MVPA-Light

MVPA-Light implements multivariate pattern analysis methods for classification and regression of neuroimaging data to investigate brain function using MEG and fMRI.

Key Features:

• Multidimensional Data Analysis: Provides functions for handling multi-dimensional datasets, including generalization (time x time) and searchlight analyses for neuroimaging data.
• Cross-validation and Hyperparameter Tuning: Includes cross-validation and hyperparameter tuning routines for model evaluation and selection.
• Nested Preprocessing: Supports nested preprocessing integrated into the modeling pipeline.
• Statistical Significance Testing: Computes classification and regression metrics and assesses their statistical significance.
• Modularity and Extensibility: Offers a modular architecture that enables addition of new methods or custom analyses.
• Integration Capabilities: Interfaces with LIBSVM and LIBLINEAR and integrates with the FieldTrip neuroimaging toolbox.

Scientific Applications:

• Magnetoencephalography (MEG): Enables classification and regression analyses on MEG datasets.
• Functional Magnetic Resonance Imaging (fMRI): Supports multivariate and searchlight analyses on fMRI datasets.
• Cognitive Neuroscience and Neuroimaging Studies: Facilitates investigations of brain function and structure using classification and regression approaches.

Methodology:

Implements native classifiers and regression models with modern optimization algorithms; provides cross-validation, hyperparameter tuning, nested preprocessing, computation of classification and regression metrics with statistical significance assessment, generalization (time x time) and searchlight analyses; interfaces with LIBSVM and LIBLINEAR and integrates with FieldTrip.