Patient-specific computational models of retinal prostheses.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Ophthalmology

Background:

• Retinal prostheses aim to restore vision by stimulating inner retinal neurons.
• Variability in electrode perception and phosphene characteristics necessitates manual thresholding.
• Computational models offer a potential solution for automating device programming and predicting variability.

Purpose of the Study:

• To investigate inter-electrode variability in visual perception among epiretinal prosthesis users.
• To develop and validate patient-specific computational models for predicting perceptual thresholds.
• To explore the relationship between anatomical factors and perceptual responses.

Main Methods:

• Statistical analysis of perceptual thresholds across seven retinal prosthesis users.
• Development of patient-specific field-cable models using optical coherence tomography (OCT) images.
• Correlation analysis between anatomical parameters (retinal thickness, electrode-retina distance, impedance) and perceptual thresholds.

Main Results:

• Retinal thickness beneath the electrode significantly correlated with perceptual threshold across participants.
• Electrode-retina distance and impedance showed individual-specific correlations with perceptual threshold.
• Patient-specific models accurately predicted perceptual thresholds for 80% of participants and could predict retinal activity and phosphene size.

Conclusions:

• Retinal thickness is a significant predictor of perceptual threshold in retinal prostheses.
• Patient-specific computational models, built from OCT data, can accurately predict visual perception and aid in device programming.
• These in silico models can optimize stimulation settings, reducing clinical trial-and-error.