Reliability of Extraocular Muscle Measurements on Magnetic Resonance Imaging in Graves' Disease: A two-rater case study

Thyroid eye disease (TED) is based mainly on clinical and ophthalmological findings, but imaging exams such as magnetic resonance imaging (MRI) is helpful on its identification and follow-up when the disease is clinically active. The main characteristic of TED is the enlargement of extraocular muscles that can be easily assessed with good quality of imaging but it is still challenging to evaluate it in the scenario of symmetric mild disease, where the patient-specific normal measure parameter is lost. Considering these challenges, the objective of this study was to evaluate two different raters with similar clinical experiences on measuring MR images to obtain metrics of extraocular musculature and investigate the method reliability isolated from the clinical perspective. A total of 31 MR images were collected between 2019 and 2021. The images are from 3 groups: Patients with Graves Disease (GD) and active inflammatory ophthalmopathy, patients with GD and inactive ophthalmopathy and healthy control group without evidence of autoimmune thyroid disease. The radiologists measured the inferior rectus, superior rectus, medial rectus, lateral rectus, superior oblique and inferior oblique muscle thickness, as well the proptosis divided by eye. They also identified the increase of extraconal fat, intraconal fat and changes in lacrimal gland. Both raters were blinded to each other measurements and to the patient group. On an overview, we could identify differences in the results for each rater. One rater had more reliable results evaluating eye muscles, while the other had better results assessing other eye structures, which can be due to the experience with the disease and with the imaging analysis. We concluded that the method is not reliable as a standalone evaluation while there is no gold standard on how to perform it, and must continue to be used as supplementary information to corroborate clinical and laboratorial findings.