Abstract
Purpose: :
Previous investigations have observed a high failure rate of the StratusOCT retinal thickness software in patients with disease (age-related macular degeneration, diabetic retinopathy, etc.). The error rate varied with pathology and ranged from 32% to as high as 92%. The failures consisted of inability to properly locate a portion of the anterior (ILM) or posterior (outer segment junction) retinal interface. Recently, the manufacturer developed a new software tool that allows manual correction of the retinal interfaces. The aim of this study was to evaluate the inter-operator reproducibility of the manual retinal interface drawing feature in optical coherence tomographic images of patients with age-related macular degeneration (ARMD).
Methods: :
We retrospectively included exams from the last 50 subjects with ARMD. Two trained OCT imaging experts and physicians (FM and NN) independently used the built-in interactive edit layer command to correct retinal thickness measurements with the StratusOCT software (version 5.0). We compared retinal volume and center retinal thickness between both imaging experts and against the standard retinal thickness software.
Results: :
We found excellent agreement between both experts. The correlation coefficient between both experts was 0.992 for center thickness and 0.999 for retinal volume. Additionally, we found that 80% of the B-scans contained detection errors (161 artifacts and 389 errors in 300 B-scans). In 16 out of 50 patients the volume error was larger than 3%. The average volume errors as measured by two experts were -0.09 ± 0.34 (range -1.53 to 0.61) and -0.07 ± 0.32 (range -1.43 to 0.61) mm3, respectively. In percent, the volume errors ranged from -14.4% to 7.8% and -13.4% to 7.8%, respectively. In our study population the average retinal volume was 6.92 ± 1.45 mm3 (range 5.11 - 12.59). The manual edited retinal volume for both experts were 7.01±1.60 (range 4.95 - 13.44) and 7.00 ± 1.59 (range 4.96-13.35) mm3, respectively. The computer-generated center thickness was 2.52 ± 133 (104-659). The manual edited center thickness measurements were 254 ± 139 (99 - 717) and 250 ± 140 (95 - 717), respectively.
Conclusions: :
The manual edit layer function is highly reproducible and offers an easy method of correcting retinal thickness volume measurements in the StratusOCT. Manual editing is recommended for longitudinal analysis of OCT measurements, in particular in patients with pathological changes.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • age-related macular degeneration • diabetic retinopathy