Sensitivity limitations in in vivo retinal imaging with ...



3D OCT Imaging in Clinical Settings: Toward Quantitative Measurements of Retinal Structures.

Robert J. Zawadzki1, Alfred R. Fuller2, Mingtao Zhao3, David F. Wiley2, Stacey Choi1, Bradley A. Bower3 Bernd Hamann2 , Joseph A. Izatt3 and John S. Werner1

1Department of Ophthalmology & Vision Science, UC Davis, Sacramento, CA

2Institute for Data Analysis and Visualization (IDAV) and Department of Computer Science, UC Davis, Davis, CA

3Department of Biomedical Engineering, Duke University, Durham, NC

In this presentation, we will describe a state-of-the-art Fourier domain OCT system using a super-luminescent diode centered at 840 nm with 50 nm optical bandwidth (6 μm axial resolution) operating at 18,000 A-scans/s (for 50 μs line exposure time) to acquire three-dimensional (3D) data from the human retina over the macula and optic nerve head. We used a fiber-based scanning con-focal Fourier-domain OCT instrument. A water vial as well as a pair of glass prisms were used in the reference arm to compensate for optics and eye water dispersion in the sample arm. A scanning head with X and Y galvo-scanners (similar to one used in commercial OCT instruments, Stratus Zeiss Meditec) was used in the sample arm. The detection channel of the OCT system consisted of 100 mm focal length fiber collimator, Wasatch Photonics Holographic diffraction grating (1200 l/mm), 150mm focal length imaging objective (air spaced triplet, Thorlabs) and 2048 pixels Camera Link line scan CCD camera from Atmel, to permit 2 mm axial scanning depth. Due to the high acquisition speed of the instrument, each volume was acquired in 5.5s, consisting of 200x500 A-scans covering a square region of the retina. B-scans were registered to remove motion artifacts and post-processed with customized 3D visualization and analysis software. Our analysis software included standard 3D visualization techniques along with a support vector machine (SVM) algorithm that allows a user to semi-automatically segment different retinal structures and layers. Figure 1 shows an example of a 3D visualization of the optical nerve head retinal region covering 6x6mm of a healthy human subject as well as the nerve fiber layer extracted from this volume.

[pic] [pic]

Figure 1. 3D visualization of the optic nerve head (left) and nerve fiber layer extracted from the 3D volume using custom SVM segmentation (right) as seen from the back of retina.

Our system makes possible precise measurements of the retinal layer thickness as well as volumetric measurements of structures of interest. Our software has been tested successfully in clinical settings for its efficacy in assessing 3D retinal structures in healthy as well as diseased cases. This may facilitate diagnosis and treatment monitoring of retinal diseases. Besides presenting results obtained with our system, we will discuss visualization and quantification methods for retinal structures, and point out potential sources of discrepancies. We also make recommendations for optimal 3D data display.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download