Evaluation of electroretinography (ERG) parameters as a biomarker for ADHD

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.

Detecting Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder Using Multimodal Time-Frequency Analysis with Machine Learning Using the Electroretinogram from Two Flash Strengths

PURPOSE

Autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) are conditions that similarly alter cognitive functioning ability and challenge the social interaction, attention, and communication skills of affected individuals. Yet these are distinct neurological conditions that can exhibit diverse characteristics which require different management strategies. It is desirable to develop tools to assist with early distinction so that appropriate early interventions and support may be tailored to an individual's specific requirements. The current diagnostic procedures for ASD and ADHD require a multidisciplinary approach and can be lengthy. This study investigated the potential of electroretinogram (ERG), an eye test measuring retinal responses to light, for rapid screening of ASD and ADHD.

METHODS

Previous studies identified differences in ERG amplitude between ASD and ADHD, but this study explored time-frequency analysis (TFS) to capture dynamic changes in the signal. ERG data from 286 subjects (146 control, 94 ASD, 46 ADHD) was analyzed using two TFS techniques.

RESULTS

Key features were selected, and machine learning models were trained to classify individuals based on their ERG response. The best model achieved 70% overall accuracy in distinguishing control, ASD, and ADHD groups.

CONCLUSION

The ERG to the stronger flash strength provided better separation and the high frequency dynamics (80-300 Hz) were more informative features than lower frequency components. To further improve classification a greater number of different flash strengths may be required along with a discrimination comparison to participants who meet both ASD and ADHD classifications and carry both diagnoses.

Sex-dependent variations of retinal function and architecture in a neurofibromatosis type I mouse model with normal vision

We aimed to characterize the structure and function of the early visual system of the neurofibromatosis type 1 (NF1) mouse model, a syndromic model of autism spectrum disorders (ASD). We used Nf1+/- mice and WT littermates and performed retinal structural analysis by optical coherence tomography (OCT), and functional assessment by electrophysiological recordings. We then performed behavioral visual tests using optomotor response (OMR) and sensitivity to visual stimulus familiarity. From the structural analysis, we found increased thickness for ganglion cell layer-inner plexiform layer (GCL-IPL) and outer nuclear layer (ONL) in male Nf1+/- mice compared with WT littermates. Regarding retinal electrophysiology, female Nf1+/- mice exhibited increased amplitudes for the second oscillatory potential (OP2) compared with WT littermates. Nevertheless, both Nf1+/- and WT mice presented normal visual acuity as measured by OMR and were able to exhibit regular visual stimulus familiarity responses. While structural sex-dependent changes are in line with previous results for brain anatomic measures, the subtle sex-dependent changes in oscillatory activity may relate to GABAergic neurotransmission changes found in NF1. Overall, these structural and functional changes do not seem to translate into visual behavioral alterations.

Retinal fundus imaging as biomarker for ADHD using machine learning for screening and visual attention stratification

Attention-deficit/hyperactivity disorder (ADHD), characterized by diagnostic complexity and symptom heterogeneity, is a prevalent neurodevelopmental disorder. Here, we explored the machine learning (ML) analysis of retinal fundus photographs as a noninvasive biomarker for ADHD screening and stratification of executive function (EF) deficits. From April to October 2022, 323 children and adolescents with ADHD were recruited from two tertiary South Korean hospitals, and the age- and sex-matched individuals with typical development were retrospectively collected. We used the AutoMorph pipeline to extract retinal features and used four types of ML models for ADHD screening and EF subdomain prediction, and we adopted the Shapely additive explanation method. ADHD screening models achieved 95.5%-96.9% AUROC. For EF function stratification, the visual and auditory subdomains showed strong (AUROC > 85%) and poor performances, respectively. Our analysis of retinal fundus photographs demonstrated potential as a noninvasive biomarker for ADHD screening and EF deficit stratification in the visual attention domain.

Validating skin electrodes: Paving the way for non-invasive ERG use in psychiatry

BACKGROUND

Electroretinography (ERG) shows promise for identifying psychiatric biomarkers. The gold-standard approach relies on Dawson-Trick-Litzkow (DTL) electrodes and desktop equipment, but high costs and the expertise required for reliable electrode placement limit its implementation. This study evaluates a cost-effective alternative: a handheld ERG device paired with less invasive self-adhesive skin electrodes, which require minimal training.

METHODS

ERG responses to cone and rod luminance stimuli were recorded from 49 participants: 15 controls (8 women, 7 men), 18 with bipolar disorder (12 women, 6 men), and 16 with schizophrenia (4 women, 12 men). Each participant underwent ERG testing with both electrode types.

RESULTS

Skin electrodes produced significantly smaller amplitudes and shorter latencies than DTL electrodes, except for longer scotopic b-wave latency. Women showed higher amplitudes and shorter latencies than men for both electrode types, with the photopic a-wave amplitude relative difference doubling when using skin electrodes. Reproducibility between eyes was high for both electrode types, though slightly lower for photopic a-wave with skin electrodes (ICC 0.76 vs. 0.86 for DTL).

CONCLUSION

Skin electrodes paired with handheld ERG devices offer a viable, accessible alternative to traditional DTL electrodes paired with desktop equipment. This approach has the potential to expand the applicability of ERG in clinical settings by addressing barriers like cost, complexity, and invasiveness, while highlighting the need to consider sex differences in ERG assessments, particularly with skin electrodes.

Remodeling the light-adapted electroretinogram using a bayesian statistical approach

OBJECTIVE

To present a remodeling of the electroretinogram waveform using a covariance matrix to identify regions of interest and distinction between a control and attention deficit/hyperactivity disorder (ADHD) group. Electroretinograms were recorded in n = 25 ADHD (16 male; age 11.9 ± 2.7 years) and n = 38 (8 male; age 10.4 ± 2.8 years neurotypical control participants as part of a broad study into the determining if the electroretinogram could be a biomarker for ADHD. Flash strengths of 0.6 and 1.2 log cd.s.m- 2 on a white 40 cd.m- 2 background were used. Averaged waveforms from each eye and flash strength were analyzed with Bayesian regularization of the covariance matrices using 100 equal length time intervals. The eigenvalues of the covariance matrices were ranked for each group to indicate the degree of complexity within the regularized waveforms.

RESULTS

The correlation matrices indicated less correlation within the waveforms for the ADHD group in time intervals beyond 70 msec. The eigenvalue plots suggest more complexity within the ADHD group compared to the control group. Consideration of the correlation structure between ERG waveforms from different populations may reveal additional features for identifying group differences in clinical populations.

Listening habits and subjective effects of background music in young adults with and without ADHD

Adults listen to an average of 20.7 hours of music per week, according to a study conducted across 26 countries. Numerous studies indicate that listening to music can have beneficial effects on cognitive performance and emotional well-being. Music listening habits may vary depending on individual needs and listening contexts. However, a limited number of studies have specifically examined the patterns of background music usage during various more or less cognitive activities, especially among individuals with attentional difficulties related to ADHD. This study primarily aimed to compare music listening habits during daily activities that are more and less cognitive (e.g., studying, problem-solving versus cleaning, engaging in sports) between neurotypical young adults and those screened for ADHD (respondents who were identified as likely having ADHD based on the number of self-reported symptoms). To achieve this, 434 young adults aged 17 to 30 responded to an online survey. The results indicate that certain listening habits differ significantly between the neurotypical and ADHD-screened groups. The ADHD-screened group reports significantly more background music listening during less cognitive activities and while studying, compared to the neurotypical group. The results also reveal a difference in the proportion of individuals preferring stimulating music between the groups: ADHD-screened individuals report significantly more frequent listening to stimulating music, regardless of the activity type (more or less cognitive). Other aspects of music listening are common to both groups. Regardless of the group, more respondents reported preferring to listen to relaxing, instrumental, familiar and self-chosen music during more cognitive activities, whereas for less cognitive activities, more individuals mentioned preferring to listen to music that is stimulating, with lyrics, familiar and self-chosen. Overall, the results confirm that most young adults listen to music during their daily activities and perceive positive effects from this listening.

Spectral Analysis of Light-Adapted Electroretinograms in Neurodevelopmental Disorders: Classification with Machine Learning

Electroretinograms (ERGs) show differences between typically developing populations and those with a diagnosis of autism spectrum disorder (ASD) or attention deficit/hyperactivity disorder (ADHD). In a series of ERGs collected in ASD (n = 77), ADHD (n = 43), ASD + ADHD (n = 21), and control (n = 137) groups, this analysis explores the use of machine learning and feature selection techniques to improve the classification between these clinically defined groups. Standard time domain and signal analysis features were evaluated in different machine learning models. For ASD classification, a balanced accuracy (BA) of 0.87 was achieved for male participants. For ADHD, a BA of 0.84 was achieved for female participants. When a three-group model (ASD, ADHD, and control) the BA was lower, at 0.70, and fell further to 0.53 when all groups were included (ASD, ADHD, ASD + ADHD, and control). The findings support a role for the ERG in establishing a broad two-group classification of ASD or ADHD, but the model's performance depends upon sex and is limited when multiple classes are included in machine learning modeling.

Retinal electrophysiology in central nervous system disorders. A review of human and mouse studies

The retina and brain share similar neurochemistry and neurodevelopmental origins, with the retina, often viewed as a "window to the brain." With retinal measures of structure and function becoming easier to obtain in clinical populations there is a growing interest in using retinal findings as potential biomarkers for disorders affecting the central nervous system. Functional retinal biomarkers, such as the electroretinogram, show promise in neurological disorders, despite having limitations imposed by the existence of overlapping genetic markers, clinical traits or the effects of medications that may reduce their specificity in some conditions. This narrative review summarizes the principal functional retinal findings in central nervous system disorders and related mouse models and provides a background to the main excitatory and inhibitory retinal neurotransmitters that have been implicated to explain the visual electrophysiological findings. These changes in retinal neurochemistry may contribute to our understanding of these conditions based on the findings of retinal electrophysiological tests such as the flash, pattern, multifocal electroretinograms, and electro-oculogram. It is likely that future applications of signal analysis and machine learning algorithms will offer new insights into the pathophysiology, classification, and progression of these clinical disorders including autism, attention deficit/hyperactivity disorder, bipolar disorder, schizophrenia, depression, Parkinson's, and Alzheimer's disease. New clinical applications of visual electrophysiology to this field may lead to earlier, more accurate diagnoses and better targeted therapeutic interventions benefiting individual patients and clinicians managing these individuals and their families.