NeuroKit2 Documentation

The Texas Resting-state EEG dataset contains EEG recordings from 22 subjects. Each subject has 72 EEG channels recorded during an 8-minute resting task. The task consists of interleaved 1-minute intervals of 'eyes open' (4 mins total) and 'eyes closed' (4 mins total) conditions. Participants were instructed to remain relaxed, alert, and awake throughout the recording.

Dataset Specifications:

• Subjects: 22
• Channels: 72 EEG channels
• Task Duration: 8 minutes per subject (4 mins eyes open, 4 mins eyes closed)
• Task Structure: 1-minute interleaved intervals
• Note: Due to a technical error, one participant has only 4 minutes of recording time.

The neurokit2.microstates module provides a suite of functions for analyzing M/EEG microstates. The workflow typically involves segmenting the signal, classifying microstates, and computing dynamic properties.

Key functions include:

• microstates_segment(): Segments the signal into microstate components.
• microstates_classify(): Classifies the segments into specific microstate types.
• microstates_plot(): Visualizes the microstate analysis results.
• microstates_clean(): Cleans the microstate data.
• microstates_dynamic(): Computes dynamic properties of the microstates.
• microstates_findnumber(): Determines the optimal number of microstates.
• microstates_peaks(): Identifies microstate peaks.
• microstates_static(): Computes static microstate properties.

NeuroKit2 follows a consistent naming convention based on the signal type and the desired goal. This allows users to switch between different signal types easily while using the same functional logic.

• Consistency: Functions follow the pattern signaltype_functiongoal(). For example, to clean an ECG signal, you use ecg_clean(). Common goals include *_clean(), *_findpeaks(), *_process(), and *_plot().
• Accessibility: High-level "master" functions like *_process() handle the entire pipeline (cleaning, preprocessing, and processing) using sensible defaults.
• Flexibility: Mid-level functions like *_clean() or *_rate() allow advanced users to build custom pipelines with granular control over parameters.

Control flow allows code to execute differently based on conditions.

Crucial Rule: Indentation Python uses indentation (usually 4 spaces or one TAB) to define code blocks. A colon : marks the start of a block (like an if statement), and the subsequent lines must be indented. Incorrect indentation will cause an error.

If/Else Syntax:

x = 5
if x < 3:
    print("lower")
else:
    print("higher")

Functions are reusable blocks of code that typically take an input and return a transformed output. In Python, you call a function by its name followed by parentheses. Common built-in functions include str() (converts to string), int() (converts to integer), and print() (outputs to console). The range(n) function is frequently used in loops to create a sequence of integers from 0 to n-1.

x = 3
x = str(x)
print(x)

# Using range in a loop
for i in range(3):
    print(i)

Dictionary Indexing

Access elements in a dictionary using their key inside square brackets:

mydict = {"A": 1, "B": 2, "C": 3}
x = mydict["B"]
print(x)  # Output: 2

List Indexing (Zero-based)

In Python, indexing starts at 0. The first element is at index 0, the second element is at index 1, and so on.

mylist = [1, 2, 3]
x = mylist[1]  # Accesses the 2nd element
print(x)       # Output: 2

Containers allow you to store multiple variables. The two primary types are:

• Lists: Accessed by order (index). Created with square brackets [].
• Dictionaries: Accessed by a key (name). Created with curly brackets {}.

mylist = [1, 2, 3]
mydict = {"A": 1, "B": 2, "C": 3}

Physiological data analysis in NeuroKit2 typically falls into two categories:

1. Event-related analysis: Refers to physiological changes occurring immediately in response to a specific event (e.g., stimulus presentation). This is epoch-based, where short chunks of signal (epochs) are time-segmented and locked to a stimulus. Use bio_analyze() to compute rate changes, peak characteristics, and phase characteristics.

2. Interval-related analysis: Refers to physiological characteristics occurring over longer periods (seconds to days) without specific time-locked events (e.g., resting state, watching a movie). Use bio_analyze() to compute rate characteristics (like variability metrics) and peak characteristics over the entire duration.

You can create a boolean mask by applying a condition to a NumPy array. This mask can then be used to subset the array or modify specific elements that meet the condition.

import numpy as np

myvector = np.array([1, 2, 3, 2, 1])

# Create a boolean mask
mask = myvector <= 2

# Subset the array using the mask
subset = myvector[mask]

# Modify elements in-place using a condition
myvector[myvector <= 2] = 6

While Python lists can hold multiple types, NumPy arrays (often called vectors when one-dimensional) are optimized containers that hold a single data type. A key advantage of arrays is vectorization: mathematical operations applied to an array are automatically propagated to every element, avoiding the need for manual loops.

import numpy as np

# Convert a list to a vector
mylist = [1, 2, 3]
myvector = np.array(mylist)

# Vectorized operation: adds 1 to every element
myvector = myvector + 1

Variables are names used to refer to values or entities. You assign a value to a variable using the = sign.

x = 2
y = 3
anothervariable = x * y
print(anothervariable)