# Sleep Spindle Based on the dominant frequency of a segment around a spindle, each is classified as a slow spindle (9-10 Hz; 10 – 12 Hz) or a fast spindle (13 – 15 Hz; 12 – 14 Hz), which are believed to occur during different phases of slow oscillations (< 1 Hz). Measuring sleep spindles and analyzing their relationship to behavior and cognition may provide insight into how these neural events influence memory performance, as well as provide diagnostic measurements for various sleep disorders. <br />The objectives of the current study include implementing the filter based and thresholding (FBT) approaches to classify spindles using a short period of high definition EEG recordings. The FBT approach is designed to classify spindles quickly with flexible parameters that capture temporal and spectral variations of the EEG representations of spindles (e.g., frequency, duration, amplitude, etc.) and serves as a classification bench mark for further investigation of machine learning models. # Goal of the project: 1. Provide an open source tool for visualizing raw EEG data and power spectral density of the EEG data. 2. Provide an open source tool for detecting spindle with particular range of frequency. 3. Provide an open source pipeline to automatically detect spindles and visualize spindles. 4. Provide preliminary descriptive statistics of detected spindle results. 5. Provide standards for manually annotating sleeping stages, spindles, and k-complex. 6. Provide interpretable features for detecting sleep spindles and k-complexes. # Next step 1. To model features that are used to annotate sleeping stages using power spectral density computed by a moving window in length of 10 seconds with 5 seconds overlapping. 2. Implementing CNN and autoencoder to extract features that contribute to the core features of the signal. <br />[Raw Data](https://osf.io/chav7/) <br />[Method](https://osf.io/fc3u5/) <br />[Results](https://osf.io/h68a3/)