Biological parametric mapping: A statistical toolbox for multimodality brain image analysis

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Neuroscience

Background:

• Emergence of multiple brain Magnetic Resonance (MR) imaging modalities.
• Current analysis methodologies are often modality-specific.
• Limitations in comparing across modalities using region-of-interest analyses hinder sophisticated neuroscience questions.

Purpose of the Study:

• To develop a toolbox for multimodal brain image analysis.
• To enable voxel-by-voxel comparisons across different imaging modalities.
• To facilitate the investigation of complex neuroscience hypotheses.

Main Methods:

• Development of the biological parametric mapping (BPM) toolbox in Matlab.
• Utilizes a voxel-wise application of the general linear model.
• Integrates information from multiple modalities as regressors.
• Incorporates statistical inference for correlation fields for enhanced accuracy.
• User-friendly interface for analyses including correlation, ANCOVA, and multiple regression.
• High integration with Statistical Parametric Mapping (SPM) software for visualization and inference.

Main Results:

• Demonstrated the potential of BPM using in vivo data.
• Enabled sophisticated voxel-wise multimodal correlation analyses.
• Provided a framework for advanced neuroimaging research integrating diverse data types.

Conclusions:

• Biological Parametric Mapping (BPM) offers a powerful solution for multimodal brain image analysis.
• The toolbox overcomes limitations of modality-specific and region-of-interest approaches.
• BPM facilitates more advanced and accurate neuroscience investigations by enabling voxel-wise cross-modal comparisons.