advertisement
Editors Selection IGR 24-1
Progression: Computer-assisted detection of visual field changes
Comment by Kouros Nouri-Mahdavi on:
50957 Progression of patterns (POP): a machine classifier algorithm to identify glaucoma progression in visual fields, Goldbaum MH; Lee I; Jang G et al., Investigative Ophthalmology and Visual Science, 2012; 53: 6557-6567
Goldbaum and co-investigators recently explored the utility of machine learning classifiers (MLC) for detection of visual field progression in glaucoma. Previously, the same team of investigators demonstrated that glaucoma patients could be reliably separated from normal subjects using a type of MLC called variational Bayesian independent component mixture model or VIM and that patterns of visual field loss could be categorized in seven different axes. The investigators have now expanded their work to explore the utility of this Progression of Patterns (PoP) analysis for detection of glaucoma worsening. They defined confidence limits for variability of visual field trends along the seven axes based on a group of stable glaucoma patients tested five times over a short period of time. They further evaluated sensitivity trends in the above-mentioned seven axes for each visual field series in patients with definite or suspected glaucoma. The axis showing the most prominent change was selected and linear regression analysis was performed in that axis (using VIM units). The methods introduced are of significant potential interest because of two main reasons. Firstly, limits of variability for trends over time were estimated not only based on the average variability in the subset of stable glaucoma patients, but also by taking into account the patient's own variability over time. Secondly, although the authors' approach provides us with binary progressing vs. nonprogressing results, it also produces a 'degree of confidence' score that is potentially useful by providing clinicians with the probability of progression. Most glaucoma patients progress. The main question is how fast and how reliable the observed trend is. The results indicated that the PoP approach led to detection of a similar proportion of progressing eyes at same high level of specificity compared to linear regression analysis of mean deviation and Visual Field Index and that it may provide additional information since it can also yield information with regard to topography of change in visual field series. Goldbaum and colleagues' approach is an important addition to our current armamentarium for detection of functional deterioration in glaucoma and the investigators are to be commended for that.
Comments
The comment section on the IGR website is restricted to WGA#One members only. Please log-in through your WGA#One account to continue.Log-in through WGA#One
