System for accelerated study of epileptic seizures developed at TSU in collaboration with IHD and NF RAS

The development analyzes electrocorticogram, recording phases of deep sleep and epileptic seizures with an accuracy of up to 88%

Specialists from the Siberian Artificial Intelligence Center of Tomsk State University (TSU), together with the Institute of Higher Nervous Activity and Neurophysiology (IHD and NF) of the Russian Academy of Sciences, have created a system for the automatic processing of brain activity data in laboratory rats. The development will accelerate research into absence epilepsy, a form of the disease characterized by brief loss of consciousness.

Experimental rat
Experimental rat

The system analyzes the electrocorticogram (ECoG), recording phases of deep sleep and epileptic seizures with an accuracy of up to 88%. Previously, the decryption of such data was performed manually and took a significant amount of time. The algorithm uses signals from only one electrode installed in the frontal lobe of the brain, which simplifies the process.

The use of automatic algorithms greatly facilitates the analysis of such records, reducing time costs by tens of times with minimal loss of accuracy. The development and implementation of such methods for solving time-consuming tasks that require large expenditures of time and effort of specialists is extremely in demand in scientific activities.
Ivan Lazarenko, employee of the behavioral neurobiology group of IHD and NF RAS

The development became a continuation of cooperation after the team of the center won the "Digital Breakthrough. Season: Artificial Intelligence" hackathon. The system is based on the CatBoost algorithm in combination with wavelet transform, which provides high processing speed - up to 25 seconds per hour of recording.

We trained the model on a data library marked up by experts from IHD and NF RAS. We made an application, now you can upload ECoG data into it. The system performs the entire data processing process without the participation of an expert with a minimum of errors.
Alexander Kovalev, one of the developers of the automated system, an employee of the Siberian AI Center of TSU

According to WHO, about 50 million people worldwide suffer from epilepsy. Studies on WAG/Rij line rats help to study the mechanisms of disease development. The results of the work are planned to be published in a scientific journal.

Read more materials on the topic:

Automated temperature control system in urban heating networks developed at Perm Polytechnic

Software and hardware complex for production automation Factory director developed at MIPT

A new method accelerating the training of distributed neural network models has been developed at MIPT with the participation of Sberbank

Now on home