Statistical_Modeling.md

graph LR
    GLM["GLM"]
    Mass_Univariate_Analysis["Mass Univariate Analysis"]
    GLM -- "uses" --> Maskers
    GLM -- "uses" --> Image_Processing
    GLM -- "interacts with" --> Reporting
    Mass_Univariate_Analysis -- "uses" --> Maskers
    Mass_Univariate_Analysis -- "uses" --> Image_Processing
    GLM -- "uses" --> Mass_Univariate_Analysis
    Mass_Univariate_Analysis -- "interacts with" --> Reporting
    GLM -- "uses" --> _Utils
    Mass_Univariate_Analysis -- "uses" --> _Utils

Loading

Details

The Statistical Modeling component in Nilearn is a cornerstone for neuroimaging data analysis, providing robust tools for statistical inference, primarily through the General Linear Model (GLM) and mass univariate analysis. Its design adheres to the "Scientific Computing Library" architectural pattern by offering modular, functionally grouped, and extensible statistical functionalities.

GLM

This component implements the core General Linear Model framework for analyzing fMRI data. It encompasses functionalities for constructing design matrices, fitting first-level (individual subject) and second-level (group) GLMs, defining and computing statistical contrasts, and applying appropriate statistical thresholding to results.

Related Classes/Methods:

• nilearn/glm/first_level/first_level.py
• nilearn/glm/second_level/second_level.py
• nilearn/glm/first_level/design_matrix.py
• nilearn/glm/contrasts.py
• nilearn/glm/thresholding.py

Mass Univariate Analysis

This component provides tools for performing voxel-wise statistical inference across the entire brain, addressing the multiple comparisons problem inherent in neuroimaging data. It primarily focuses on permutation-based methods, such as Permuted Ordinary Least Squares (OLS), to derive statistically significant results without strong parametric assumptions.

Related Classes/Methods:

• nilearn/mass_univariate/permuted_least_squares.py

FAQ