Researchers at the University of Louisville have developed an artificial intelligence (AI) system that demonstrates the potential to diagnose autism in children between the ages of 24 and 48 months with a remarkable accuracy rate of 98.5%. The findings, presented at the annual meeting of the Radiological Society of North America in Chicago, highlight the system’s ability to analyze specialized MRIs of the brain for early detection of autism.
Innovative AI system for early diagnosis
The three-stage AI system, devised by a multi-disciplinary team, utilizes diffusion tensor MRI (DT-MRI) to detect how water travels along white matter tracts in the brain. DT-MRI, a specialized technique, provides insights into the connectivity between different brain regions. The AI system involves isolating brain tissue images from DT-MRI scans, extracting imaging markers indicating connectivity levels, and utilizing a machine-learning algorithm to compare patterns in the brains of children with autism against those with typically developing brains.
Urgent need for objective diagnosis
Current diagnostic tools for autism are subjective, particularly when evaluating individuals near the borderline between autism and typical development. The AI system aims to address this challenge by offering an objective and accurate technology for early autism diagnosis. Dr. Ayman El-Baz, the study co-author and chair of the bioengineering department at the University of Louisville, emphasizes the urgent need for such advancements in autism diagnosis.
Significance for early intervention
Early diagnosis of autism is crucial for effective therapeutic intervention, with better outcomes, including increased independence and higher IQs, observed when intervention occurs before the age of three. The researchers applied their methodology to DT-MRI brain scans of 226 children, with promising results that could lead to earlier access to evidence-based intervention for children with autism.
Potential impact on clinical workflow
The AI system could potentially reduce the workload of psychologists by up to 30%. The proposed diagnostic workflow involves an initial assessment by the AI system, followed by a confirmatory session with a psychologist to guide parents on the next steps. This streamlined approach could enhance efficiency in the diagnostic process, facilitating quicker access to intervention for children with autism.
Expert opinions and considerations
While the findings are promising, some experts emphasize the need for further validation and broader applicability. Dr. Leandra Berry from Texas Children’s Hospital acknowledges the potential of diagnostic technology for early autism detection but suggests that additional studies are required to replicate the findings. The study’s focus on children aged 24 to 48 months raises questions about the technology’s utility in younger children.
Dr. Diana Robins, Director of the A.J. Drexel Autism Institute, stresses the importance of including children with other developmental delays in the study sample to ensure accurate differentiation. Additionally, considerations about commonly co-occurring diagnoses, such as attention-deficit/hyperactivity disorder and intellectual disabilities, highlight the complexity of autism spectrum disorders.
Future steps and commercialization
The researchers express their intention to commercialize the AI software and seek clearance from the Food and Drug Administration (FDA). This step is critical to ensuring the technology’s accessibility for widespread use. While the potential benefits are evident, experts caution that the technology is not yet ready for general public use, emphasizing the need for thorough validation and addressing practical challenges, such as the availability of expensive imaging tests and the limited access to technology in certain regions.
The University of Louisville’s AI system presents a promising step forward in the early diagnosis of autism, offering an objective and efficient approach. The potential to reduce psychologists’ workload and expedite therapeutic intervention underscores the significance of such technological advancements in the field of autism research. However, experts call for cautious optimism, emphasizing the need for further validation and addressing practical considerations before widespread implementation.
