Researchers Utilize AI to Uncover Brain Abnormalities in ADHD Diagnoses among Adolescents
Scientists have successfully employed artificial intelligence (AI) to analyze specialized brain MRI scans of adolescents with and without attention-deficit/hyperactivity disorder (ADHD). The study, which was presented at the annual meeting of the Radiological Society of North America (RSNA), revealed significant abnormalities in nine brain white matter pathways in teenagers diagnosed with ADHD. This groundbreaking research could pave the way for more objective diagnostic tools for the prevalent disorder, which affects an estimated 5.7 million children and teenagers in the United States alone.
ADHD is a widespread disease that typically manifests in childhood and continues into adulthood. Its impact on an individual’s quality of life and ability to function within society can be immense. In recent years, the disorder has become increasingly prevalent among today’s youth due to the widespread availability of distracting devices such as smartphones.
According to Justin Huynh, a co-author of the study and a research specialist at the University of California, San Francisco, ADHD is notoriously difficult to diagnose, relying primarily on subjective self-reported surveys. Huynh emphasizes the urgent need for more objective metrics in ADHD diagnosis and highlights the gap that this research aims to address.
This study marks the first time that deep learning, a type of AI, has been utilized to identify ADHD markers within the multi-institutional Adolescent Brain Cognitive Development (ABCD) Study. The ABCD Study comprises brain imaging data, clinical surveys, and other pertinent information from over 11,000 adolescents across 21 research sites in the United States. The researchers focused on diffusion-weighted imaging (DWI), a specific type of MRI, to extract fractional anisotropy (FA) measurements from 30 primary white matter tracts in the brain. FA measures the flow of water molecules along these white matter tract fibers.
An artificial intelligence model was trained using the FA values of 1,371 individuals, following which it was evaluated on a separate group of 333 patients, 193 of whom were diagnosed with ADHD and 140 who were not. ADHD diagnoses were confirmed using the Brief Problem Monitor evaluation, a renowned rating instrument that tracks a child’s functioning and response to interventions.
The researchers discovered significantly higher FA values in nine white matter tracts among patients with ADHD using their AI model. These findings offer unprecedented insights into the detailed MRI signatures associated with ADHD. Remarkably, the abnormalities observed in these white matter tracts align with the known symptoms of the disorder, further validating the AI’s accuracy.
Moving forward, the researchers intend to gather additional data from the remaining participants within the ABCD dataset to ascertain the effectiveness of alternative AI models. The ultimate goal is to develop imaging biomarkers that can quantitatively and objectively diagnose ADHD, a step that could revolutionize the diagnostic framework for this complex disorder.
Pierre F. Nedelec, Samuel Lashof-Regas, Michael Romano, Leo P. Sugrue, and Andreas M. Rauschecker also contributed to this groundbreaking study.
The implications of this research are significant. While traditional methods of diagnosing ADHD rely primarily on subjective measures, the utilization of AI and advanced imaging techniques could provide more objective and accurate diagnostic tools. This could potentially lead to earlier interventions and enhanced management of the disorder, significantly improving the lives of millions of individuals affected by ADHD.
