BrainFD

BrainFD computes fractal-dimension measurements of intracranial brain volume from T1-weighted MRI to quantify structural complexity and detect volume changes associated with neurodegeneration.

Key Features:

• Fractal Dimension Calculation: Calculates fractal dimension (FD) as a metric of brain structural complexity for precise intracranial volume measurement.
• Box-Counting Algorithm: Implements the box-counting algorithm to compute FD with implementations in Python and JAVA.
• Cross-Subject and Cross-Session Comparisons: Enables comparison of FD across subjects and multiple MRI sessions for longitudinal and cross-sectional analyses.
• Clinical Dementia Rating (CDR) Integration: Creates imaging profiles based on Clinical Dementia Rating (CDR) scores to relate FD measures to cognitive status.
• Heterogeneous Data Model Training: Supports training of models on heterogeneous multi-session data to evaluate FD variations for each participant.

Scientific Applications:

• Classification of Cognitive Impairment: Distinguishes CDR-based classes using FD measures, reported as 100% accuracy in the presented study.
• Longitudinal Monitoring: Monitors intracranial brain volume changes across MRI sessions to assess progression of atrophy.
• Imaging Biomarker Development: Provides a potential low-cost personalized imaging biomarker for intracranial brain volume loss associated with neurodegeneration.
• Detection of Subtle Structural Variations: Detects subtle variations in imaging data relevant to structural brain changes for clinical evaluation.

Methodology:

FD was computed using the box-counting algorithm (Python and JAVA) on T1-weighted MRI images from the OASIS-3 database; analyses included a cross-sectional experiment categorizing data into two CDR classes and a heterogeneous data model training experiment using multi-session scans per participant to evaluate FD variations.