Closed-loop Control of Cortical Visual Neuroprosthetic

Background

Visual neuroprosthetics represent a promising frontier in restoring sight to individuals with visual impairments. As part of the NeuraViPeR project, we are investigating advanced methods to optimize electrically stimulated visual perceptions (phosphenes) through a closed-loop system. Our approach utilizes novel flexible electrodes for neural recording during brain stimulation experiments in rats, allowing us to study and optimize the relationship between electrical stimulation and perceived visual sensations.

Current challenges in visual prosthetics include optimizing stimulation patterns and developing real-time adaptive systems that can respond to neural feedback. This project leverages experimental data from rat models and simulation to advance our understanding and control of prosthetic vision.

Project Objectives

The master’s project focuses on developing and implementing closed-loop control strategies for visual neuroprosthetics. Key objectives include:

1. Analysis of cerebral recordings to optimize visual stimulation parameters
2. Development of neural network-based algorithms for stimulation pattern optimization
3. Implementation of real-time processing algorithms for closed-loop control

Methodology

The project will utilize:

• Neural recordings from rat experiments using flexible electrodes
• Phosphene simulator for training and validation
• Hardware platforms (Jetson/FPGA) for real-time implementation

Students can focus on one of several possible directions:

1. Analysis of cerebral recordings for closed-loop stimulation optimization
2. Development of DNN-based control algorithms using the phosphene simulator
3. Investigation of temporal dynamics in phosphene perception using SNNs
4. Implementation of real-time processing on embedded platforms

Requirements

• Strong programming skills (Python, C++)
• Background in signal processing and neural networks
• Knowledge of neural recording analysis (preferred)
• Experience with embedded systems (for hardware implementation track)
• Basic understanding of neuroscience principles
• Familiarity with deep learning frameworks (PyTorch/TensorFlow)