I. Introduction
Biological neurons can be considered highly efficient signal processing units [1], [2] and hold significant potential for future artificial intelligence research. Although numerous methods based on electrical signal detection or optical imaging have been utilized to study the dynamics of in vitro neuronal networks, these methods have certain limitations. For example, methods based on electrical signal recording suffer from low spatial resolution and the mixing of signals from multiple neurons [3]. On the other hand, optical methods, such as calcium imaging, encounter issues with low temporal resolution [4]. Additionally, electrical stimulation techniques for controlling neural network activity face challenges with low spatial resolution.