Zeiss CIRRUS HD-OCT 500, Руководство пользователя

Страница: 310 / 352

Appendix
2660021169012 Rev. A 2017-12 CIRRUS HD-OCT User Manual
B-2
Group 5 – Macular edema for which treatment was planned,
Group 6 – No retinal pathology.
Any subjects with a primary diagnosis that placed them within Groups 1 through 4, for
whom treatment of macular edema was scheduled, were categorized into Group 5.
Two (2) 200x200 scans and two (2) 512x128 scans of the study and fellow eyes were
acquired using the CIRRUS HD-OCT instrument during a single visit. Retinal thickness in
every subfield (based on the ETDRS 6 mm grid centered on the fovea; see
"ETDRS
Position" on page 8-7 ) was calculated using CIRRUS 3.0 software for each of the
scans. The scans were reviewed to identify scans with poor image quality due to poor
signal strength, poor scan placement within the axial field of view of the instrument
resulting in missing data, and shifts in location between scans prior to analyzing the
repeatability or reproducibility data. Scans with more than 10% missing data or data
missing from the center, very large shifts (greater than 3 mm), and very poor image quality
were excluded from the analysis as these factors would preclude accurate assessments of
repeatability between scans.
Each subject also underwent a Stratus Fast Macula scan of the study eye.
Data Analysis
Accuracy was assessed by having 14 trained clinicians perform hand segmentations of
selected B-scans from a single scan of each type from each subject. Layers segmented by
CIRRUS HD-OCT were compared to the hand–segmentations.
Agreement of the CIRRUS HD-OCT analysis with Stratus OCT was assessed by comparing
the average retinal thickness in nine retinal subfields.
Repeatability of the average measurements for each of the nine subfields was assessed
using analysis of variance. Repeatability was assessed with and without the use of an
algorithm that registers a repeated scan to a prior scan, and with and without aligning the
subfields to the subject’s fovea for each scan. Both of these capabilities were introduced
with the CIRRUS Version 4.0 software.
Results and Discussion
Accuracy
The CIRRUS HD-OCT internal limiting membrane (ILM) and retinal pigment epithelium
(RPE) segmentations were scored as accurate if software–segmentations and hand–
segmentations agreed, for 100% of A-scans that were evaluated, where agreement was
defined as being within 16
μ m for the central 1 mm of the scan and within 32 μ m
elsewhere in the scan. The accuracy of segmentations was found to depend on layer (RPE
or ILM) and disease category, and is summarized below in
Table B-1 and Table B-2 .