Transforming 3D MRI to 2D Feature Maps Using Pre-Trained Models for Diagnosis of Attention Deficit Hyperactivity Disorder

According to the World Health Organization (WHO), approximately 5% of children and 2.5% of adults suffer from attention deficit hyperactivity disorder (ADHD). This disorder can have significant negative consequences on people’s lives, particularly children. In recent years, methods based on artificial intelligence and neuroimaging techniques, such as MRI, have made significant progress, paving the way for more reliable diagnostic methods than traditional clinical approaches. In this paper, we present our machine learning based analytical approach, more precisely, a novel feature extraction technique, to more accurately and reliably diagnose ADHD using clinically available neuroimaging data. Utilizing the ADHD200 dataset, which encompasses demographic factors and anatomical MRI scans in a diverse ADHD population, our study focuses on leveraging modern machine learning methods. The preprocessing stage employs a pre-trained Visual Geometry Group16 (VGG16) network to extract Two-Dimensional (2D) feature maps from anatomical 3D MRI data, thereby reducing computational complexity and enhancing efficiency. The inclusion of personal attributes, such as age, gender, intelligence quotient, and handedness, strengthens the diagnostic models. Four deep-learning architectures, Convolutional Neural Network 2D (CNN2D), CNN1D, Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU), were explored for the analysis of the MRI data, with and without inclusion of clinical characteristics. A 10-fold cross-validation revealed that the LSTM model, which incorporated both MRI data and personal attributes, had the best performance, achieving higher accuracy (0.8637) and area under the curve (AUC:0.9025). Our findings demonstrate that the proposed approach, which extracts 2D features from 3D MRI images and integrates these features with clinical characteristics, may be useful in diagnosis of ADHD with high accuracy.