Optical coherence tomography noise modeling and fundamental bounds on human retinal layer segmentation accuracy
(Conference Presentation)

Theodore B. DuBose; Peyman Milanfar; Joseph A. Izatt; Sina Farsiu

doi:10.1117/12.2214777

26 April 2016 Optical coherence tomography noise modeling and fundamental bounds on human retinal layer segmentation accuracy (Conference Presentation)

Theodore B. DuBose, Peyman Milanfar, Joseph A. Izatt, Sina Farsiu

Author Affiliations +

Proceedings Volume 9697, Optical Coherence Tomography and Coherence Domain Optical Methods in Biomedicine XX; 96971U (2016) https://doi.org/10.1117/12.2214777
Event: SPIE BiOS, 2016, San Francisco, California, United States

Abstract

The human retina is composed of several layers, visible by in vivo optical coherence tomography (OCT) imaging. To enhance diagnostics of retinal diseases, several algorithms have been developed to automatically segment one or more of the boundaries of these layers. OCT images are corrupted by noise, which is frequently the result of the detector noise and speckle, a type of coherent noise resulting from the presence of several scatterers in each voxel. However, it is unknown what the empirical distribution of noise in each layer of the retina is, and how the magnitude and distribution of the noise affects the lower bounds of segmentation accuracy. Five healthy volunteers were imaged using a spectral domain OCT probe from Bioptigen, Inc, centered at 850nm with 4.6µm full width at half maximum axial resolution. Each volume was segmented by expert manual graders into nine layers. The histograms of intensities in each layer were then fit to seven possible noise distributions from the literature on speckle and image processing. Using these empirical noise distributions and empirical estimates of the intensity of each layer, the Cramer-Rao lower bound (CRLB), a measure of the variance of an estimator, was calculated for each boundary layer. Additionally, the optimum bias of a segmentation algorithm was calculated, and a corresponding biased CRLB was calculated, which represents the improved performance an algorithm can achieve by using prior knowledge, such as the smoothness and continuity of layer boundaries. Our general mathematical model can be easily adapted for virtually any OCT modality.

Conference Presentation

Citation Download Citation

Theodore B. DuBose, Peyman Milanfar, Joseph A. Izatt, and Sina Farsiu "Optical coherence tomography noise modeling and fundamental bounds on human retinal layer segmentation accuracy (Conference Presentation)", Proc. SPIE 9697, Optical Coherence Tomography and Coherence Domain Optical Methods in Biomedicine XX, 96971U (26 April 2016); https://doi.org/10.1117/12.2214777

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
PRESENTATION

WATCH
PRESENTATION SAVE TO MY LIBRARY

GET CITATION

KEYWORDS

Optical coherence tomography

Image segmentation

Mathematical modeling

Algorithm development

Coherence imaging

Retina

Speckle

Show All Keywords

Keywords/Phrases

Search In:

Publication Years