ISMRM25 - Development & Implementation of Artificial Intelligence in Pediatric Magnetic Resonance Imaging: Current Uses, Challenges & Future Directions

ISMRM & ISMRT Annual Meeting & Exhibition • 10-15 May 2025 • Honolulu, Hawai'i

Development & Implementation of Artificial Intelligence in Pediatric Magnetic Resonance Imaging: Current Uses, Challenges & Future Directions

Member-Initiated Session

Thursday, 15 May 2025

313B

13:15 - 15:15

Moderators: Matthew Barkovich

Session Number: MIS-06

No CME/CE Credit

Overview:
The Pediatric MR Study Group Member Initiated Symposium on AI in Pediatric MRI is a collaboration with the Society for Pediatric Radiology (SPR) MRI Committee and the World Federation of Pediatric Imaging (WFPI).
The unique challenges of the development and incorporation of artificial intelligence (AI), both machine learning and deep learning, in magnetic resonance imaging in pediatric populations is the focus of this member-initiated symposium. A primary focus on AI in routine, clinical workflow will include how AI may improve exam scheduling and protocoling, sequences optimization, image segmentation in infants and children at different developmental stages, and image acquisition speed. Generalizability and the “value add” of the clinical translation of AI into practice will be discussed including improvements in reporting and enhanced efficiency in the communication of findings (i.e., “closing the loop”). How AI may improve detection and diagnosis, treatment management, and prognostication will be explored. Research applications will focus on image quality improvements and post-processing as well as integration of genetics, -omics, and clinical data into application development for clinical use. Challenges of generalizability in pediatrics and unique challenges of rare diseases will be explored including a need for multicentre registries, multicentre collaboration, and data sharing.

Target Audience:
Clinical radiologists, MR technologists, researchers, and trainees (MD, MD-PhD, and PhD residents, fellows, and post- docs).

Learning Objectives:
Upon completion of this activity, participants should be able to:

• Recognize the unique challenges of AI development and use in pediatric MR imaging related to anatomic differences in neonates, infants, young children, and adolescents and its impact on clinical translation;
• Develop a framework for understanding how AI may improve routine MR clinical practice in paediatrics with faster imaging and workflow improvements from exam scheduling to protocoling;
• Through a deeper dive in current applications, challenges, and future directions in pediatric MR neuroimaging with a focus of cancer genetics, identify issues with generalizability, and need for multicenter collaboration and data sharing; and
• Incorporate AI tools into research with a particular focus on analytics and image quality.

13:15		Challenges of Image Segmentation in Infants & Children Maria Deprez
13:45		How Deep Learning May Speed Up Image Acquisition Patricia Johnson
14:15		Challenges in Generalizability of Neuroimaging AI in Routine Use in Pediatrics Risto Filippi
14:45		AI Improvement in Image Quality & Analysis in Research & Clinical Practice Yohan Jun