In a groundbreaking study presented at the annual meeting of the Radiological Society of North America (RSNA), researchers utilized artificial intelligence (AI) to analyze specialized brain MRI scans of adolescents, aiming to identify objective markers of attention-deficit/hyperactivity disorder (ADHD).
Prevalence of ADHD
According to the Centers for Disease Control and Prevention, ADHD is a common disorder often diagnosed in childhood, persisting into adulthood. In the U.S., an estimated 5.7 million children and adolescents between the ages of 6 and 17 have been diagnosed with ADHD. The increasing prevalence of ADHD in today’s youth, attributed to factors such as smartphones and distracting devices, underscores the importance of effective diagnosis and intervention.
Challenges in ADHD Diagnosis
Application of Deep Learning AI
This study represents a pioneering effort in applying deep learning, a form of artificial intelligence, to identify ADHD markers. The researchers leveraged the expansive ABCD dataset, encompassing brain imaging, clinical surveys, and other data from over 11,000 adolescents across 21 research sites in the U.S. The study focused on a specialized MRI technique called diffusion-weighted imaging (DWI).
The researchers in this study used AI to analyze a type of brain imaging data known as Fractional Anisotropy (FA) values. FA is a measure derived from Diffusion Tensor Imaging (DTI), a type of Magnetic Resonance Imaging (MRI) that allows researchers to visualize and quantify the diffusion of water molecules in the brain’s white matter tracts. White matter tracts are bundles of myelinated nerve fibers that connect different regions of the brain and allow for communication between them.
In the brain, water molecules tend to diffuse along the direction of these fibers, and this diffusion can be measured and quantified using DTI. The FA value is a measure of the directionality of this diffusion. A higher FA value indicates more directional diffusion, which is typically associated with healthier and more intact white matter tracts.
Methodology
The research team selected 1,704 individuals from the ABCD dataset, including both adolescents with and without ADHD. Using DWI scans, fractional anisotropy (FA) measurements were extracted along 30 major white matter tracts in the brain. FA is a crucial measure indicating how water molecules move along the fibers of white matter tracts.
Deep Learning AI Model
The FA values from 1,371 individuals were utilized to train a deep-learning AI model, which was subsequently tested on 333 patients, including 193 diagnosed with ADHD and 140 without. ADHD diagnoses were established using the Brief Problem Monitor assessment, a rating tool monitoring a child’s functioning and responses to interventions.
Key Findings
Through the implementation of AI, the researchers identified significant elevations in FA values in nine white matter tracts in patients diagnosed with ADHD. These findings represent an unprecedented level of detail in the MRI signatures associated with ADHD, aligning with the symptoms of the disorder.
Future Directions
Huynh emphasizes the significance of this discovery and the need for further exploration. The research team plans to extend their analysis to the entire ABCD dataset, comparing the performance of additional AI models. The ultimate goal is to establish imaging biomarkers that can be employed in a quantitative, objective diagnostic framework for ADHD.
Conclusion
The integration of artificial intelligence in the analysis of specialized brain MRI scans has unveiled distinct differences in white matter tracts associated with ADHD. This breakthrough provides a promising avenue for developing objective diagnostic tools, addressing the current challenges posed by the subjective nature of existing diagnostic tests. As researchers continue to delve into the vast dataset and refine AI models, the potential for revolutionizing ADHD diagnosis becomes increasingly tangible, offering hope for improved early detection and intervention.
reference link : https://press.rsna.org/timssnet/media/pressreleases/14_pr_target.cfm?ID=2470
