Skip to content

Node Reference

NeuroPype ships with a vast collection of over 800 data-processing nodes, organized into the packages listed below. Each package page lists the nodes it contains, along with a one-line description and a link to the full documentation for each node.

This same per-node documentation is also visible in Pipeline Designer when a node is selected, and documentation for each node parameter appears as a tooltip when hovering over the parameter in the node's properties window.

Signal processing

Signal processing packages contain nodes for neural and bio signal processing, implementing domain-specific algorithms for acquiring, processing, and analyzing physiological signals across a range of modalities:

  • Connectivity — nodes which estimate information-flow measures between multiple signal channels.
  • Feature Extraction — nodes which extract a variety of features from signal data.
  • Signal Processing — nodes which perform signal-specific functions and algorithms.
  • Source Localization — EEG source-space processing nodes to solve the problem of estimating source activity given EEG/MEG observations and operating on the resulting solutions.
  • Spectral — estimation and transform nodes operating on the data frequency spectrum (Loreta, Beamforming, etc.).
  • Neural — nodes which include a variety of BCI-related algorithms (regularized common spatial filters, etc.)
  • Cardiac — nodes which process ECG/PPG data and compute cardiac measures (HRV, HR, Respiration, etc.)
  • Eye Tracking — nodes which process gaze data and compute fixation and saccade events for multi-modal processing.
  • NIRS — nodes implementing domain-specific methods for processing near-infrared spectroscopy (fNIRS) data.
  • Diagnostics — nodes for computing signal quality and sanity-checking data.

AI training and inference

AI training and inference packages support both classical and modern machine learning workflows, including model fitting, evaluation, and probabilistic modeling:

  • Machine Learning — nodes implementing supervised machine learning algorithms on signal data.
  • Deep Learning — nodes for building, training, and running inference with neural networks within processing workflows.
  • Bayesian — nodes for specifying statistical models and performing Bayesian inference to quantify and propagate uncertainty through analyses.
  • Optimization — nodes for gradient computation, loss functions, penalties, constraints, and solving optimization problems.
  • Distributions — nodes for creating probability distributions and computing their quantities such as mean, variance, log-probability, and sampling.
  • Random — nodes for drawing samples from random distributions and managing random seeds and chainable keys.

Mathematics and statistics

Mathematics and statistics packages offer general-purpose numerical operations on arrays and tensors, as well as standard statistical estimators:

  • Array — nodes providing a full suite of matrix operations on raw arrays.
  • Elementwise Math — nodes with standard math operators and expressions operating on data packets on an element-by-element basis.
  • Tensor Math — nodes which implement mathematical operations that can be applied to any kind of data tensor.
  • Statistics — nodes which implement a range of standard and advanced statistical estimators, including ancova, mancova, etc.

Programming and workflow

Programming and workflow packages provide control flow, data formatting, and pipeline construction tools for building complex processing workflows:

  • Programming — nodes that implement standard programming control flow and data structure handling.
  • Control Flow — nodes for implementing non-linear control flow such as looping, conditional processing, and functions within pipelines.
  • Formatting — nodes which change the format or shape of data blocks without altering numeric values, including segmentation and streaming playback.
  • Markers — nodes for adding and manipulating event markers used in event-related segmentation and processing.
  • Utility — nodes providing a variety of useful signal processing functions such as dejittering timestamps, fixing chunk size irregularities, etc.
  • Special Purpose — nodes supporting specialized workflows that do not yet have their own package.

Reporting, presentation and I/O

Reporting, presentation and I/O packages handle generating visual data analysis reports, data acquisition from hardware and networks, import/export from numerous formats, and realtime visualization:

  • Reporting — nodes for generating data analysis reports from computed results, including figure generation and template-based report synthesis.
  • Network — nodes for receiving input and sending output over various protocols, including LSL, TCP, OSC.
  • File I/O — nodes for file-related functions (import, export, record, etc.)
  • Device I/O — nodes for transmitting data via vendor-specific hardware interfaces.
  • Visualization — nodes for generating real-time visualization of signal data (head model, connectivity, time series, etc.)