Megahertz ultra-wide-field swept-source retina optical coherence tomography compared to current existing imaging devices

Background To investigate the image quality of wide-angle cross-sectional and reconstructed fundus images based on ultra-megahertz swept-source Fourier domain mode locking (FDML) OCT compared to current generation diagnostic devices. Methods A 1,050 nm swept-source FDML OCT system was constructed running at 1.68 MHz A-scan rate covering approximately 70 degrees field of view. Twelve normal eyes were imaged with the device applying an isotropically dense sampling protocol (1,900x1,900 A-scans) with a fill factor of 100 %. Obtained OCT scan image quality was compared with two commercial OCT systems (Heidelberg Spectralis and Stratus OCT) of the same 12 eyes. Reconstructed en-face fundus images from the same FDML-OCT data set were compared to color fundus, infrared and ultra-wide-field scanning laser images (SLO). Results Comparison of cross-sectional scans showed a high overall image quality of the 15x averaged FDML images at 1.68 MHz [overall quality grading score: 8.42 +/- 0.52, range 0 (bad)-10 (excellent)] comparable to current spectral-domain OCTs (overall quality grading score: 8.83 +/- 0.39, p=0.731). On FDML OCT, a dense 3D data set was obtained covering also the central and mid-peripheral retina. The reconstructed FDML OCT en-face fundus images had high image quality comparable to scanning laser ophthalmoscope (SLO) as judged from retinal structures such as vessels and optic disc. Overall grading score was 8.36 +/- 0.51 for FDML OCT vs 8.27 +/- 0.65 for SLO (p=0.717). Conclusions Ultra-wide-field megahertz 3D FDML OCT at 1.68 MHz is feasible, and provides cross-sectional image quality comparable to current spectral-domain OCT devices. In addition, reconstructed en-face visualization of fundus images result in a wide-field view with high image quality as compared to currently available fundus imaging devices. The improvement of >30x in imaging speed over commercial spectral-domain OCT technology enables high-density scan protocols leading to a data set for high quality cross-sectional and en-face images of the posterior segment.