Korean team develops standardized tool to analyze brain signal data across different chip makers

A South Korean research team has developed a new analytical technology that allows electrical signals measured from brain neurons to be analyzed together under a single standard, even when the data comes from different equipment makers.

The Korea Institute of Toxicology said Wednesday that a research team led by Park Sun-hyun, director of the institute's Animal Alternative Research Center, has developed an open-source analysis tool called CASCADE. The team, working with Soonchunhyang University, Sungkyunkwan University and Pusan National University, designed CASCADE to process multi-electrode array, or MEA, data from different manufacturers using a single method.

Multi-electrode array chips are devices that use multiple electrodes to simultaneously measure electrical signals generated by cultured neurons or brain organoids. While widely used in neurotoxicity assessments and animal-alternative testing, the chips have long posed a challenge. Each manufacturer uses its own data format and analysis method, making it difficult to directly compare research results across platforms.

CASCADE processes data from five different manufacturers within a single analytical framework. From raw data input, it can analyze spikes and firing rates — the electrical signals generated by neurons — as well as network activity. It can even measure "criticality" and "neuronal avalanches," all with a single line of code.

Criticality refers to a balanced state in which a neural network is neither too quiet nor overly excited. In this state, information transmission and the range of responses are believed to be maximized, making criticality a useful indicator of a neural network's information-processing capacity. A neuronal avalanche refers to a phenomenon in which the activity of one neuron cascades through neighboring neurons. Researchers analyze the state of the network through the scale and duration of these cascades.

The research team verified the tool's performance using six real-world datasets from five different manufacturers. Even with different equipment and file formats, CASCADE produced criticality values under a consistent standard, and its spike-detection performance matched that of some manufacturers' proprietary software.

The tool also improved analysis speed and efficiency. A full analysis took about five to 20 seconds on a standard laptop, and even large datasets of up to 8.6 gigabytes were processed using no more than 352 megabytes of memory.

The significance of the research lies in standardizing the entire process — from raw data to neural network activity analysis to criticality classification. This makes it easier to compare results from animal-alternative neurotoxicity assessments using cultured neurons and brain organoids, while improving reproducibility and reliability, according to the researchers.

Going forward, the tool is expected to help detect early side effects that pharmaceuticals or chemical substances may have on the nervous system. It could also be used in "biotech intelligence" research aimed at quantifying the learning and information-processing capabilities of cultured neural networks.

"Scientific comparison only becomes possible when signals generated by neurons can be read by the same standard, no matter what equipment measured them," said Park, the center director. "CASCADE is our way of making that common standard freely available, and we hope it can help advance animal-alternative neurotoxicity assessment and biotech intelligence research to the next stage."