Normalization of Time Domain Optical Coherence Tomography (TD-OCT) Retinal Nerve Fiber Layer (RNFL) Thickness Measurements at Variable Scan Locations to a Virtual Universal Center Location Using Three-Dimensional (3D) Spectral Domain (SD-) OCT Data

Abstract

 

Purpose:

 

To develop and test a method of normalizing TD-OCT RNFL thickness measurements obtained at variable scan locations to a virtual universal center location using a retinal nerve fiber bundle distribution (RNFBD) pattern detected on the corresponding 3D SD-OCT images.

 

Methods:

 

Twelve eyes of 12 healthy subjects and 7 eyes of 7 glaucoma subjects were enrolled. A set of nine TD-OCT (Stratus OCT; Carl Zeiss Meditec, Inc., Dublin, CA (CZMI)) circumpapillary scans (one centered and 8 intentionally off-centered) were obtained for each eye. One 3D SD-OCT (Cirrus HD-OCT; CZMI) optic nerve head cube scan was also obtained at the same visit. RNFBD pattern was modeled by detecting the major RNFBD curvatures (one for each in superior and inferior hemi-field) on a SD-OCT cube data for each eye. RNFL thickness measurements from off-centered TD-OCT scans were normalized by using the modeled RNFBD pattern and the matched scan location within the corresponding SD-OCT cube data. Algorithm performance was assessed by comparing global and sectoral RNFL thickness measurement imprecisions with and without normalization.

 

Results:

 

RNFL thickness measurement imprecision was statistically significantly lower with normalization than without in all sectors except for global mean (Figure).

 

Conclusions:

 

The developed normalization method reduced the RNFL thickness measurement variability caused by variable scan locations. This method may be useful for longitudinal glaucoma progression analysis.  

 

Keywords: imaging/image analysis: clinical • image processing • computational modeling