Skip to main content

Table 4 Feature selection using mRMR method

From: Wavelet-based Gaussian-mixture hidden Markov model for the detection of multistage seizure dynamics: A proof-of-concept study

Order

V F /W c

Frequency Range (Hz)

1

0.497

8 - 15

2

0.436

15 - 40

3

0.400

100 - 250

4

0.331

4 - 8

5

0.308

40 - 100

6

0.308

250 - 400

7

0.266

<4

  1. The frequency bands associated with the two highest mutual information quotients (VF/Wc) were used to create the reduced feature space. In our analysis of chronic seizures versus non-seizure activities, the alpha (8 - 15 Hz) and beta (15 - 40 Hz) activities were found to be the most relevant for seizure detection.
Back to article page

BioMedical Engineering OnLine

ISSN: 1475-925X

Contact us