Abstract
Purpose: :
Analysis of the multifocal electroretinogram (mfERG) is relatively simple when recordings are obtained under optimal conditions however interpretation can be subjective in the case of less cooperative patients. This becomes relevant when a grading of recording quality is required for example in multi-centre clinical trials. An objective, automated method of assessing the quality, during and after a mfERG test would be of benefit.We previously showed that analysis of the uncorrelated mfERG data in the Fourier domain reveals distinctive patterns corresponding to muscle movement and blinking/eye motion during testing1. The aim was therefore to grade the integrity of a recording as excellent, moderate, noisy or unreportable using the Fourier domain.
Methods: :
90 mfERG recordings were selected 30 of which were of a high quality. The remaining 60 contained unacceptable levels of noise and were divided into 2 groups: muscle noise and eye movement/blinking noise. The uncorrelated data of each test was transformed to the Fourier domain. Mean and standard deviation frequency profiles were calculated to define limits for excellent and unreportable recordings. The region between these limits was divided in two, providing limits for moderate and noisy.To test the system 50 additional mfERG recordings were selected. The quality of each mfERG recording was independently examined and classed as one of the four categories listed above. The automated system then graded each recording. The results of each were compared.
Results: :
There was an 80% agreement between the expert and the system. Of the remaining 20% the classifications given by the expert and the system were within one category of each other.
Conclusions: :
It is possible to grade the integrity of a recording reasonably well in the Fourier domain. The discrepancies between the expert and the system are however the subject of further investigation.1Foulis A, Parks S, Keating D, A Multilayered Approach to the Automatic Analysis of multifocal ERG Waveforms: Analysis of the Fourier Domain Profile, ARVO 2009, Program No. 4517, Poster No. 683
Keywords: electrophysiology: non-clinical • electrophysiology: clinical