A comparison among different techniques for human ERG signals processing and classification

Synthetic electroretinogram signal generation using a conditional generative adversarial network

PURPOSE

The electroretinogram (ERG) records the functional response of the retina. In some neurological conditions, the ERG waveform may be altered and could support biomarker discovery. In heterogeneous or rare populations, where either large data sets or the availability of data may be a challenge, synthetic signals with Artificial Intelligence (AI) may help to mitigate against these factors to support classification models.

METHODS

This approach was tested using a publicly available dataset of real ERGs, n = 560 (ASD) and n = 498 (Control) recorded at 9 different flash strengths from n = 18 ASD (mean age 12.2 ± 2.7 years) and n = 31 Controls (mean age 11.8 ± 3.3 years) that were augmented with synthetic waveforms, generated through a Conditional Generative Adversarial Network. Two deep learning models were used to classify the groups using either the real only or combined real and synthetic ERGs. One was a Time Series Transformer (with waveforms in their original form) and the second was a Visual Transformer model utilizing images of the wavelets derived from a Continuous Wavelet Transform of the ERGs. Model performance at classifying the groups was evaluated with Balanced Accuracy (BA) as the main outcome measure.

RESULTS

The BA improved from 0.756 to 0.879 when synthetic ERGs were included across all recordings for the training of the Time Series Transformer. This model also achieved the best performance with a BA of 0.89 using real and synthetic waveforms from a single flash strength of 0.95 log cd s m-2.

CONCLUSIONS

The improved performance of the deep learning models with synthetic waveforms supports the application of AI to improve group classification with ERG recordings.

Novel Machine-Learning Based Framework Using Electroretinography Data for the Detection of Early-Stage Glaucoma

Purpose

Early-stage glaucoma diagnosis has been a challenging problem in ophthalmology. The current state-of-the-art glaucoma diagnosis techniques do not completely leverage the functional measures' such as electroretinogram's immense potential; instead, focus is on structural measures like optical coherence tomography. The current study aims to take a foundational step toward the development of a novel and reliable predictive framework for early detection of glaucoma using machine-learning-based algorithm capable of leveraging medically relevant information that ERG signals contain.

Methods

ERG signals from 60 eyes of DBA/2 mice were grouped for binary classification based on age. The signals were also grouped based on intraocular pressure (IOP) for multiclass classification. Statistical and wavelet-based features were engineered and extracted. Important predictors (ERG tests and features) were determined, and the performance of five machine learning-based methods were evaluated.

Results

Random forest (bagged trees) ensemble classifier provided the best performance in both binary and multiclass classification of ERG signals. An accuracy of 91.7 and 80% was achieved for binary and multiclass classification, respectively, suggesting that machine-learning-based models can detect subtle changes in ERG signals if trained using advanced features such as those based on wavelet analyses.

Conclusions

The present study describes a novel, machine-learning-based method to analyze ERG signals providing additional information that may be used to detect early-stage glaucoma. Based on promising performance metrics obtained using the proposed machine-learning-based framework leveraging an established ERG data set, we conclude that the novel framework allows for detection of functional deficits of early/various stages of glaucoma in mice.

Oscillatory Potentials in Achromatopsia as a Tool for Understanding Cone Retinal Functions

Achromatopsia (ACHM) is an inherited autosomal recessive disease lacking cone photoreceptors functions. In this study, we characterize the time-frequency representation of the full-field electroretinogram (ffERG) component oscillatory potentials (OPs), to investigate the connections between photoreceptors and the inner retinal network using ACHM as a model. Time-frequency characterization of OPs was extracted from 52 controls and 41 achromat individuals. The stimulation via ffERG was delivered under dark-adaptation (DA, 3.0 and 10.0 cd·s·m-2) to assess mixed rod-cone responses. The ffERG signal was subsequently analyzed using a continuous complex Morlet transform. Time-frequency maps of both DA conditions show the characterization of OPs, disclosing in both groups two distinct time-frequency windows (~70-100 Hz and >100 Hz) within 50 ms. Our main result indicates a significant cluster (p < 0.05) in both conditions of reduced relative power (dB) in ACHM people compared to controls, mainly at the time-frequency window >100 Hz. These results suggest that the strongly reduced but not absent activity of OPs above 100 Hz is mostly driven by cones and only in small part by rods. Thus, the lack of cone modulation of OPs gives important insights into interactions between photoreceptors and the inner retinal network and can be used as a biomarker for monitoring cone connection to the inner retina.

Advanced Bioelectrical Signal Processing Methods: Past, Present, and Future Approach-Part III: Other Biosignals

Analysis of biomedical signals is a very challenging task involving implementation of various advanced signal processing methods. This area is rapidly developing. This paper is a Part III paper, where the most popular and efficient digital signal processing methods are presented. This paper covers the following bioelectrical signals and their processing methods: electromyography (EMG), electroneurography (ENG), electrogastrography (EGG), electrooculography (EOG), electroretinography (ERG), and electrohysterography (EHG).

Continuous wavelet transform analysis of ERG in patients with diabetic retinopathy

PURPOSE

Diabetic retinopathy (DR) is one of the leading causes of blindness worldwide. Non-proliferative diabetic retinopathy (NPDR) is a stage of the disease that contains morphological and functional disruption of the retinal vasculature and dysfunction of retinal neurons. This study aimed to compare time and time-frequency-domain analysis in the evaluation of electroretinograms (ERGs) in subjects with NPDR.

METHOD

The ERG responses were recorded in 16 eyes from 12 patients with NPDR and 24 eyes from 12 healthy subjects as the control group. The implicit time, amplitude, and time-frequency-domain parameters of photopic and scotopic ERGs were analyzed.

RESULTS

The implicit times of b-waves in the dark-adapted 10.0 (P = 0.0513) and light-adapted 3.0 (P = 0.0414) were significantly increased in the NPDR group. The amplitudes of a- and b-wave showed a significantly decreased dark-adapted 10.0 (P = 0.0212; P = 0.0133) and light-adapted 3.0 (P = 0.0517; P = 0.0021) ERG of the NPDR group. The Cohen's d effect size had higher values in the amplitude of dark-adapted 10.0 b-wave (|d|= 1.8058) and amplitude of light-adapted 3.0 b-wave (|d|= 1.9662). The CWT results showed that the frequency ranges of the dominant components in dark-adapted 10.0 and light-adapted 3.0 ERG were decreased in the NPDR group compared to the healthy group (P < 0.05). The times associated with the NDPR group's dominant components were increased compared to normal eyes in both dark-adapted 10.0 and light-adapted 3.0 ERG (P < 0.05). All Cohen's d effect sizes of the implicit times and dominant frequency components were on a large scale (|d|> 1).

CONCLUSION

These findings suggest that the time and time-frequency parameters of both photopic and scotopic ERGs can be good indicators for DR. However, time-frequency-domain analysis could present more information might be helpful in the assessment of the DR severity.

Time-Frequency Analysis of Photopic Negative Response in CRVO Patients

PURPOSE

The PhNR is driven by retinal ganglion cells (RGCs). Therefore, the function of RGCs could be objectively evaluated by analyzing the PhNR. The aim of this article is to determine the effect of central retinal vein occlusion (CRVO) on PhNR and RGCs performances.

METHODS

Seventeen patients with CRVO were included. Full-field photopic ERGs, including PhNR, were recorded and compared with the fellow normal eyes. ERG signals were analyzed based on the standard time-domain analyses of the PhNR as well as a continuous wavelet transform (CWT) to extract time-frequency components that correspond to the PhNR using MATLAB. We obtained the main frequencies and their occurrence time from CWT.

RESULTS

All a-wave, b-wave, and PhNR amplitudes of CRVO eyes showed a significant reduction compared to those of the fellow eyes (P < .01, P < .001, and P < .001, respectively). The peak times of a-wave, b-wave, and PhNR were increased significantly in the CRVO eyes (P = .04, P = .04, and P = .003, respectively). The dominant f₃ frequency, which corresponds to the PhNR in CRVO patients, showed a more significant decrease (P < .001) compared to other dominant frequencies (f₀, f₁, and f₂). The occurrence time of f₃ (t₃) was significantly higher in the CRVO eyes (P < .001). Time-domain of the PhNR was also affected in CRVO patients (P < .001).

CONCLUSION

CWT allows quantifications of ERG responses, especially for PhNR. The PhNR was severely affected in CRVO eyes implicating loss of RGCs. CWT might demonstrate the severity of CRVO more precisely and identify diagnostically significant changes of ERG waveforms that are not resolved when the analysis is only limited to the time-domain measurements.

A novel method for electroretinogram assessment in patients with central retinal vein occlusion

PURPOSE

Central retinal vein occlusion (CRVO) is the second most common retinal vascular disorder after diabetic retinopathy that affects the eyes. We propose a method for distinction of normal and central CRVO eyes based on electroretinogram (ERG).

METHODS

Seventeen patients with CRVO in one eye were analyzed. Their ERG signals were collected in six different stimuli, including four records in the darkness (dark-adapted 0.01, dark-adapted 3.0, dark-adapted oscillatory potentials, and dark-adapted 10) and two records in brightness (light-adapted 3.0 and light-adapted 30 Hz flicker). Nonlinear features such as Hurst exponent (HE) and approximate entropy (ApEn) were extracted from healthy and CRVO eyes. Finally, a parabolic mapping and two criteria (theta angle and the density of points) were proposed to distinguish the groups.

RESULTS

For ApEn, the P values of dark-adapted 3.0 oscillatory (P = 0.0433) and flicker (P = 0.0425) confirmed significant differences between the groups. For HE, the P values of dark-adapted 3.0 oscillatory (P = 0.0421) and flicker 30 Hz (P = 0.0402) confirmed differences between the healthy and CRVO groups. The P values of theta angle for dark-adapted 3.0 (P = 0.0199), dark-adapted oscillatory (P = 0.0265), dark-adapted 10.0 (P = 0.0166), light-adapted 3.0 (P = 0.0411), and flicker (P = 0.0361) showed significant differences. Using the density criterion, the statistical test demonstrated a significant difference between the groups in dark-adapted 3 (P = 0.0038), dark-adapted oscillatory (P = 0.0102), dark-adapted 10.0 (P = 0.0071), light-adapted 3.0 (P = 0.0319), and flicker 30 Hz (P = 0.0076).

CONCLUSION

The proposed features have made it possible to distinguish between healthy and CRVO eyes. This method could be helpful in some cases with no definite diagnosis or to estimate the severity of CRVO.

Time and frequency components of ERG responses in retinitis pigmentosa

PURPOSE

To evaluate the effects of retinitis pigmentosa (RP) on time, frequency, and time-frequency components of Xenon flash ERG signals using Fourier and wavelet transforms.

METHODS

Xenon flash ERG was done in 18 eyes of nine RP patients and 20 normal eyes. After examining latency and amplitude, Fourier and wavelet transforms were performed using MATLAB software. Then, we extracted the mode frequency from the Fourier transform and main frequencies and their occurrence time from the wavelet transform. Finally, mean differences were analyzed using statistical tests.

RESULTS

The results indicated increased latency and reduced ERG wave amplitude, no significant inter-group difference in the average mode frequency, and significant reduction in main signal frequencies and their increased occurrence times. Also one or two of the three main frequencies had disappeared in more advanced cases.

CONCLUSION

Retinitis pigmentosa can induce changes in ERG time and time-frequency components. Impacted areas can be identified more accurately by wavelet transform and converting scales to frequencies.

Wavelet decomposition analysis in the two-flash multifocal ERG in early glaucoma: a comparison to ganglion cell analysis and visual field

PURPOSE

To further improve analysis of the two-flash multifocal electroretinogram (2F-mfERG) in glaucoma in regard to structure-function analysis, using discrete wavelet transform (DWT) analysis.

METHODS

Sixty subjects [35 controls and 25 primary open-angle glaucoma (POAG)] underwent 2F-mfERG. Responses were analyzed with the DWT. The DWT level that could best separate POAG from controls was compared to the root-mean-square (RMS) calculations previously used in the analysis of the 2F-mfERG. In a subgroup analysis, structure-function correlation was assessed between DWT, optical coherence tomography and automated perimetry (mf103 customized pattern) for the central 15°.

RESULTS

Frequency level 4 of the wavelet variance analysis (144 Hz, WVA-144) was most sensitive (p < 0.003). It correlated positively with RMS but had a better AUC. Positive relations were found between visual field, WVA-144 and GCIPL thickness. The highest predictive factor for glaucoma diagnostic was seen in the GCIPL, but this improved further by adding the mean sensitivity and WVA-144.

CONCLUSIONS

mfERG using WVA analysis improves glaucoma diagnosis, especially when combined with GCIPL and MS.

Witnessing the first sign of retinitis pigmentosa onset in the allegedly normal eye of a case of unilateral RP: a 30-year follow-up

PURPOSE

A patient initially presented with constricted visual field, attenuated retinal vasculature, pigmentary clumping and reduced ERG in OS only, suggestive of unilateral retinitis pigmentosa (RP). This patient was subsequently seen on eight occasions (over three decades), and, with time, the initially normal eye (OD) gradually showed signs of RP-like degeneration. The purpose of this study was to evaluate which clinical modality (visual field, funduscopy or electroretinography) could have first predicted this fate.

METHODS

At each time points, data obtained from our patient were compared to normative data using Z tests.

RESULTS

At initial visit, all tests were significantly (p < 0.05) altered in OS and normal in OD. Visual field and retinal vessel diameter in OD reduced gradually to reach statistical significance at the 5th visit and 6th visit (21 and 22 years after the first examination, respectively). In OD, the amplitude of the scotopic and photopic ERGs reduced gradually and was significantly smaller than normal at the 2nd visit (after 11 years) and 3rd visit (after 18 years), respectively. When the photopic ERG was analyzed using the discrete wavelet transform (DWT), we were able to detect a significant change at the 2nd visit (after 11 years) instead of the 3rd visit (18 years).

CONCLUSIONS

Our study allowed us to witness the earliest manifestation of an RP disease process. The ERG was the first test to detect significant RP changes. A significantly earlier detection of ERG anomalies was obtained when the DWT was used, demonstrating its advantage for early detection of ERG changes.

Wavelet analysis of pulse-amplitude-modulated chlorophyll fluorescence for differentiation of plant samples

Pulse-amplitude-modulated (PAM) chlorophyll fluorescence (ChlF) from photosystem II (PSII) of plants has been routinely measured for the analysis of photosynthesis and environmental changes. PAM ChlF from PSII is non-stationary and has time-varying frequency characteristics; however, existing analysis of PAM ChlF has been limited to selected characteristic values in the time domain. Wavelet transform is recognized as an efficient tool for analyzing non-stationary signals. In this research, an attempt was made to analyze PAM ChlF through wavelet transform. Features of PAM ChlF signals were computed from wavelet decomposition to classify two tree species and to detect chilling and detachment stresses. The wavelet-based features were compared with the commonly-used maximal PSII efficiency Fv/Fm. Both the wavelet-based features and Fv/Fm could effectively classify two tree species, but the former showed superiority than the latter in detecting the stresses. Wavelet transform revealed chilling stress earlier than Fv/Fm and detected detachment stress Fv/Fm failed to show. The results show that wavelet transform is a useful technique for analysis of PAM ChlF.

Empirical mode decomposition and neural network for the classification of electroretinographic data

The processing of biosignals is increasingly being utilized in ambulatory situations in order to extract significant signals' features that can help in clinical diagnosis. However, this task is hampered by the fact that biomedical signals exhibit a complex behavior characterized by strong nonlinear and non-stationary properties that cannot always be perceived by simple visual examination. New processing methods need be considered. In this context, we propose a signal processing method, based on empirical mode decomposition and artificial neural networks, to analyze electroretinograms, i.e., the retinal response to a light flash, with the aim to detect and classify retinal diseases. The present application focuses on two retinal pathologies: achromatopsia, which is a cone disease, and congenital stationary night blindness, which affects the photoreceptoral signal transmission. The results indicate that, under suitable conditions, the method proposed here has the potential to provide a powerful tool for routine clinical examinations, since it is able to recognize with high level of confidence the eventual presence of one of the two pathologies.