My research focuses on the development of translational artificial intelligence (AI) methods for pediatric care. My background in computational physics, statistics and engineering, and my experience in pediatric imaging led me to develop AI-based quantitative solutions for complex diagnostic problems to help clinicians improve the quality of care for kids.
I’m building a lab that does translational AI research with expertise in imaging, text analytics and time-series problems for leveraging radiology data to improve pediatric care and outcomes. I work to develop pipelines to efficiently curate medical imaging data, generate ground truth for complex diagnostic problems and develop AI models for abdominal and chest imaging applications. I also work to devise clinical implementation and monitoring strategies for tracking AI model performance and their interaction with clinicians. My goals are to provide AI-based quantitative metrics for accurate clinical diagnosis, precise outcome prediction and improved clinical workflow.
Several projects I’m involved with utilize deep learning techniques. I am currently working on projects involving automatic identification and contouring of skeletal muscle mass for sarcopenia diagnosis, a multi-institutional protocol-agnostic MRI liver segmentation algorithm and an automatic detector for non-diagnostic neck/airway X-ray exams in the emergency department. In addition, I’ve developed and validated whole-body tissue segmentation models for pediatric CT scans using traditional machine learning and radiomic feature selection algorithms.
I am also an Associate Director of the AI imaging core at Cincinnati Children’s, a multidisciplinary effort to establish infrastructure, policy and procedures for future AI imaging applications. Readers interested in my research can contact me for more details. I am always open to new ideas and collaborations. I have been a researcher for over five years, and began my work at Cincinnati Children’s in 2017.