Toward OLED-controlled live neural networks

Toward OLED-controlled live neural networks

Technology News |
In a paper titled "Arrays of microscopic organic LEDs for high-resolution optogenetics" published in the Advance Sciences journal, scientists at the University of St Andrews detail how they leveraged very fine pixel pitch OLEDs in place of a Petri dish to individually trigger living cells and observe their electric activity.
By Julien Happich


Working with Dr Gareth Miles from the School of Psychology and Neuroscience, Professor Gather and his team used OLEDs as a substrate to manipulate individual, live cells from a human embryonic kidney cell line that were tweaked to produce a light-sensitive protein. Upon exposure to blue OLED light from pixels directly underneath the cell, the researchers stimulated the electric activity of individual targeted cells, while neighbouring cells remained in the dark and stayed inactive.

The real first here, was that the researchers used a 20mm2 OLED microarray (from Fraunhofer FEP, based on a CMOS backplane featuring 230,000 individually addressable pixels) with 6×9μm2 pixels, smaller than the actual cells under study. This enabled the team to optically stimulate not only discrete cells but also different parts of a given cell.

The OLED microarray with cells adhered on top of the array (not drawn to scale). The microarray is connected to a high-definition multimedia interface (HDMI) driver with a flexible connector. Each pixel of the array can be turned on and off by the driver and the CMOS backplane. Light-induced changes in cell membrane current are measured with a patch clamp electrode (voltage clamp mode, whole-cell configuration). The cross section on the right shows the layer structure of the OLED array.

To ensure the shortest optical path (from the OLED pixels used as a substrate) to the cells under study while protecting the OLED active layers, the team only applied a 1.5μm thin-film encapsulation barrier (three layers of Al2O3 and two layers of polymer), which they had reported, was enough to keep the OLED functional over several days without noticeable degradation even when fully immersed into a salt buffer solution. The microarray was then bonded to a flexible flat cable that connects to a custom HDMI driver interface.

In their experimental setup, the team used a standard cell line to test their approach, with a micro-electrode placed manually to collect the electric response from a given cell. By lighting discrete OLED pixels underneath the modified photo-switchable cells, the researchers were able to switch the cell’s membrane potential conductivity states, testing different ion channel activation and deactivation kinetics.

Testing different ion channel activation and deactivation kinetics through localized illumination.

Depending on the cells used, these ranged from 1ms for activation and 21ms for deactivation to bi-stable switching characteristics (the conductive channel being opened or closed by different wavelength exposures). In fact, the emission spectrum of the OLEDs was tuned to match the spectral response of the cells under study.

Probing dozens or even thousands of cells in parallel would quickly become tedious if only relying on individually placed electrodes, recognized Gather in an interview with eeNews Europe, but the professor regarded this OLED-based experiment as only a first step.

“The next step would be to laminate optical or electrical sensors so that when actuator cells start to fire, we could detect the cells’ response and interactions across full networks” he said, hinting at the study of neuronal networks.

“As soon as you can have a parallel readout, you can probe one cell and see how it can be excited by other cells”.

In fact, not only this OLED-approach could be used to study neural networks, it could even be used to actively tweak such networks, playing on different ion channel activation and deactivation kinetics.

“Rather than having a probe on one neuron, with mechanically flexible micro-OLEDs, you could envisage bio-implants in real brains, wrapped around the surface of the brain”, Gather said, adding that with a closed loop and learning algorithms, you could probe parts of the brain and try to correlate all firing actions, advancing knowledge of the neuronal dysfunction that underlies devastating neurodegenerative conditions such as Alzheimer’s Disease, Parkinson’s Disease and Motor Neurone Disease. Today, such studies are performed with bulkier micro-electrodes.

Ultimately, you could dynamically control parts of the brain, say to silence the parts that are causing an epileptic seizure, Gather concluded.

Access the full paper on Science Advances at:

Related articles:

100-channel neural-recording IC achieves record low power consumption

Brain implants near milestone: FDA about to give green light

Synaptic transistor learns as it switches

Linked Articles
eeNews Europe