independent component analysis
Mathematical technique for separating mixed neural signals into statistically independent components, commonly used for artifact removal and source separation in neuroimaging.
Full Definition
Independent Component Analysis (ICA) is a computational technique that separates multivariate signals into additive, statistically independent components. In neuroimaging, ICA is widely used to decompose complex neural data into meaningful components representing distinct neural processes, artifacts, or noise sources. The method assumes that observed signals are linear mixtures of independent source signals and uses higher-order statistics to achieve separation. ICA is particularly valuable for removing artifacts (eye movements, cardiac signals, muscle activity) from EEG/MEG data and for identifying functional networks in fMRI data. The technique requires careful interpretation as components may represent noise, artifacts, or genuine neural activity.
Usage
Usage note: Commonly abbreviated as ICA; distinguish from PCA (Principal Component Analysis).
In Context
- "Independent component analysis successfully isolated eye blink artifacts from the EEG data." — Signal processing methodology
- "ICA decomposition revealed 20 independent components, including the default mode network." — Resting-state fMRI analysis