Automatic Sleep Spindle Detection and Genetic Influence Estimation Using Continuous Wavelet Transform

Abstract

Mounting evidence for the role of sleep spindles in neuroplasticity has led to an increased interest in these non-rapid eye movement (NREM) sleep oscillations. It has been hypothesized that fast and slow spindles might play a different role in memory processing. Here, we present a new sleep spindle detection algorithm utilizing a continuous wavelet transform (CWT) and individual adjustment of slow and fast spindle frequency ranges. Eighteen nap recordings of ten subjects were used for algorithm validation. Our method was compared with both a human scorer and a commercially available SIESTA spindle detector. For the validation set, mean agreement between our detector and human scorer measured during sleep stage 2 using kappa coefficient was 0.45, whereas mean agreement between our detector and SIESTA algorithm was 0.62. Our algorithm was also applied to sleep-related memory consolidation data previously analyzed with a SIESTA detector and confirmed previous findings of significant correlation between spindle density and declarative memory consolidation. We then applied our method to a study in monozygotic (MZ) and dizygotic (DZ) twins, examining the genetic component of slow and fast sleep spindle parameters. Our analysis revealed strong genetic influence on variance of all slow spindle parameters, weaker genetic effect on fast spindles, and no effects on fast spindle density and number during stage 2 sleep.

Keywords: EEG; automatic detection; heritability; sleep spindle; twins.

Figure 1

Algorithm detection scheme.

Figure 2

Complex Morlet wavelet with central frequency f₀ = 2 used in the analysis. Presented wavelet corresponds to 14 Hz frequency.

Figure 3

The adjustment scheme of individual spindle frequency range. (A) The outcome of spindle activity scan which resulted in two vectors of spindle activity over frequency range separately for frontal channel F3A2 (vecslow: green color) and parietal channel P3A2 (vecfast: blue color). (B) In both activity vectors the value in 9 Hz was set to zero, vectors were smoothed and 50% of mean spindle activity (dashed black line) was added to both of them. (C) Vector (vec_rel) showing a relation of spindle activity between frontal EEG and parietal EEG, computed according to “Spindle Activity Comparison” Section. (D) Smoothed vec_rel. First, algorithm localized minimum and maximum (black dots). Localized minimum in vec_rel was set as slow spindle central frequency (green square). Localized maximum in vec_rel was a starting point to estimate fast spindle frequency ranges using vec_fast. Local maximum in vec_fast was set as fast spindle central frequency (blue square). Ranges of fast (dashed blue lines) and slow (dashed green lines) spindle frequency were estimated according to “Spindle Activity Comparison” Section. First frequency bin below slow spindle range in which spindle activity was higher in the parietal channel was set as frequency in which slow spindles are unlikely (stop_detect: red dashed line).

Figure 4

Distribution of detected sleep spindles in 0.1 Hz frequency bins in monozygotic (MZ) twin pair number 10. Analysis was performed separately for stage 2 and slow wave sleep (SWS). Each row of plots represents one recording night. Column Activity Scan shows the result of pre-analysis performed to localize slow and fast spindle frequency ranges. During activity scan spindles were detected in two EEG derivations: parietal channel P3A2 (blue color) and frontal channel F3A2 (green color). Information from activity scan was used to set frequency range of fast spindles (light blue color), slow spindles (light green color) and range in which spindles should not be detected anymore (light red color). Localized frequency ranges were used to detect sleep spindles in four EEG derivations, which are presented in distinct columns: FP1A2, F3A2, C3A2 and P3A2. Blue color depicts sleep spindles detected with wavelets in fast spindle frequency range, green color depicts sleep spindles detected with wavelets in slow spindle frequency range whereas orange color depicts sleep spindles detected with combined slow and fast spindle frequency ranges.

Figure 5

The scheme of spindle detection. (A) EEG signal from C3A2 derivation during SWS in twin 10a during night 2 (localization of spindle frequency ranges and overall results of spindle detection in twin 10a are presented in Figure 4). (B) The result of continuous wavelet transform (CWT) in time and frequency domain. Red color depicts WT result using wavelet corresponding to 9 Hz frequency. Events with this frequency are not classified as spindles. Green color depicts WT using wavelets corresponding to 10.4–12 Hz frequency range. Events detected in this frequency range are classified as slow spindles (light green color). Blue color depicts WT using wavelets corresponding to 12.5–13.5 Hz frequency range. Events detected in this frequency range are classified as fast spindles (light blue color).

Figure 6

Sensitivity-precision plot showing how these two measures depend on spindle detection thresholds. Sleep spindles were scored in C3A2 EEG channel. (A) ROC-like plot of sensitivity vs. precision, (B) the sum of sensitivity and precision according to detection thresholds variety. We had two detection thresholds in our algorithm: spindle peak threshold (SP) set as 80 times basic threshold (BT) and spindle activity threshold (SA) set as 55 times BT (calculation of BT is described in “Threshold Setup” Section). We illustrate how performance changes according to SP, where y axis shows multiplication rate of BT used to obtain SP, but for each iteration values of both thresholds were changed together to always keep the same ratio between them (SP = 1.45 × SA). Black circles connected with black line mark sensitivity and precision obtained for thresholds chosen for our algorithm.

Figure 7

Validation set of 18 nap EEG recordings. Agreement in sleep spindle detection during stage 2 in C3A2 EEG derivation between our algorithm, human visual scorer and SIESTA automatic spindle detector. On y axis there are presented: (A) total number of detected sleep spindles for each recording, (B) spindle density for each recording, (C) the kappa coefficient of scorers agreement for each recording. Subject id and number of nap recording are presented on x axis.

Figure 8

Venn diagram showing in numbers of detected spindles, how spindles detected by each scorer overlapped with spindles detected by other scorers. Sleep spindles were detected during stage 2 in C3A2 EEG derivation.

Figure 9

Relation between declarative memory performance and spindle density computed by two algorithms: SIESTA spindle detector and CWT detector during (A) stage 2 sleep and (B) SWS. Sleep spindles were detected in C4A1 EEG derivation.