Sensitivity and Specificity of Heidelberg Retina Tomograph (HRT) Topographic Change Analysis (TCA) Parameters

Abstract

 

Purpose:

 

To determine the sensitivity and specificity of HRT TCA parameters for detecting change in glaucoma and healthy eyes.

 

Methods:

 

We determined 99%, 95% and 90% specificity cut-offs for several TCA parameters based on the variability in measurements from 1008 permuted longitudinal series derived from 18 healthy eyes imaged > 4 times each (Moorfields data). Each eye contributed 56 pseudo-random longitudinal series to the 1008 total. Specificity of each chosen cut-off was assessed in a separate group of 21 healthy eyes (from UCSD Diagnostic Innovations in Glaucoma Study, DIGS, imaged > 3 times). Sensitivity was assessed in 36 patient eyes (DIGS) with known progression by SITA SAP GPA or stereophotograph assessment, imaged > 3 times over > 2 years. We also determined the number of patient eyes (DIGS) identified as progressed by TCA analysis, but not GPA or stereophotography (n=204, imaged > 3 times over > 2 years). TCA parameters investigated were: cluster size (CSZ), area (CA) and volume (CVOL) of the largest (by area) progressed cluster both inside the optic disc margin (ins) and for the entire image (ent). All progressed clusters were repeatable based on the repeatability requirements for HRT HEYEX software.

 

Results:

 

Sensitivities/specificities (and cut-offs used) are shown below. 118 (58%) patients, stable by GPA and stereophotography, showed progression by CA^ins at 90% specificity (TCA parameter with highest sensitivity/specificity trade off). Given the high specificity of CA^ins in healthy eyes, we suspect many of these eyes may represent progression undetected by standard methods.  

 

Conclusions:

 

Sensitivity of HRT TCA parameters is fair, while specificity is good. Measurements inside the optic disc generally out-perform those over the entire image. TCA detects change in many patient eyes that appear stable by GPA and stereophotography, possibly suggesting early change detection.

 

Keywords: optic nerve • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)