100099: Compressive Spike Sorting for Neural Implants


Many companies sell neurophysiology data acquisition systems that are bulky, wired, and very expensive, requiring large computational power via PCs to sort neural spikes. In addition, the spike-sorting task is a fundamental step in the design of clinically viable brain machine interfaces for restoring normal sensory and motor function of disabled people. Many of these systems require high electrode channel count and large number of cells to operate efficiently and reliably.



Michigan State University’s technology is an algorithm for sorting neural spikes fired by multiple nerve cells and simultaneously recording them using a single or an arrow of microelectrodes implanted adjacent to the cell population. The algorithm is hardware optimized to be readily implemented in ultra-low power and miniaturized electronic circuits that are feasible to include in a fully implanted system.


The technology classifies multiple spike waveforms (spike sorting) to permit extracting spike trains of individual neurons from the recorded mixture of signals and reduces the ultra-high communication bandwidth needed to transmit the recorded raw data to an external PC. It also extracts information reliably from single cell activity to characterize brain function in daily life and in subjects suffering from many neurological diseases and disorders, such as Parkinson’s disease and epilepsy.



·         Decompression not necessary: Sorts the spikes within the compressed representation of the raw data without the need to decompress and sort in the traditional sense.

·         Suited for large-scale interfacing: Well suited for large-scale interfacing with the nervous system where hundreds (or potentially thousands) of cells would need to be monitored simultaneously to infer useful information about neural activity.

·         Real-time operation: Operates in real time, making it suitable for closed loop brain machine interface system design in clinical applications.

·         Preserves biological information: Preserves all the biological information in the recorded data.

·         Adaptive and flexible: Adaptive to changes in spike waveform shapes typically observed during long-term experiments/clinical use.

·         Suitable for implantable systems: The technology is hardware friendly, making it suitable for large-scale integration in implantable microsystems.



·         Neurotechnology

·         Assistive technology for people with severe paralysis or spinal cord injury or prosthetic limbs

·         Brain machine interfaces for restoring normal sensory and motor function of disabled people


IP Protection Status

Patent pending

Patent Information:


For Information, Contact:

Raymond DeVito
Technology Manager
Michigan State University - Test