Compact semiconductor light-emitting diodes for dynamic imaging of neuronal circuitry

A fundamental goal of neuroscience is to study how networks of neurons process information collectively. The study of population activity using conventional patch-clamp technique is limited to sampling the electrical activity of at most few cells. Optical techniques, such as voltage-sensitive dye (VSD) imaging and flash photolysis methods, offer a complementary non-invasive approach to study the activity of neuronal networks. While in the VSD imaging, a potentiometric fluorescent probe (molecular dye) detects the change in membrane potential and reports it as a change in its intensity or spectrum, in the flash photolysis technique, "caged" neurotransmitters that are released with UV pulses are used to trigger the neurons locally and hence map out the connectivity of complex networks. These optical techniques are usually dependent on excitation sources such as bulky incoherent lamps or large frame lasers. In this thesis, compact diode pumped lasers and custom-designed UV and blue/green light emitting diodes (LEDs) are introduced as flexible means to perform dynamic imaging of neural circuitry.; High-efficiency gallium-nitride-based LEDs having individual element sizes comparable to typical biological cells have been fabricated and operated in proximity-illumination mode for individual neurons. Periodic arrays of these emitters have been designed and fabricated to image the activity from a network of cultured hippocampal neurons. The green emitter arrays were used to record the VSD activity from cultured hippocampal neurons and the UV LEDs were used to trigger the activity of the neurons by locally uncaging glutamate, thus eliminating the need for expensive lamps and lasers. With direct electrical control of each LED, a spatially periodic multi-neuronal target can be excited in a predesigned temporal sequence, thereby providing a new approach to dynamic recording of small neural circuits. For cells that have been cultured on patterned periodic templates, this approach is part of our long term goal to develop a new type of dynamical imaging approach to neural networks, as well as to achieve an active "chip scale" interface between a neural and a man-made (optoelectronic) circuit.