Machine learning approach in EEG-based determination of BBB state for rats
K.S. Sergeev, Saratov State University, Russia; N.I. Semenova, FEMTO-ST Institut, Université Bourgogne Franche-Comté 15B avenue des Montboucons Besançon Cedex, 25030, France; A.V. Slepnev, Saratov State University, Russia
Abstract
In this work we considered two-channel EEG data obtained from healthy rats in different conditions: awake, sleep, and after audial impact. Based on these EEG realizations the multilayer feed-forward artificial neural network were trained (methods of training dataset preparation based on raw EEG are in focus).
The using of this network to recognize the degree of similarity of EEG fragments after audial impact and EEG fragments of normal sleeping rats showed that more than half of such fragments are determined by the ANN as similar. This similarity allows us to assume that there are a number of processes (including the opening of the BBB), occurring in a similar way during normal sleep and audial impact.
Speaker
K.S. Sergeev (kssergeev@mail.ru)
Saratov State University
Russia
Discussion
Ask question
