advertisement

Editors Selection IGR 12-1

Structure Tests and Anatomy: Disc progression and HRT

Comment by Gerhard Zinser on:

25182 Clinical evaluation of the proper orthogonal decomposition framework for detecting glaucomatous changes in human subjects, Balasubramanian M; Bowd C; Weinreb RN et al., Investigative Ophthalmology and Visual Science, 2010; 51: 264-271

Find related abstracts

Balasubramanian et al. (186) investigated the performance of a novel method to detect optic disc progression in longitudinal HRT examinations, proper orthogonal decomposition (POD), and compared it to the performance of the topographic change analysis (TCA). In this study, POD and TCA showed similar areas under the ROC curves for separating progressing discs from discs of stable normal subjects (0.86) and from discs of non-progressing glaucoma patients (0.68). This is remarkable, as POD is designed to give a result after the first follow-up examination, while TCA requires two or more follow-up examinations to ensure adequate specificity. However, there might be a cost to the earlier result. In the interesting range at around 80% sensitivity, the specificity of POD appears to be lower (~60%) than that of TCA (~80%).

In its current implementation, POD does not provide spatial information about where in the optic disc the topographic change occurred. It would be important to know that, because much information can be gained from the geometric pattern of change. It can only be encouraged that the authors take this new method one step further and include spatially resolved information.

POD is designed to give a result after the first follow-up examination, while TCA requires two or more follow-up examinations to ensure adequate specificity

The data presented raises a very important question: Why is the diagnostic accuracy (of both, POD and TCA) to differentiate progressors from non-progressing patients so much lower than the diagnostic accuracy to differentiate progressors from normal subjects? As a possible answer, the authors suggest the problem might be in the gold standard used to identify the non-progressing glaucoma patients (SAP and expert-grading of optic disc stereophotographs). This is an important consideration when investigating the performance of new diagnostic methods, and it will certainly be beneficial to revisit this data after some additional years of follow-up.

• ← Previous
• Next →