Diagnosis of multiple sclerosis using multifocal ERG data feature fusion.

The purpose of this paper is to implement a computer-aided diagnosis (CAD) system for multiple sclerosis (MS) based on analysing the outer retina as assessed by multifocal electroretinograms (mfERGs). MfERG recordings taken with the RETI-port/scan 21 (Roland Consult) device from 15 eyes of patients diagnosed with incipient relapsing-remitting MS and without prior optic neuritis, and from 6 eyes of control subjects, are selected. The mfERG recordings are grouped (whole macular visual field, five rings, and four quadrants). For each group, the correlation with a normative database of adaptively filtered signals, based on empirical model decomposition (EMD) and three features from the continuous wavelet transform (CWT) domain, are obtained. Of the initial 40 features, the 4 most relevant are selected in two stages: a) using a filter method and b) using a wrapper-feature selection method. The Support Vector Machine (SVM) is used as a classifier. With the optimal CAD configuration, a Matthews correlation coefficient value of 0.89 (accuracy = 0.95, specificity = 1.0 and sensitivity = 0.93) is obtained. This study identified an outer retina dysfunction in patients with recent MS by analysing the outer retina responses in the mfERG and employing an SVM as a classifier. In conclusion, a promising new electrophysiological-biomarker method based on feature fusion for MS diagnosis was identified.

Diagnostic ability of multifocal electroretinogram in early multiple sclerosis using a new signal analysis method.

PURPOSE: To determine if a novel analysis method will increase the diagnostic value of the multifocal electroretinogram (mfERG) in diagnosing early-stage multiple sclerosis (MS). METHODS: We studied the mfERG signals of OD (Oculus Dexter) eyes of fifteen patients diagnosed with early-stage MS (in all cases < 12 months) and without a history of optic neuritis (ON) (F:M = 11:4), and those of six controls (F:M = 3:3). We obtained values of amplitude and latency of N1 and P1 waves, and a method to assess normalized root-mean-square error (FNRMSE) between model signals and mfERG recordings was used. Responses of each eye were analysed at a global level, and by rings, quadrants and hemispheres. AUC (area under the ROC curve) is used as discriminant factor. RESULTS: The standard method of analysis obtains further discrimination between controls and MS in ring R3 (AUC = 0.82), analysing N1 waves amplitudes. In all of the retina analysis regions, FNRMSE value shows a greater discriminating power than the standard method. The highest AUC value (AUC = 0.91) was in the superior temporal quadrant. CONCLUSION: By analysing mfERG recordings and contrasting them with those of healthy controls it is possible to detect early-stage MS in patients without a previous history of ON.

Exploring the methods of data analysis in multifocal visual evoked potentials.

PURPOSE: The multifocal visual evoked potential (mfVEP) provides a topographical assessment of visual function, which has already shown potential for use in patients with glaucoma and multiple sclerosis. However, the variability in mfVEP measurements has limited its broader application. The purpose of this study was to compare several methods of data analysis to decrease mfVEP variability. METHODS: Twenty-three normal subjects underwent mfVEP testing. Monocular and interocular asymmetry data were analyzed. Coefficients of variability in amplitude were examined using peak-to-peak, root mean square (RMS), signal-to-noise ratio (SNR) and logSNR techniques. Coefficients of variability in latency were examined using second peak and cross-correlation methods. RESULTS: LogSNR and peak-to-peak methods had significantly lower intra-subject variability when compared with RMS and SNR methods. LogSNR had the lowest inter-subject amplitude variability when compared with peak-to-peak, RMS and SNR. Average latency asymmetry values for the cross-correlation analysis were 1.7 ms (CI 95 % 1.2-2.3 ms) and for the second peak analysis 2.5 ms (CI 95 % 1.7-3.3 ms). A significant difference was found between cross-correlation and second peak analysis for both intra-subject variability (p < 0.001) and inter-subject variability (p < 0.001). CONCLUSIONS: For a comparison of amplitude data between groups of patients, the logSNR or SNR methods are preferred because of the smaller inter-subject variability. LogSNR or peak-to-peak methods have lower intra-subject variability, so are recommended for comparing an individual mfVEP to previous published normative data. This study establishes that the choice of mfVEP data analysis method can be used to decrease variability of the mfVEP results.

A signal-to-noise-ratio-based analysis of multifocal visual-evoked potentials in multiple sclerosis risk assessment.

OBJECTIVE: To study the value of using the signal-to-noise ratio (SNR) of multifocal visual-evoked potentials (mfVEPs) in assessment of subjects at risk of developing multiple sclerosis (MS). METHODS: MfVEP signals were obtained from 15 patients with radiologically isolated syndrome (RIS), from 28 patients with clinically isolated syndrome (CIS), from 28 with clinically definite MS and from 24 control subjects. The CIS and MS groups were divided into two subgroups: those with eyes affected by optic neuritis (ON) and those without (non-ON). The mfVEPs' SNR was obtained for both the whole visual field and at various eccentric rings. The area under the curve (AUC) was calculated by comparing the control subjects' mfVEP SNR values with those of the RIS, CIS and MS groups. RESULTS: In whole visual field analysis, risk of developing MS increased as SNR decreased (SNRCONTROL=0.70, SNRRIS=0.62, SNRCIS-nonON=0.54, SNRCIS-ON=0.40, SNRMS-nonON=0.52, SNRMS-ON=0.40). Ring 5 (9.8°-15° eccentricity) was most affected by the SNR decrease, as indicated by its higher AUC values (AUCFULL_EYE=0.81, AUCRING_5=0.89). A significant relationship (Spearman correlation, ρRING_5=0.61) between SNR values and disability severity on the Expanded Disability Status Scale (EDSS) was observed in clinically definite MS patients. CONCLUSION: A new method based on analysis of the SNR of mfVEP signal amplitude improves assessment of patients at risk of developing MS. SIGNIFICANCE: Improved mfVEP assessment of MS-risk patients was achieved by using SNR values at 9.8°-15° eccentricity of the visual field.

Continuous-wavelet-transform analysis of the multifocal ERG waveform in glaucoma diagnosis.

The vast majority of multifocal electroretinogram (mfERG) signal analyses to detect glaucoma study the signals' amplitudes and latencies. The purpose of this paper is to investigate application of wavelet analysis of mfERG signals in diagnosis of glaucoma. This analysis method applies the continuous wavelet transform (CWT) to the signals, using the real Morlet wavelet. CWT coefficients resulting from the scale of maximum correlation are used as inputs to a neural network, which acts as a classifier. mfERG recordings are taken from the eyes of 47 subjects diagnosed with chronic open-angle glaucoma and from those of 24 healthy subjects. The high sensitivity in the classification (0.894) provides reliable detection of glaucomatous sectors, while the specificity achieved (0.844) reflects accurate detection of healthy sectors. The results obtained in this paper improve on the previous findings reported by the authors using the same visual stimuli and database.

Software for analysing multifocal visual evoked potential signal latency progression.

BACKGROUND: This paper describes a new non-commercial software application (mfVEP(2)) developed to process multifocal visual-evoked-potential (mfVEP) signals in latency (monocular and interocular) progression studies. METHOD: The software performs analysis by cross-correlating signals from the same patients. The criteria applied by the software include best channels, signal window, cross-correlation limits and signal-to-noise ratio (SNR). Software features include signal display comparing different tests and groups of sectors (quadrants, rings and hemispheres). RESULTS: The software's performance and capabilities are demonstrated on the results obtained from a patient with acute optic neuritis who underwent 9 follow-up mfVEP tests. Numerical values and graphics are presented and discussed for this case. CONCLUSIONS: The authors present a software application used to study progression in mfVEP signals. It is also useful in research projects designed to improve mfVEP techniques. This software makes it easier for users to manage the signals and allows them to choose various ways of selecting signals and representing results.

Filtering multifocal VEP signals using Prony's method.

BACKGROUND: This paper describes use of Prony's method as a filter applied to multifocal visual-evoked-potential (mfVEP) signals. Prony's method can be viewed as an extension of Fourier analysis that allows a signal to be decomposed into a linear combination of functions with different amplitudes, damping factors, frequencies and phase angles. METHOD: By selecting Prony method parameters, a frequency filter has been developed which improves signal-to-noise ratio (SNR). Three different criteria were applied to data recorded from control subjects to produce three separate datasets: unfiltered raw data, data filtered using the traditional method (fast Fourier transform: FFT), and data filtered using Prony's method. RESULTS: Filtering using Prony's method improved the signals' original SNR by 44.52%, while the FFT filter improved the SNR by 33.56%. The extent to which signal can be separated from noise was analysed using receiver-operating-characteristic (ROC) curves. The area under the curve (AUC) was greater in the signals filtered using Prony's method than in the original signals or in those filtered using the FFT. CONCLUSION: filtering using Prony's method improves the quality of mfVEP signal pre-processing when compared with the original signals, or with those filtered using the FFT.

Filtering mfVEP signals using Wiener filter.

The multifocal visual-evoked-potential (mfVEP) signals are filtered using the Wiener filter combined with a Fast Fourier Transform and their signal-to-noise ratios are compared against those of unfiltered signals (RAW data) and those of signals filtered using the traditional method (FFT data). The Wiener filter improves the original signals' SNR by 37.49%, while the FFT improves the SNR by 20.41%. This gain is achieved by selecting the best channel in each sector of the visual field. In conclusion, filtering using the Wieners method improves the quality of mfVEP signal pre-processing when compared against the original signals, or against filtering using the FFT.

Improved measurement of intersession latency in mfVEPs.

PURPOSE: The purpose of the study is to present a method (Selfcorr) by which to measure intersession latency differences between multifocal VEP (mfVEP) signals. METHODS: The authors compared the intersession latency difference obtained using a correlation method (Selfcorr) against that obtained using a Template method. While the Template method cross-correlates the subject's signals with a reference database, the Selfcorr method cross-correlates traces across subsequent recordings taken from the same subject. RESULTS: The variation in latency between intersession signals was 0.8 ± 13.6 and 0.5 ± 5.0 ms for the Template and Selfcorr methods, respectively, with a coefficient of variability CV_TEMPLATE = 15.83 and CV_SELFCORR = 5.68 (n = 18, p = 0.0002, Wilcoxon). The number of analyzable sectors with the Template and Selfcorr methods was 36.7 ± 8.5 and 45.3 ± 8.7, respectively (p = 0.0001, paired t test, two tailed). CONCLUSIONS: The Selfcorr method produces smaller intersession mfVEP delays and variability over time than the Template method.