (E.L.C.) after photography.

Decompression surgery without lid surgery was performed as an electiveprocedure and performed by one surgeon (P.A.D.R.). Criteria for surgery includedpatients with clinically stable, thyroid-related ophthalmopathy, no priorhistory of any eyelid or orbital surgery, and cosmetically significant proptosis.Stable thyroid-related ophthalmopathy was defined as no detectable changesduring serial clinical examinations in the patient’s visual acuity,color-plate measurements, proptosis as measured by a Hertel exophthalmometer,strabismus, or lid retraction for at least 6 months. In 4 men and 15 women(34 eyes), bony decompression of the medial wall and posterior medial floorof the orbit was performed via a transcaruncular approach. The thick bonystrut present between the maxillary sinus and ethmoid sinus at the maxillo-ethmoidjunction was preserved to prevent hypoglobus. The medial floor of the orbitwas removed up to the medial border of the infraorbital neurovascular bundle.Two patients (4 eyes) had bony decompression performed in conjunction withthe otolaryngology service via an endoscopic endonasal approach. Only patientswith surgery performed via these 2 approaches were included because theseapproaches avoid scarring and inadvertent injury to the eyelids that mightalter their position.

Patients were asked to return at 3 months postoperatively for follow-upclinical examination. At that time, patients were photographed using the previouslydescribed protocol, and the same preoperative examiner repeated Hertel measurements.

All acquired images were transferred to a personal computer for digitalanalysis. Digital images were analyzed using NIH ImageJ 1.27 software.6 After threshold and contrast adjustment, digital photographswere calibrated by assuming each cornea to have a horizontal diameter of 12mm. The horizontal diameter of the cornea was measured using the computerto highlight the cornea on the basis of color contrast differences betweenthe conjunctiva and the cornea and calculating the numerical width of thecornea in pixels. This reference pixel width was then assigned a value of12 mm for calibration. A reference point was marked at the medial canthalangle. The magic wand tool was then used to create a selection by tracingthe junction of the eyelid margin and the conjunctiva. To trace this junction,the pointer was placed on top of the medial canthus reference point, and themagic wand tool was activated. The result is an imaginary turtle that startsmoving to the right, looking for an edge. Once the turtle finds the edge,the turtle follows the edge until it returns to the starting point (Figure 1). Using this tracing, the x and y coordinates of the peak of the uppereyelid were calculated. The horizontal distance to the peak of the upper eyelidwas measured from the previously placed reference point. In addition, a verticalline through the centroid (ie, the center point of the selection, the averageof the x and y coordinatesof all pixels in the selection) of the palpebral fissure was used to measurethe palpebral fissure height. The reproducibility of the palpebral fissurearea measurements resulted in a standard deviation of ± 0.03%using this automated tracing technique. Use of the magic wand tracing toolto perform automated tracing of the photographs at the junction of the conjunctivaand the lid margin, based on contrast differences as measured by the computer,results in minimization of investigator bias and an increase in the reproducibilityand repeatability of these measurements.

Figure 1.
Image not available

A sample of tracing and collectionof data using a digital image and ImageJ software. The white line delineatesthe measured palpebral fissure area (270.07 mm2). The palpebralfissure height was 14.23 mm in this predecompression example.


The average preoperative Hertel measurement was 25.1 mm. The averageHertel measurement 3 months after surgery was 21.6 mm. The average reductionin proptosis after bony decompression surgery was 3.5 mm (P<.001).

The average exposed surface area of the palpebral fissure before decompressionsurgery was 292.92 mm2. Three months after surgery, the averageexposed surface area of the palpebral fissure was 236.8 mm2. Theaverage reduction in exposed surface area in the palpebral fissure was 56.12mm2 (P<.001) (Figure 2).

Figure 2.
Image not available

Preoperative (A) and postoperative(B) photographs of a patient after decompression. Note the shift of the upperlid peak toward the medial canthus and the reduction in the palpebral fissureheight and the exposed ocular surface area. The star marks the upper lid peak;the line helps demonstrate the shift in the upper lid peak after surgery.The palpebral fissure area decreased from 241.58 mm2 to 180.42mm2. The palpebral fissure height decreased from 12.27 mm to 9.38mm after decompression.

The average height of the palpebral fissure was 14.21 mm before surgeryand 12.5 mm after surgery, with an average 1.71-mm reduction in the verticalheight of the palpebral fissure after decompression surgery (P = .002).

The average distance from the peak of the upper eyelid to the medialcanthus reference point was 9.1 mm. The average distance from the peak ofthe upper eyelid to the medial canthus reference point 3 months after surgerywas 7.8 mm. The net average shift of the upper eyelid peak toward the medialcanthus was 1.3 mm (P = .05).

Larger reductions in proptosis resulted in a larger shift in the peakof the upper eyelid toward the medial canthus reference point.


Upper eyelid retraction and its lateral accentuation is one of the mostcommon ophthalmic changes caused by Graves disease. The origin of this malpositionis multifactorial; it may be caused by overaction of the Müller muscleor the levator muscle, by levator fibrosis and adhesions, or by axial proptosisof the globe.2,4,5

With axial globe displacement, the lateral sclera projects further beyondthe plane of the anterior orbital rim than the medial sclera, accentuatingthe lateral flare in eyelid retraction. When these anatomical factors arecombined with a retracted upper eyelid position due to shortened or fibroticretractors, the greatest effect is on the portion of the eyelid farthest beyondthe plane of the anterior entrance to the orbit. The phenomenon of increasedlateral retraction occurs, in part, because the eye in primary position looksmedially away from the orbital axis and projects more lateral sclera.5

In 1998, van den Bosch et al7 demonstratedthat eyelid position was directly affected by globe position. Eyelid contourwas straighter in patients who underwent upper eyelid recession surgery anddecompression surgery than in the nondecompressed population.

Several confounding factors from the study by van den Bosch et al7 were addressed in our study to demonstrate that lidposition and lateral flare are a direct function of globe position. To generatean axial reduction in proptosis and avoid hypoglobus, decompression surgerywas performed on the medial wall and inferomedial floor of the orbit, withpreservation of the thick bony strut between the maxillary sinus and the ethmoidsinus at the maxillo-ethmoid junction. Eyelid measurements were taken beforeand after decompression surgery, prior to any adjunctive lid recession procedures.Precautions were taken to assure consistency in photography, and automatedimage analysis software was used to minimize human measurement errors andinvestigator bias. Even with these precautions, however, it is difficult toaddress the fact that lid height largely depends on patient effort and thatsuch effort is highly variable.

Reducing axial proptosis of the eye with decompression surgery shiftedthe peak of the upper eyelid toward the medial canthus an average of 1.3 mm.More significantly, patients demonstrated an average 1.71-mm reduction inthe height of the palpebral fissure and a 56.2-mm2 reduction inthe palpebral fissure surface area. The amount of shift in the lid peak towardthe medial canthus is a function of the reduction in proptosis after decompressionsurgery (Figure 3). Reduction in proptosisafter surgery also results in a corresponding decrease in the exposed ocularsurface area (Figure 4).

Figure 3.
Image not available

The amount of shift in the uppereyelid peak position in relation to the amount of reduction in exophthalmosafter decompression surgery.

Figure 4.
Image not available

The amount of reduction in the exposedocular surface area in relation to the amount of reduction in exophthalmosafter decompression surgery.

Therefore, performing decompression surgery prior to any adjunctivelid recession procedures may significantly reduce lid retraction, lateralflare, and ocular surface area exposure, depending on the amount of reductionin proptosis. Patients who experience large reductions in proptosis afterdecompression surgery may attain a more natural and functional eyelid andpalpebral fissure appearance without undergoing other adjunctive procedures.Consequently, decompression surgery should be considered before any otherlid surgery in the rehabilitation of any patient with Graves disease, notjust in patients with optic neuropathy, extreme proptosis, or exposure keratopathy.

Back to top
Article Information

Correspondence: Dr Chang, Doheny Eye Institute,1450 San Pablo St, DEI 4705, Los Angeles, CA 90033 (

Financial Disclosure: None.

Frueh  BRMusch  DCGarber  FW Lid retraction and levator aponeurosis defects in Graves’ eyedisease. Ophthalmic Surg 1986;17216- 220
Grove  AS  Jr Eyelid retraction treated by levator marginal myotomy. Ophthalmology 1980;871013- 1018
Small  RG Enlargement of levator palpebrae superioris muscle fibers in Graves’ophthalmopathy. Ophthalmology 1989;96424- 430
Waller  RR Eyelid malpositions in Graves’ ophthalmopathy. Trans Am Ophthalmol Soc 1982;80855- 930
Lemke  BN Anatomic considerations in upper eyelid retraction. Ophthal Plast Reconstr Surg 1991;7158- 166
Rasband  W ImageJ. 1.27 Springfield National Institutes of Health2002;
van den Bosch  WATjon-Fo-Sang  MJLemij  HG Eyeball position in Graves orbitopathy and its significance for eyelidsurgery. Ophthal Plast Reconstr Surg 1998;14328- 335