This seminar, presented by James S. Duncan, PhD, and hosted by Maciej Mazurowski, PhD, explores advanced neuroimaging analysis in autism spectrum disorders (ASD), with a focus on identifying objective biomarkers and predicting response to behavioral therapies such as Pivotal Response Treatment (PRT). Dr. Duncan presents a range of methodologies, beginning with a Bayesian framework to detect atypical brain connectivity and moving toward machine learning approaches that use convolutional and recurrent neural networks for classification and prediction. The talk also discusses early work involving dynamic causal modeling as a complementary approach to traditional functional connectivity methods. These techniques offer promising tools for individualized diagnosis, treatment planning, and monitoring in ASD. This seminar originally aired on March 7th, 2023. This session was part of the monthly Spark: AI Health Initiative for Medical Imaging seminar series. Spark highlights cutting-edge work in medical imaging research from Duke and beyond, with a mission to develop, validate, and implement AI tools for clinical imaging. The series fostered collaboration between technical and clinical experts. VIEW
