An innovative artificial intelligence (AI) system, highlighted at the annual Radiological Society of North America (RSNA) meeting, exhibits an impressive 98.5% accuracy in diagnosing autism in children aged 24 to 48 months. This AI system scrutinizes specialized brain MRIs, particularly DT-MRI scans. It isolates brain tissue images and extracts imaging markers indicative of connectivity between different brain regions. Through a machine learning algorithm, the AI compares these markers in the brains of children with autism to those in typically developing brains. As autism often stems from faulty brain connections, DT-MRI effectively captures these aberrations, shedding light on symptoms like impaired social communication and repetitive behaviors.
In their study, researchers applied this methodology to DT-MRI brain scans from the Autism Brain Imaging Data Exchange-II, involving 226 children aged 24 to 48 months—126 with autism and 100 typically developing children. The AI technology demonstrated exceptional performance, boasting 97% sensitivity, 98% specificity, and an overall accuracy of 98.5% in identifying children with autism.
The key to early intervention lies in capitalizing on brain plasticity—the brain's capacity to normalize function through therapy. Delays in diagnosing infants and young children with autism often result from various factors, such as limited resources at testing centers. According to the researchers, their AI system not only facilitates precise autism management but also streamlines assessment and treatment processes, potentially reducing time and costs associated with these interventions. The AI system generates a comprehensive report detailing affected neural pathways, the anticipated impact on brain functionality, and a severity grade for guiding early therapeutic intervention.
https://www.prnewswire.com/news-releases/novel-ai-system-could-diagnose-autism-much-earlier-301993093.html
