Repeatability and Reproducibility of Anterior Chamber Angle Measurement with Swept-Source Optical Coherence Tomography in Patients with Primary Angle Closure Suspect

Purpose: To evaluate the inter- and intra-observer reliability of anterior chamber (AC) angle measurements obtained by swept-source optical coherence tomography (SS-OCT). Methods: Forty-eight consecutive patients diagnosed with primary angle closure suspect (PACS) were included. Three masked observers at different training levels (one glaucoma specialist, one ophthalmology resident, and one pre-medical college student) measured 192 SS-OCT images of the PACS patients. One observer (the glaucoma specialist) repeated measurements 1 week later. SS-OCT parameters included: Anterior segment volume, including corneal, AC, and iris volume; anterior segment dimensions, including AC depth and width (ACD, ACW), and lens vault (LV); and angle parameters, including angle opening distance (AOD), angle recess area (ARA), trabecular iris space area (TISA), and the trabecular iris angle (TIA). Intraclass correlation coefficients (ICCs) were used to measure reliability. Results: For inter-observer reproducibility, ICCs of corneal, AC, and iris volumes were 0.952 to 0.998. ICCs of ACD at all axes were above 0.989. ICCs of ACW and LV were smallest in the 90 degrees-270 degrees axis (0.751 and 0.768) but not significantly different from other axes. ARA, TISA, and TIA at all angles had significantly smallest ICCs 250 mu m from the scleral spur compared with 500 mu m and 750 mu m. The ICCs comparing observers with different training levels had similar ranges and followed similar trends. For intra-observer repeatability, the smallest ICC was 0.843. Decreasing AC depth correlated with increased inter-observer reproducibility. Conclusions: We found excellent intra-observer repeatability for all SS-OCT parameters. Angle measurements have more variation among the observers when taken 250 mu m from the scleral spur. Shallow AC might lead to more variability for angle parameters. Non-expert observers may be recruited for high-quality image grading with standard training.