Purpose: To compare automated retinal thickness values generated by the fast macular thickness maps (FMTM) and customized 6-radial line scans (RLS) versus manual retinal measurements on Stratus optical coherence tomography (OCT) (Carl Zeiss Meditec, Dublin, CA).
Design: Prospective, observational case series.
Participants: Patients with subfoveal choroidal neovascularization (CNV) caused by age-related macular degeneration (AMD), diabetic macular edema (DME), or branch/central retinal vein occlusion (RVO).
Methods: Patients were prospectively imaged using the FMTM and customized RLS patterns on Stratus OCT at the same sitting. Each scan was evaluated for errors in retinal segmentation (i.e., correct retinal boundaries [CRB]). Automated values were recorded while central retinal thickness measurements were determined manually for both patterns. The presence or absence of epiretinal phenomenon, cystoid spaces, pigment epithelial detachment, and subretinal fluid was also noted.
Main outcome measures: Errors in retinal segmentation at and outside the fovea (i.e., CRB) and percentage of automated values within a clinically acceptable margin (+/-25 mum) of the manual central retinal thickness.
Results: A total of 147 eyes of 147 patients (95 eyes with exudative AMD, 41 eyes with DME, and 11 eyes with macular edema caused by RVO) were included. For wet AMD, the total number of CRB at the fovea and outside the fovea was 363 (63.7%) and 360 (63.2%), respectively, in FMTM and 428 (75.1%) and 426 (74.7%), respectively, in RLS (P<0.0001 for both). For DME and RVO, the total number of CRB at the fovea and outside the fovea was 274 (87.8%) and 256 (82.1%), respectively, in FMTM and 287 (92.0%) and 270 (86.5%), respectively, in RLS (P = 0.11, P = 0.15, respectively). Some 40% and 56% of automated foveal center point thicknesses on FMTM and RLS, respectively, were within +/-25 mum of the manual central retinal thickness for AMD (P = 0.042), versus 94% and 81% for DME and RVO, respectively (P = 0.07).
Conclusions: For exudative AMD, the RLS protocol provides fewer segmentation errors than the FMTM protocol, and its automated retinal thickness values (e.g., foveal center point, central subfield) correlate better with manual retinal thickness measurement than FMTM. In DME and RVO, however, both protocols provide similar and low segmentation errors, and their automated results are close to manual measurements.
Financial disclosure(s): Proprietary or commercial disclosure may be found after the references.