MIT camera chip improves views of neural behavior

11 Jun 2024

Individually programmable pixels maximize signal strength when studying rapid brain activity.

A project at MIT's Picower Institute for Learning and Memory has developed a sensor chip suited in particular to enhanced imaging of neural activity.

Described in Nature Communications, the chip represents "a new take on standard CMOS technology used in scientific imaging," according to the MIT team.

Recent advances in high-speed wide-field fluorescence microscopy have allowed researchers to capture biological processes with exceptional resolution. However, low signal-to-noise ratio (SNR) at high frame rates has limited researchers' ability to detect a class of particularly faint fluorescent events.

Changes to the imaging optics and computational data processing can improve things, but the basic trade-offs of speed and SNR are fundamentally linked to the image sensor, according to the MIT team, and to the standard CMOS approach of all pixels turning on and off at the same time.

Doing that ensures that faster sampling means capturing less light, so the new chip enables each pixel's timing to be controlled individually. That way, neighboring pixels can essentially complement each other to capture all the available light without sacrificing speed.

MIT designed this architecture and its "pixelwise" programmability as a way to improve visualization of neural voltage spikes that neurons use to communicate with each other, and also of the more subtle, momentary fluctuations in voltage that constantly occur between those spiking events.

"Measuring with single-spike resolution is really important as part of our research approach," said Matthew Wilson from MIT Department of Brain and Cognitive Sciences (BCS). "Thinking about the encoding processes within the brain, single spikes and the timing of those spikes is important in understanding how the brain processes information."

Real-time imaging of freely moving animals

The project's programmable exposures CMOS (PE-CMOS) sensor was designed to use six transistors per pixel, considerably simplifying the pixel-level circuitry compared to that needed for pixel control in earlier versions of the same principle.

This minimalist design allows for independent programming of pixel exposures without sacrificing photodiode area to the circuits, ensuring high sensitivity under low light conditions, according to MIT. A high-speed interface on the PE-CMOS chip then allows custom exposure and sampling patterns to be configured at the pixel level, and updated instantaneously via on-chip control.

"Since many fluorescence proteins are predominantly expressed on the cell membrane, we can encircle the bright outer contour of the cell region of interest with fast pixels having short exposure times and high sampling rates, while employing slower pixels in the dimmer regions to enhance the SNR," said MIT in its paper.

In trials on mice the new chip was able to effectively double the normal SNR, by staggering the start times of neighboring pixels and having them stay on for longer periods, detecting neural spiking activities that the conventional sensor missed. PE-CMOS could also capture both fast spiking and slower sub-threshold changes, by varying the exposure durations of neighboring pixels between 15.4 and 1.9 milliseconds.

MIT's aim is to move towards brain-wide real-time measurements of activity in distinct types of neurons in animals, even as they are freely moving and learning how to navigate mazes. The development of genetically encoded voltage indicators, labels that make cells glow as their voltage changes in real-time, should be a good clinical match with the capabilities of the PE-CMOS chip.

"We are already working on the next iteration of chips with lower noise, higher pixel counts, time-resolution of multiple kHz, and small form factors for imaging in freely behaving animals," commented Matt Wilson.