Spatiotemporal integration of visual stimuli and its relevance to the use of a divisional power supply scheme for retinal prosthesis

A wireless photovoltaic retinal prosthesis is currently being studied with the aim of providing prosthetic vision to patients with retinitis pigmentosa (RP) and age-related macular degeneration (AMD). The major challenge of a photovoltaic device is its limited power efficiency. Our retinal prosthetic design implements a unique divisional power supply scheme (DPSS) system that provides the electrical power generated by all of the solar cells to only a subset of electrodes at any moment in time. The aim of the present study was to systematically characterize the spatiotemporal integration performance of the system under various DPSS conditions using human subjects and a psychophysical approach. A 16x16 pixels LED array controlled by Arduino was used to simulate the output signal of the DPSS design, and human performance under different visual stimulations at various update frequencies was then used to assess the spatiotemporal capability of retinal prostheses. The results showed that the contrast polarity of the image, image brightness, and division number influenced the lower limit of the update frequency of the DPSS system, while, on the other hand, visual angle, ambient light level, and stimulation order did not affect performance significantly. Pattern recognition by visual persistence with spatiotemporal integration of multiple frames of sparse dots is a feasible approach in retinal prosthesis design. These findings provide an insight into how to optimize a photovoltaic retinal prosthesis using a DPSS design with an appropriate update frequency for reliable pattern recognition. This will help the development of a wireless device able to restore vision to RP and AMD patients in the future.当前学界正对无线光伏视网膜假体（wireless photovoltaic retinal prosthesis）开展研究，以期为色素性视网膜炎（retinitis pigmentosa, RP）及年龄相关性黄斑变性（age-related macular degeneration, AMD）患者提供假体视觉。该类光伏器件的核心挑战在于功率效率受限。本研究提出的视网膜假体设计采用了独特的分区供电方案（divisional power supply scheme, DPSS）系统，该系统可将所有太阳能电池产生的电能仅供给任一时刻的部分电极。本研究旨在通过人体受试者与心理物理学方法，系统表征该系统在不同DPSS条件下的时空整合性能。本研究采用Arduino控制的16×16像素发光二极管（light-emitting diode, LED）阵列模拟DPSS设计的输出信号，并通过不同更新频率视觉刺激下的人类视觉表现，评估视网膜假体的时空性能。结果显示，图像对比度极性、图像亮度与分区数量会影响DPSS系统更新频率的下限；而视角、环境光照水平及刺激顺序对性能无显著影响。基于视觉暂留与多帧稀疏点时空整合的模式识别方法，在视网膜假体设计中具备可行性。上述研究结果为如何通过采用合适更新频率的DPSS设计优化光伏视网膜假体以实现可靠的模式识别提供了理论依据。这将助力未来研发可帮助RP与AMD患者恢复视力的无线视觉假体设备。