Eyeball descending identification using MRI-based spatial coordinates in thyroid-associated orbitopathy patients with unilateral upper eyelid retraction

Introduction

Upper eyelid retraction (UER) is the most common clinical manifestation of thyroid-associated orbitopathy (TAO), and occurs in approximately 80% of TAO patients (1,2). In some cases, UER may be the sole manifestation of the disease (3). UER is characterized by an abnormally high resting position of the upper eyelid, which can lead to conjunctival hyperaemia, and carries a high risk of conjunctival infections (2,4). Further, patients with UER may display surprised or afraid expressions (5).

In rare cases, UER occurs unilaterally (6). The changed appearance of the eyelids can influence the perceived position of the eyeballs. According to clinical findings and previous research, the affected eyeball appears to be positioned lower than the healthy eyeball (Figure 1) (7-10). However, there is currently a lack of analysis addressing whether the impaired eyeball is positioned objectively lower in patients with UER (11).

Figure 1 Images demonstrating the appearance of eyes without upper eyelid retraction and with unilateral upper eyelid retraction.

Orbital magnetic resonance imaging (MRI) can provide more objective and comprehensive information on the internal structure of orbits than digital photographs (12,13). With its high soft tissue resolution, MRI is widely used to evaluate TAO in clinical practice (12,14). However, the inevitable positional skew that occurs during scan positioning can significantly affect the judgment of binocular position differences on conventional magnetic resonance (MR) images (11,15,16). With the proper application of image registration, three-dimensional (3D) reconstruction, and medical image visualization tools (17-19), it is possible to spatially position the centroid and bounding box of the segmented eyeball, allowing for an objective and accurate evaluation of the eyeball position.

Thus, this study aimed to assess the phenomenon of eyeball descent in TAO patients with unilateral UER and to investigate the associated factors via the 3D reconstruction of MR images. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1659/rc).

Methods

Patients

The retrospective cross-sectional study was conducted in accordance with the Health Insurance Portability and Accountability Act (HIPAA) and the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Review Committee of Peking University People’s Hospital (No. 2022PHB123-001), and the requirement of individual consent for this retrospective analysis was waived.

From December 2019 to September 2022, patients were considered eligible for inclusion in the study if they met the following inclusion criteria: (I) had a diagnosis of TAO based on Bartley’s diagnostic criteria (20); (II) presented clinically with unilateral UER (≥2 mm) (21); and (III) had undergone a standardized orbital MRI examination at Peking University People’s Hospital. Patients were excluded from the study if they met any of the following exclusion criteria: (I) the MR image quality was inadequate for further analysis; (II) the proptosis difference between the eyeballs was >2 mm (22); and/or (III) they had previously undergone orbital surgery. (This study sought to minimize the effect of exophthalmos and orbital surgery on the assessment of UER and eyeball position). Healthy control participants were considered eligible for inclusion in the study if they met the following inclusion criteria: (I) had undergone a standardized orbital MRI examination that demonstrated no obvious abnormality; and (II) had no history of orbital trauma, surgery, orbital disease, or neuro-ophthalmic conditions affecting the orbital structure.

Clinical data

The clinical data of the study participants were collected by an investigator who was blinded to the MRI results. Age, sex, symptom duration, and the clinical activity score of TAO were recorded for the patients in the TAO group. Symptom duration was defined as the time between the appearance of the first symptoms and the MRI scan.

The patient was instructed to look straight ahead, and the small convexities on the Hertel exophthalmometer (66 Vision Tech Co., Ltd., Suzhou, China) were positioned on the temporal orbital rims of both eyes. When the two scales on the exophthalmometer aligned in the plane mirror, the scale value corresponding to the corneal tip image was read, recording the degree of proptosis. Moreover, margin reflex distance 1 was also assessed. Information on smoking and a history of TAO (including duration and treatment) was obtained through both investigator-initiated telephone follow-ups and patients’ medical records.

MRI acquisition and image processing

Standardized orbital MRI examinations were performed using a 3.0T system with a 16-channel head coil (Ingenia, Philips Healthcare, the Netherlands). Participants were instructed to lie in a supine position, and look forward, with their eyes closed naturally. A rice bag (400 g) was placed on their eyes to keep them still. Coronal and oblique T2-weighted sequences with fat suppression were used (Table 1), along with conventional structural MRI sequences (axial T1-weighted imaging and axial T2-weighted imaging with and without fat suppression). The axial scanning line was parallel to the auditory canthus, and the oblique sagittal line was parallel to the optic nerve.

Measurement of descent distance

The MRI data were anonymized and provided in Digital Imaging and Communication in Medicine (DICOM) format. The data were then imported into the 3D slicer tool (https://www.slicer.org/, National Institutes of Health, version 5.0.3). To reduce the influence of head movement, coronal T2-weighted spectral pre-saturation with inversion recovery (SPIR) orbital images from each participant were registered using advanced normalization tools (ANTS) based on images from a single participant randomly selected from the control group. The segmented volumes were transformed to the registered positions. The 3D slicer tool was then used to manually outline the eyeballs layer by layer. The segmented data of bilateral eyeballs was extracted as connected components (Figure 2A). The segmented eyeballs data were extracted using spatial coordinates of the registered head space, preserving the distances of the segmented eyeballs to the registered MRI scan boundary. The registered segments were then imported into MATLAB (MathWorks, Inc. version R2020a), and the spatial coordinates of the vertical (i.e., along the axial direction) highest points (A₁, B₁) and the lowest points (A₂, B₂) of the eyeballs were extracted (Figure 2B). The line segments between the highest and lowest points of eyeballs to the top boundary of the registered volume were denoted as AA₁, AA₂, BB₁, and BB₂, respectively (Figure 2B). The descent distance (ΔH) was calculated in MATLAB as the vertical difference between the centroid of the eyeballs in a single patient or control participant, measured along the direction perpendicular to the plane as expressed in Eq. [1]:

ΔH=12(BB2⇀-BB1⇀)−12(AA2⇀-AA1⇀)[1]

Figure 2 Illustrations of the methods used for the eyeball measurements. (A) The process of the eyeball segmentation was employed to facilitate subsequent importation into MATLAB; (B) schematic diagram showing the equation for measuring eyeball descent distance; (C) illustration of the measurements of the LPS, SR, and IR muscle thickness; (D) segmentation of the LPS-SR complex on the oblique sagittal plane; the corresponding volume value was automatically calculated by the software. LPS, levator palpebrae superioris; SR, superior rectus; IR, inferior rectus.

Other imaging features

The thickness of the levator palpebrae superioris (LPS), superior rectus (SR), and inferior rectus (IR) muscles was measured on the oblique sagittal T2-weighted SPIR orbital images using the 3D slicer tool (Figure 2C), and the difference in the thickness between the impaired side and the healthy side was measured. The volume of the LPS-SR complex was automatically calculated after manual segmentation using the 3D slicer tool (Figure 2D).

Other imaging features were also recorded, including the protrusion distances from the orbital rims, the thickness of other extraocular muscles, and optic nerve involvement. Inviwo (https://inviwo.org/) is an open-source software framework for rapid visualization prototyping (18). Using the visualization modules of Inviwo, coronal orbital MR images were used for the 3D visualization reconstruction of images to better observe the relative positions between the eyeballs (Figure 3).

Figure 3 Magnetic resonance image and three-dimensional visualization reconstruction image based on a TAO patient with unilateral upper eyelid retraction. (A) Coronal T2-weighted image from a TAO patient with right upper eyelid retraction; (B) the reconstruction of this image using inviwo software showed the changes in eyeball position more clearly. TAO, thyroid-associated orbitopathy.

Statistical analysis

The categorical variables are reported as the number and percentage. The quantitative variables with a non-normal distribution are reported as the mean ± standard deviation, or the median (interquartile range). The student’s t-test was used to compare the descent distance differences for the eyeballs in the patient group and those in the healthy control group. Spearman correlation coefficients were computed to assess the strength and direction of the associations. A linear regression analysis was performed to examine the relationship between variables, with shaded areas indicating 95% confidence intervals for the fit. All statistical analyses were performed using RStudio software (RStudio, PBC, version 2022.07.2+576). Scatter plots and linear regressions were generated using the R package ggplot2 (https://ggplot2.tidyverse.org/, version 3.4.0). A priori power analysis using G*Power software (Franz Faul, university Kiel, Germany, version 3.1.9.4) indicated that a total sample size of 54 (27 per group) was required to detect a large effect size (d=1.19) with a power of 0.95 and an alpha level of 0.01 for a two-tailed independent t-test.

Results

Demographics

A total of 47 TAO patients with unilateral UER, and 85 healthy control participants were initially considered for inclusion in the study (Figure 4). Once the exclusion criteria were applied, 30 TAO patients with unilateral UER and 67 healthy control participants were included in the final sample. The mean age of the patients was 34.27±10.51 years (range, 20–71 years), while that of the healthy control participants was 46.87±16.25 years (range, 21–74 years). The TAO group comprised 4 male and 26 female patients, while the healthy control group comprised 35 male and 32 female participants.

Figure 4 Flow diagram of study. TAO, thyroid-associated orbitopathy; LPS, levator palpebrae superioris; IR, inferior rectus; SR, superior rectus; UER, upper eyelid retraction; MRI, magnetic resonance imaging; 3D, three dimensional.

In the TAO group, 5 of the 30 patients were in the active phase (with clinical activity scores ≥3) (Figure 5). In terms of severity, 8 patients had mild TAO, while most had moderate-to-severe TAO. Additionally, four patients were smokers. None of the patients had hypotropia or a history of ophthalmic surgery in this study. Proptosis was measured using the Hertel exophthalmometer, and the average margin reflex distance 1 of the affected eye was 18.5±2.1 and 4.67±0.93 mm. The average contralateral proptosis was 17.8±2.1 mm, and there was no significant difference in exophthalmos between the affected eyeball and the contralateral eyeball (P=0.204). The median symptom duration among patients was 2 months (range: 3 days to 25 months), with 25 patients (83.3%) demonstrating a short symptom duration (<18 months), and the remaining 5 patients (16.7%) demonstrating a long symptom duration (≥18 months).

Figure 5 Heatmap and bar charts of the clinical characteristics of TAO patients with unilateral UER. CAS, clinical activity score; M, month; TAO, thyroid-associated orbitopathy; UER, upper eyelid retraction.

Eyeball descent distance

In the TAO patients with unilateral UER, the average descent distance of the impaired eyeball compared with the healthy eyeball was 1.192±1.159 mm (P<0.001). In the healthy control participants, the height difference of the eyeballs was −0.067±0.938 mm (P=0.560). Eyeball descent did not occur in the affected eyeballs of 5 patients (−0.400±0.418 mm), 4 of whom had a symptom duration of <3 months.

MR features of LPS, SR, and IR muscles

A total of 28 patients (93.3%) presented with LPS muscle hypertropia on the impaired side (with UER) (4.964±1.607) versus on the healthy side (without UER) (3.741±1.319 mm; P=0.002). Additionally, 24 patients had a thicker SR muscle on the impaired side (4.373±1.610 mm) than the healthy side (3.467±0.872 mm; P=0.010). There was no difference in the IR muscle thickness between the impaired (4.018±0.812 mm) and healthy eyeballs (4.010±0.633 mm) (P=0.968). The LPS-SR complex volume was higher on the impaired side (1.808±0.764 cm³) than the healthy side (1.228±0.403 cm³; P<0.001) among all patients.

Factors associated with eyeball descent distance

There was a significant positive correlation between the descent distance in the impaired eyes and the increased LPS muscle thickness (Spearman’s rho =0.803; P<0.001). However, the descent distance was not correlated with an increased SR muscle thickness (Spearman’s rho =0.012; P=0.950), an increased IR muscle thickness (Spearman’s rho =0.035; P=0.853), or an increased LPS-SR complex volume (Spearman’s rho =0.025; P=0.894).

The simple linear regression model confirmed that an increased LPS muscle thickness was correlated with the descent difference (P<0.001); however, no significant correlations were found between the other variables and descent difference (all P>0.01) (Figure 6).

Figure 6 Scatter plots with fitted linear models showing the correlation between the descent distance and other parameters in thyroid-associated orbitopathy patients with unilateral upper eyelid retraction. The shaded areas indicate 95% confidence intervals. IR, inferior rectus; LPS, levator palpebrae superioris; SR, superior rectus.

Discussion

In this study, 83.3% of the TAO patients with unilateral UER had impaired eyeball descent, and the descent distance was significantly correlated with an increased thickness of the LPS muscle. These findings showed that impaired eyeball descent is an objective phenomenon, and the distance of this descent can be measured.

To achieve anatomically accurate and objective estimations of distances in this study, we extracted spatial coordinates from the bounding box, which were extracted from connected components based on the 3D segmentation of T2-weighted images of both eyeballs after registration (23). In addition, 3D visualization allowed us to clearly display the height difference between the eyeballs. The results confirmed previous clinical observations that the impaired eyeball is objectively lower than the non-affected eyeball, and this appearance is not simply a visual deviation caused by an increase in the marginal reflex distance (22,24). In previous studies, an upward rotation of the impaired eyeball has also been observed in patients with UER (22,24). This may be caused by levator muscle contracture due to inflammation and fibrosis (25). According to some research, eyeball rotation rarely affects the vertical distance of the centroid of the eyeball during routine examinations (19,26). Thus, we compared the centroids of eyeballs rather than the marginal reflex distance to eliminate any interference that might be caused by the rotation of the eyeballs.

We also investigated the factors that may be associated with eyeball descent in this patient population. We found that the LPS-SR complex volume was higher on the impaired side than the healthy side in all the patients. This finding is consistent with the results of previous studies that an increased LPS-SR complex volume is associated with UER in TAO patients (21,22). As the structure and positions of the LPS and SR muscles differ (27), we also measured the thicknesses of these muscles separately in this study. Previous research has shown that the muscular branch to the LPS muscle passes through the SR muscle belly to innervate the LPS muscle in 12.5% of cases, and nervous branches to the LPS muscle may occasionally pierce the SR muscle (28-30). The muscle belly between the LPS muscle and SR muscle can be distinguished; however, it can be difficult to accurately segment the LPS muscle and SR muscle at the marginal level. In this study, we therefore calculated the volume of the LPS-SR complex rather than the volumes of LPS muscle and SR muscle. Moreover, we found that 93.3% of the patients presented with LPS muscle hypertropia on the impaired side, and the mean increase in thickness of the LPS muscle was greater than that of the SR muscle, indicating that patients with thicker LPS muscles may experience a greater degree of impaired eyeball descent. We also found that the descent distance of the impaired eyeball was significantly correlated with an increased thickness of the LPS muscle but not with an increased thickness of the SR or IR muscles, or an increased volume of the LPS-SR complex. The thickening of the LPS muscle and SR muscle is not always concurrent; some patients show pronounced LPS muscle thickening with minimal SR muscle change, and vice versa. These results suggest that the thickening of the LPS muscle, which is positioned near the top of the eyeball, may directly compress the eyeball, causing it to descend (31). The immunoreactive inflammation of the LPS muscle was one of the first theories proposed to explain the occurrence of UER in patients with TAO (25,31,32). Some research has shown that the LPS muscle is frequently thickened in TAO patients with UER (7,25,33,34). Among the 30 TAO patients with unilateral UER in this study, 5 did not demonstrate eyeball descent. These patients also did not demonstrate any increased thickness of the LPS muscle. The occurrence of unilateral UER among these patients could be related to Muller’s muscle, which is another factor involved in UER (21).

The limitations of this study include its observational single-center design and small sample size, which was due to the low incidence of TAO with unilateral UER. The ratio of male patients to female patients in this study was 4:26; however, this reflects clinical findings, as TAO with UER occurs more frequently in women than men (35). In addition, this study did not specifically investigate the effect of eyeball rotation and other extraocular muscles on the height of eyeballs.

Conclusions

Our findings suggest that the descent distance of the eyeball in TAO patients with unilateral UER is associated with pathological changes in the LPS muscle. When contemplating orbital decompression surgery for such patients, consideration should be given to the possible cosmetic implications, such as the increased downward displacement of the affected eye.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-24-1659/rc

Funding: This work was supported by Peking University People’s Hospital Scientific Research Development Fund (No. RDL2022-18).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-24-1659/coif). J.L. is an employee of Philips Healthcare. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The retrospective cross-sectional study was conducted in accordance with the Health Insurance Portability and Accountability Act (HIPAA) and the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Review Committee of Peking University People’s Hospital (No. 2022PHB123-001), and the requirement of individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bahn RS. Graves’ ophthalmopathy. N Engl J Med 2010;362:726-38. [Crossref] [PubMed]
2. Osaki TH, Monteiro LG, Osaki MH. Management of eyelid retraction related to thyroid eye disease. Taiwan J Ophthalmol 2022;12:12-21. [Crossref] [PubMed]
3. Chan W, Wong GW, Fan DS, Cheng AC, Lam DS, Ng JS. Ophthalmopathy in childhood Graves' disease. Br J Ophthalmol 2002;86:740-2. [Crossref] [PubMed]
4. Rana HS, Akella SS, Clabeaux CE, Skurski ZP, Aakalu VK. Ocular surface disease in thyroid eye disease: A narrative review. Ocul Surf 2022;24:67-73. [Crossref] [PubMed]
5. Guastella C, di Furia D, Torretta S, Ibba TM, Pignataro L, Accorona R. Upper Eyelid Retraction in Graves' Ophthalmopathy: Our Surgical Experience on 153 Cases of Full-Thickness Anterior Blepharotomy with Mullerectomy. Aesthetic Plast Surg 2022;46:1713-21. [Crossref] [PubMed]
6. Nugent RA, Belkin RI, Neigel JM, Rootman J, Robertson WD, Spinelli J, Graeb DA. Graves orbitopathy: correlation of CT and clinical findings. Radiology 1990;177:675-82. [Crossref] [PubMed]
7. Duan M, Xu DD, Zhou HL, Fang HY, Meng W, Wang YN, Jin ZY, Chen Y, Zhang ZH. Triamcinolone acetonide injection in the treatment of upper eyelid retraction in Graves' ophthalmopathy evaluated by 3.0 Tesla magnetic resonance imaging. Indian J Ophthalmol 2022;70:1736-41. [Crossref] [PubMed]
8. Pinas D, O B, De Keizer R, Wubbels RJ, van den Bosch WA, Paridaens D. Results of surgical correction of upper eyelid retraction in Graves' Orbitopathy. Acta Ophthalmol 2021;99:e608-13. [Crossref] [PubMed]
9. Xu D, Liu Y, Xu H, Li H. Repeated triamcinolone acetonide injection in the treatment of upper-lid retraction in patients with thyroid-associated ophthalmopathy. Can J Ophthalmol 2012;47:34-41. [Crossref] [PubMed]
10. Joos ZP, Sullivan TJ. Peri-levator palpebrae superioris triamcinolone injection for the treatment of thyroid eye disease-associated upper eyelid retraction. Clin Exp Ophthalmol 2017;45:651-2. [Crossref] [PubMed]
11. van den Bosch WA, Tjon-Fo-Sang MJ, Lemij HG. Eyeball position in Graves orbitopathy and its significance for eyelid surgery. Ophthalmic Plast Reconstr Surg 1998;14:328-35. [Crossref] [PubMed]
12. Wnuk E, Maj E, Jabłońska-Pawlak A, Jeczeń M, Rowińska-Berman K, Rowiński O. Validation of exophthalmos magnetic resonance imaging measurements in patients with Graves' orbitopathy, compared to ophthalmometry results. Pol J Radiol 2022;87:e539-44. [Crossref] [PubMed]
13. Georgouli T, Chang B, Nelson M, James T, Tanner S, Shelley D, Saldana M, McGonagle D. Use of high-resolution microscopy coil MRI for depicting orbital anatomy. Orbit 2008;27:107-14. [Crossref] [PubMed]
14. Tian H, Wang Y, Li J, Li H. Prognostic value of extraocular muscles for intraocular pressure in thyroid-associated ophthalmopathy patients. Quant Imaging Med Surg 2023;13:5727-36. [Crossref] [PubMed]
15. Pereira TS, Leite CA, Kuniyoshi CH, Gebrim EMMS, Monteiro MLR, Pieroni Gonçalves AC. A randomized comparative study of inferomedial vs. balanced orbital decompression. Analysis of changes in orbital volume, eyelid parameters, and eyeball position. Eye (Lond) 2022;36:547-54. [Crossref] [PubMed]
16. Park HH, Chun YS, Moon NJ, Kim JT, Park SJ, Lee JK. Change in eyelid parameters after orbital decompression in thyroid-associated orbitopathy. Eye (Lond) 2018;32:1036-41. [Crossref] [PubMed]
17. Fedorov A, Beichel R, Kalpathy-Cramer J, Finet J, Fillion-Robin JC, Pujol S, Bauer C, Jennings D, Fennessy F, Sonka M, Buatti J, Aylward S, Miller JV, Pieper S, Kikinis R. 3D Slicer as an image computing platform for the Quantitative Imaging Network. Magn Reson Imaging 2012;30:1323-41. [Crossref] [PubMed]
18. Jonsson D, Steneteg P, Sunden E, Englund R, Kottravel S, Falk M, Ynnerman A, Hotz I, Ropinski T. Inviwo - A Visualization System with Usage Abstraction Levels. IEEE Trans Vis Comput Graph 2020;26:3241-54. [Crossref] [PubMed]
19. Guo J, Qian J, Yuan Y, Zhang R, Huang W. A Novel Three-Dimensional Vector Analysis of Axial Globe Position in Thyroid Eye Disease. J Ophthalmol 2017;2017:7253898. [Crossref] [PubMed]
20. Bartley GB, Gorman CA. Diagnostic criteria for Graves' ophthalmopathy. Am J Ophthalmol 1995;119:792-5. [Crossref] [PubMed]
21. Davies MJ, Dolman PJ. Levator Muscle Enlargement in Thyroid Eye Disease-Related Upper Eyelid Retraction. Ophthalmic Plast Reconstr Surg 2017;33:35-9. [Crossref] [PubMed]
22. Byun JS, Lee JK. Relationships between eyelid position and levator-superior rectus complex and inferior rectus muscle in patients with Graves' orbitopathy with unilateral upper eyelid retraction. Graefes Arch Clin Exp Ophthalmol 2018;256:2001-8. [Crossref] [PubMed]
23. Szymor P, Kozakiewicz M, Olszewski R. Accuracy of open-source software segmentation and paper-based printed three-dimensional models. J Craniomaxillofac Surg 2016;44:202-9. [Crossref] [PubMed]
24. Lee JY, Bae K, Park KA, Lyu IJ, Oh SY. Correlation between Extraocular Muscle Size Measured by Computed Tomography and the Vertical Angle of Deviation in Thyroid Eye Disease. PLoS One 2016;11:e0148167. [Crossref] [PubMed]
25. Cruz AA, Ribeiro SF, Garcia DM, Akaishi PM, Pinto CT. Graves upper eyelid retraction. Surv Ophthalmol 2013;58:63-76. [Crossref] [PubMed]
26. Moon Y, Lee WJ, Shin SH, Kim JH, Lee JY, Oh SY, Lim HW. Positional Change of the Eyeball During Eye Movements: Evidence of Translatory Movement. Front Neurol 2020;11:556441. [Crossref] [PubMed]
27. Knight B, Fakoya AO, Lopez MJ, Patel BC. Anatomy, Head and Neck: Levator Palpebrae Superioris Muscle. In: Treasure Island (FL): StatPearls Publishing; 2024.
28. Haładaj R. Normal Anatomy and Anomalies of the Rectus Extraocular Muscles in Human: A Review of the Recent Data and Findings. Biomed Res Int 2019;2019:8909162. [Crossref] [PubMed]
29. Djordjević B, Novaković M, Milisavljević M, Milićević S, Maliković A. Surgical anatomy and histology of the levator palpebrae superioris muscle for blepharoptosis correction. Vojnosanit Pregl 2013;70:1124-31. [Crossref] [PubMed]
30. Flammer J, Mozaffarieh M, Bebie H. Basic Sciences in Ophthalmology: Physics and Chemistry. Berlin, Heidelberg: Springer Berlin Heidelberg, 2013.
31. Ohnishi T, Noguchi S, Murakami N, Nakahara H, Hoshi H, Jinnouchi S, Futami S, Nagamachi S, Watanabe K. Levator palpebrae superioris muscle: MR evaluation of enlargement as a cause of upper eyelid retraction in Graves disease. Radiology 1993;188:115-8. [Crossref] [PubMed]
32. Al-Qadi M, Hussain A. Influence of orbital decompression on upper eyelid retraction in Graves' orbitopathy: a systematic review and meta-analysis. Orbit 2024;43:549-54. [Crossref] [PubMed]
33. Ai LK, Hu YB, Wu Y, Man FY. The MRI manifestations of the levator muscles in thyroid associated ophthalmopathy. Ophthalmology 2014;23:326-31.
34. Parsons SR, Wilson-Pogmore A, Sullivan TJ. Percutaneous triamcinolone injection for upper eyelid retraction in thyroid eye disease. Front Ophthalmol (Lausanne) 2024;4:1388197. [Crossref] [PubMed]
35. Lazarus JH. Epidemiology of Graves' orbitopathy (GO) and relationship with thyroid disease. Best Pract Res Clin Endocrinol Metab 2012;26:273-9. [Crossref] [PubMed]