Metaheuristic Spatial Transformation (MST) for accurate detection of Attention Deficit Hyperactivity Disorder (ADHD)
Summary
A new Metaheuristic Spatial Transformation (MST) method improves Attention Deficit Hyperactivity Disorder (ADHD) detection using brain scans. This approach enhances classification accuracy for ADHD diagnosis from resting-state fMRI data.
Area of Science:
- Neuroscience
- Medical Imaging
- Artificial Intelligence
Background:
- Accurate diagnosis of neuropsychological disorders like ADHD using resting-state functional Magnetic Resonance Imaging (rs-fMRI) is challenging.
- Existing spatial transformation methods lack generalization for datasets with high variance and small sample sizes, such as ADHD datasets.
Purpose of the Study:
- To present a novel Metaheuristic Spatial Transformation (MST) approach for improved ADHD classification.
- To address the challenges of high dimensionality, low inter-class separability, small sample size, and high intra-class variability in rs-fMRI data.
Main Methods:
- The study introduces a Metaheuristic Spatial Transformation (MST) approach, framing spatial filter design as a constrained optimization problem solved by a hybrid genetic algorithm.
- Highly separable features derived from MST are used with a meta-cognitive radial basis function classifier for ADHD classification.
- Performance evaluation was conducted on the ADHD200 consortium dataset using ten-fold cross-validation.
Main Results:
- The MST-based classifier achieved a state-of-the-art classification accuracy of 72.10%, a 1.71% improvement over previous transformation-based methods.
- Significant increases in training and testing specificity were observed with the MST-based classifier compared to prior methods.
- MST effectively identifies highly discriminant transformations in datasets with high variability, small sample sizes, and numerous features.
Conclusions:
- The Metaheuristic Spatial Transformation (MST) approach enables reliable and accurate diagnosis of ADHD from rs-fMRI data.
- MST demonstrates efficacy in handling complex neuroimaging datasets characterized by high variability and limited sample sizes.
- The findings suggest MST is a valuable tool for the clinical diagnosis of neuropsychological disorders using rs-fMRI.
