Visual Prostheses in the Era of Artificial Intelligence Technology.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence
• Ophthalmology

Background:

• Technological advancements have enabled invasive visual prostheses, yet functional outcomes, particularly visual acuity, remain limited.
• Retinal and cortical prostheses (RCPs) are emerging technologies aiming to restore vision.
• Artificial intelligence (AI) presents a potential avenue for enhancing RCP performance.

Purpose of the Study:

• To review current developments in retinal and cortical prostheses (RCPs).
• To critically assess the role of AI in advancing visual prostheses.
• To systematically review AI-driven image and signal processing for improved clinical outcomes.

Main Methods:

• A systematic literature review was conducted using PubMed and Elicit.
• 455 studies were screened, with 28 included for analysis.
• Focus on AI applications in image saliency extraction and stimulation-perception consistency.

Main Results:

• AI applications in RCPs primarily focus on saliency extraction and stimulation consistency.
• Artificial neural networks were used in 14 of 28 studies, with 12 involving model training.
• While 22 studies used empirical data, 15 relied on simulated prosthetic vision.

Conclusions:

• AI algorithms demonstrate potential for optimizing prosthetic vision through improved image processing and stimulation.
• Current research heavily relies on simulations, with limited validation in real-world settings.
• Clinical validation with blind patients is essential to confirm the effectiveness of AI-enhanced visual prostheses.