A mixed integer linear programming framework for improving cortical vision prosthesis designs. (February 2023)

• Record Type: Journal Article
• Authors: Abouelseoud, Gehan; Abouelseoud, Yasmine; Shoukry, Amin; Ismail, Nour; Mekky, Jaidaa
• Is Part Of: Biomedical signal processing and control. Volume 80(2023)Part 1
• Publication Date: 2023-02
• Subjects: Electrical brain stimulation -- Cortical vision prosthesis -- Mixed integer linear programming (MILP) -- Optimization; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.104253
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Although cortical vision prosthesis design has been an active area of research since the early 70s, progress in performance remains quite limited. This paper aims to develop a rigorous mathematical framework, based on mixed integer programming, for the analysis of the cortical vision prosthesis design problem and the optimal estimation of a prosthesis setup parameters. The simulated examples illustrate the unique capabilities of the proposed strategy in testing the feasibility of the goals of a prosthesis under a set of specified design constraints. Investigation of the results of the simulated examples suggest that a number of electrodes much larger than the number of cortical locations that are to be controlled is needed to achieve an improved vision experience, approaching the quality of normal vision (to reach a 300 pixels/inch resolution a million-electrode implantation is required). Thus, future research should focus on how these electrodes can be safely implanted. Once this is possible, the proposed framework can be applied to decide their optimal locations and excitation currents. Highlights: The challenges hindering the progress of cortical vision prostheses are reviewed. A mixed integer linear programming framework is proposed for problem analysis. The framework finds both optimal electrode currents and locations. Provable optimality finds minimal prosthesis setups capable of achieving design goals. Our results suggest: million electrodes are needed for near normal vision quality.