MRI-based SIR quantitative biomarkers: a novel imaging diagnostic strategy for thyroid eye disease activity staging.

Purpose: To evaluate the diagnostic efficacy of magnetic resonance imaging (MRI)-based signal intensity ratio (SIR) between extraocular muscles and white matter/temporal muscle for staging thyroid eye disease (TED) activity, and to provide a novel approach for diagnosis in active TED patients. Methods: A number of 40 patients with TED (79 eyes) and 65 controls (65 eyes) were recruited. MRI-based parameters of four extraocular muscles, ipsilateral white matter, temporal muscle, and other clinical factors were retrospectively collected. Patients were grouped according to disease activity determined by the Clinical Activity Score (CAS), and intergroup analysis was subsequently performed based on this classification. The signal intensities were measured using fat-suppressed T2-weighted imaging (T2WI-FS) sequences from MRI. The SIR of the extraocular rectus muscle to that of the ipsilateral white matter represents SIR1, while the SIR of the extraocular rectus muscle to that of the ipsilateral temporalis muscle represents SIR2. Results: Active TED group demonstrated elevated SIR1 and SIR2 values across all rectus muscles compared to control group and inactive TED group (P<0.05), with no differences between inactive TED and controls. Receiver operator characteristic (ROC) curve analysis identified SIR2 as superior to SIR1 for most muscles, with inferior rectus SIR1 achieving the highest AUC (0.837). Based on 95% confidence intervals and cutoff values, we propose redefining TED staging: control ranges (0.936-1.019) as absolute inactive phase, active TED ranges (1.210-1.344) as absolute active phase, and transitional values (1.019-1.210) as clinical vigilance phase requiring heightened attention. The model suggests that patients currently defined as "inactive" by CAS may have subclinical inflammation, explaining paradoxical disease progression in some cases. Conclusions: The signal intensity ratio (SIR) from fat-suppressed T2-weighted (T2WI-FS) sequences serves as a reliable predictor for TED activity. The 95% confidence interval (CI) for SIR values can provide a new strategy for early diagnosis.