TY - JOUR
T1 - Using Constrained Square-Root Cubature Kalman Filter for Quantifying the Severity of Epileptic Activities in Mice
AU - Huang, Chih Hsu
AU - Wang, Peng Hsiang
AU - Ju, Ming Shaung
AU - Lin, Chou Ching K.
N1 - Funding Information:
Funding: This work was partly funded by a grant (MOST 110-2314-B-006 -080) from the Ministry of Science and Technology, Taiwan. The funding source was not involved in conceptualization, study design,datacollection,analysis orinterpretation, manuscript writingor journal submission.
Publisher Copyright:
© 2022 by the authors. Licensee MDPI, Basel, Switzerland.
PY - 2022/7
Y1 - 2022/7
N2 - (1) Background: Quantification of severity of epileptic activities, especially during electrical stimulation, is an unmet need for seizure control and evaluation of therapeutic efficacy. In this study, a parameter ratio derived from constrained square-root cubature Kalman filter (CSCKF) was formu-lated to quantify the excitability of local neural network and compared with three commonly used indicators, namely, band power, Teager energy operator, and sample entropy, to objectively determine their effectiveness in quantifying the severity of epileptiform discharges in mice. (2) Methods: A set of one normal and four types of epileptic EEGs was generated by a mathematical model. EEG data of epileptiform discharges during two types of electrical stimulation were recorded in 20 mice. Then, EEG segments of 5 s in length before, during and after the real and sham stimulation were collected. Both simulated and experimental data were used to compare the consistency and differences among the performance indicators. (3) Results: For the experimental data, the results of the four indicators were inconsistent during both types of electrical stimulation, although there was a trend that seizure severity changed with the indicators. For the simulated data, when the simulated EEG segments were used, the results of all four indicators were similar; however, this trend did not match the trend of excitability of the model network. In the model output which retained the DC component, except for the CSCKF parameter ratio, the results of the other three indicators were almost identical to those using the simulated EEG. For CSCKF, the parameter ratio faithfully reflected the excitability of the neural network. (4) Conclusion: For common EEG, CSCKF did not outperform other commonly used performance indicators. However, for EEG with a preserved DC component, CSCKF had the potential to quantify the excitability of the neural network and the associated severity of epileptiform discharges.
AB - (1) Background: Quantification of severity of epileptic activities, especially during electrical stimulation, is an unmet need for seizure control and evaluation of therapeutic efficacy. In this study, a parameter ratio derived from constrained square-root cubature Kalman filter (CSCKF) was formu-lated to quantify the excitability of local neural network and compared with three commonly used indicators, namely, band power, Teager energy operator, and sample entropy, to objectively determine their effectiveness in quantifying the severity of epileptiform discharges in mice. (2) Methods: A set of one normal and four types of epileptic EEGs was generated by a mathematical model. EEG data of epileptiform discharges during two types of electrical stimulation were recorded in 20 mice. Then, EEG segments of 5 s in length before, during and after the real and sham stimulation were collected. Both simulated and experimental data were used to compare the consistency and differences among the performance indicators. (3) Results: For the experimental data, the results of the four indicators were inconsistent during both types of electrical stimulation, although there was a trend that seizure severity changed with the indicators. For the simulated data, when the simulated EEG segments were used, the results of all four indicators were similar; however, this trend did not match the trend of excitability of the model network. In the model output which retained the DC component, except for the CSCKF parameter ratio, the results of the other three indicators were almost identical to those using the simulated EEG. For CSCKF, the parameter ratio faithfully reflected the excitability of the neural network. (4) Conclusion: For common EEG, CSCKF did not outperform other commonly used performance indicators. However, for EEG with a preserved DC component, CSCKF had the potential to quantify the excitability of the neural network and the associated severity of epileptiform discharges.
UR - http://www.scopus.com/inward/record.url?scp=85133657027&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85133657027&partnerID=8YFLogxK
U2 - 10.3390/biomedicines10071588
DO - 10.3390/biomedicines10071588
M3 - Article
AN - SCOPUS:85133657027
SN - 2227-9059
VL - 10
JO - Biomedicines
JF - Biomedicines
IS - 7
M1 - 1588
ER -