Multi-modal, Multi-measure, and Multi-class Discrimination of ADHD with Hierarchical Feature Extraction and Extreme Learning Machine Using Structural and Functional Brain MRI

Abstract

Structural and functional MRI unveil many hidden properties of the human brain. We performed this multi-class classification study on selected subjects from the publically available attention deficit hyperactivity disorder ADHD-200 dataset of patients and healthy children. The dataset has three groups, namely, ADHD inattentive, ADHD combined, and typically developing. We calculated the global averaged functional connectivity maps across the whole cortex to extract anatomical atlas parcellation based features from the resting-state fMRI (rs-fMRI) data and cortical parcellation based features from the structural MRI (sMRI) data. In addition, the preprocessed image volumes from both of these modalities followed an ANOVA analysis separately using all the voxels. This study utilized the average measure from the most significant regions acquired from ANOVA as features for classification in addition to the multi-modal and multi-measure features of structural and functional MRI data. We extracted most discriminative features by hierarchical sparse feature elimination and selection algorithm. These features include cortical thickness, image intensity, volume, cortical thickness standard deviation, surface area, and ANOVA based features respectively. An extreme learning machine performed both the binary and multi-class classifications in comparison with support vector machines. This article reports prediction accuracy of both unimodal and multi-modal features from test data. We achieved 76.190% (p < 0.0001) classification accuracy in multi-class settings as well as 92.857% (p < 0.0001) classification accuracy in binary settings. In addition, we found ANOVA-based significant regions of the brain that also play a vital role in the classification of ADHD. Thus, from a clinical perspective, this multi-modal group analysis approach with multi-measure features may improve the accuracy of the ADHD differential diagnosis.

Keywords: ADHD-200; ANOVA; extreme learning machine; global functional connectivity; hierarchical feature extraction; machine learning; neuroimaging; revised recursive feature elimination.

Figure 1

A typical global functional connectivity map for a single ADHD patient. The left column shows the axial view; middle column shows the sagittal while the rightmost column shows the coronal view. The color temperature in the connectivity maps represents the strength of the connectivity measure between different resting-state networks in the cortical region.

Figure 2

Overall classification framework of the current study. The box at the extreme left presents the feature acquisition. Right, top four boxes represent the feature selection and training of the classifier. The box at the bottom on the extreme right shows the framework to acquire the testing accuracy from the data.

Figure 3

Twelve regions with significant differences as determined by structural ANOVA. The top row shows the transverse views. The bottom figures show the lateral views. The most significant results are located in the superior frontal gyrus.

Figure 4

Four significant networks including nine significant sub-network regions acquired by ANOVA analysis of the global connectivity maps of resting state fMRI data. The figure depicts the selected region in all three sagittal, coronal, and axial views respectively.