Cape Town - 2026 ISMRM-ISMRT Annual Meeting and Exhibition

AI/DL in Pediatrics

Digital Poster

Pediatrics

Wednesday, 13 May 2026

Digital Posters Row C

14:35 - 15:30

Session Number: 562-04

No CME/CE Credit

Similar Sessions

Machine (ML) and deep learning (DL) and artificial intelligence (AI) are changing how we image children and expanding what we can learn and understand from MRI. This session highlights the use of ML/DL/AI in pediatric MRI across organ systems.

Skill Level: Intermediate

		562-04-001. Multiparametric MRI Radiomics for Noninvasive Risk Stratification in Pediatric Neuroblastoma Matthias Anders, Federico Mollica, Tom Meyer, Reda Tahan, Hedwig Deubzer, Simon Veldhoen, Corona Metz Charité-Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany Impact: Multiparametric MRI Radiomics enables noninvasive risk stratification of rare pediatric neuroblastoma. This method has the potential to minimize biopsy requirements, aid in early treatment planning, and integrate with molecular and histologic biomarkers to guide personalized therapy.
		562-04-002. Implicit neural representations for IVIM parameter estimation: application to Wilms Tumors Gerrit Arends, Joris Harbers, Tom Hendriks, Maxime Chamberland, Frank Simonis, Dennis Klomp, Matthijs Fitski, Annemieke Littooij, Chantal Tax University Medical Center Utrecht, Utrecht, Netherlands Impact: Implicit neural representations enable accurate and spatially regularized estimation of IVIM parameters in a self-supervised framework without training data, improving perfusion quantification and working towards non-invasive tumor subtype characterization of Wilms tumors.
		562-04-003. Deep Learning Reconstruction in Free Breathing Pediatric Abdominal MRI: Application to Multiband DWI and LAVA-Star Eugene Milshteyn, Sergio Valencia, Xinzeng Wang, Patricia Lan, Arnaud Guidon, Teresa Victoria, Michael Gee GE HealthCare, San Ramon, United States of America Impact: Application of deep learning reconstruction to multiband DWI and LAVA-Star can provide an avenue for robust free breathing abdominal imaging for the pediatric cohort.
		562-04-004. Neonatal Brain-Age Prediction and Interpretability Analysis Using CNN- and ViT-Based Models Dayeon Bak, Junghwa Kang, Hyun Gi Kim, Yoonho Nam Hankuk university of Foreign Studies, gyeonggi-do, Korea, Republic of Impact: This study developed a brain age prediction model and visualized its results to explore how age-related information could be represented in the neonatal brain.
		562-04-005. cSVR: Fast Convolutional Slice-to-Volume Reconstruction for Fetal Brain MRI Margherita Firenze, Sean Young, Clinton Wang, Elfar Adalsteinsson, Hyuk Jin Yun, Patricia Grant, Kiho Im, Polina Golland Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, Massachusetts, Cambridge, United States of America Impact: Our fast method supports clinical adoption of SVR and scanner-side decisions on when sufficient data for diagnostic quality 3D volume reconstruction has been acquired.
		562-04-006. Fully automated deep leaning model for evaluation of CSP development in normal fetuses using MRI Zhengyang Zhu, Feng Shi, Chenchen Yan, Bing Zhang Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School — Department of Radiology, Nanjing, China Impact: This research can help radiologists understand CSP changes when gestational age progresses and the normal development of fetal brain. The developed method can assist in identifying abnormal CSP development in clinical settings.
		562-04-007. Investigating Brain Regional Significance in Explainable AI for Pediatric MRI Anik Das, Kaue Duarte, Catherine Lebel, Mariana Bento University of Calgary, Calgary, Canada Impact: By leveraging explainable artificial intelligence (XAI) and statistical analysis, this study evaluates and quantifies the significance of specific brain regions in prenatal alcohol exposure (PAE) classification, which may support revealing critical biomarkers and facilitating effective diagnostic strategies under PAE conditions.
		562-04-008. Patch-based Unsupervised Deep Learning for Brain Anomaly Detection via Age Prediction in Fetal MRI Yingqi Hao, Mingxuan Liu, Juncheng Zhu, Hongjia Yang, Haoxiang Li, Junwei Huang, Yi Liao, Haibo Qu, Qiyuan Tian Tsinghua University, Beijing, China Impact: The proposed 3D patch-based PANDA framework enables accurate, automated detection of fetal brain anomalies from MRI, improving diagnostic efficiency and objectivity. It offers a quantitative tool for early anomaly screening and potential clinical decision support.
		562-04-009. Predicting brain age of autism spectrum disorders aged 2–6 years using routine T1- and T2-weighted MRI zun, ying hu, Rongjia Xiang, Li Kaixin, Huanyu Luo, Di Hu, Huiying Kang, Xin Fan, Huanjie Li, Yun Peng Beijing Children's Hospital, Capital Medical University, National Center for Children's Health, Beijing, China Impact: Evaluating the brain age developmental characteristics of ASD using clinical imaging sequences is more convenient, and efficient than research-grade sequences. Our findings offer a novel perspective for exploring developmental trajectories in ASD and provide theoretical support for clinical interventions.
		562-04-010. Deep Learning-Based White Matter Injury Segmentation With T1WI and T2WI in Multi-Center Infants Aged 6-24 Months Zhen Jia, Man Li, Tingting Huang, Yitong Bian, Xianjun Li, Feng Shi, jian Yang The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, China Impact: Provides a clinically applicable tool for early CP screening using routine MRI sequences.
		562-04-011. Large sample evaluation of AI accelerated Clinical Pediatric Neuroimaging Karen Kettless, Camilla Hansen, Llucia Coll, Marcel Dominik Nickel, Adam. Hansen, Melanie Ganz Siemens Healthcare A/S, Ballerup, Denmark Impact: This study shows that deep learning-based accelerated MRI can preserve or enhance image quality in pediatric imaging using 50% of conventionally acquired k-space data, supporting AI-driven protocols that reduce scan time and improve patient experience without compromising diagnostic confidence.
		562-04-012. Gadolinium-Free MR Perfusion Imaging Based on Generative Adversarial Network for Primary Intracranial Tumor Diagnosis Guoqi Lin, Lukui Xiong, Jiahao Tao, Yihua Chen, Junhuan Hong, Yuwei Pan, Pingping He, Wei Guo, Yan Su, Ming Chen, Yang Song, Dejun She The First Affiliated Hospital of Fujian Medical University, Fuzhou, China Impact: This study enables contrast-free MR perfusion imaging for brain tumors, reducing gadolinium-related risks while preserving diagnostic accuracy. It empowers radiologists to assess tumor vascularity and molecular subtypes safely, prompting new research in contrast-free functional imaging and AI-guided diagnostics.
		562-04-013. Automated Segmentation of Pediatric Hippocampal and Basal Ganglia Structures in Ultra-Low-Field Magnetic Resonance Images Toufiq Musah, Philip Nkwam, Ajay Sharma Kwame Nkrumah University of Science and Technology, Kumasi, Ghana Impact: Automated segmentation of neonatal hippocampi and basal ganglia is feasible at ultra-low field (0.064T) MRI, enabling reliable neuroanatomical assessment in resource-limited settings and paving the way for accessible early neurodevelopmental diagnostics.
		562-04-014. Clinical optimisation of AI-based MELD-GRAPH classifier to localise epileptogenic foci in pediatric focal cortical dysplasia Enrico De Vita, Annemarie Knill, Mathilde Ripart, Kiran Seunarine, Yi Jie Li, Suresh Pujar, J Helen Cross, M Zubair Tahir, Friederike Moeller, Konrad Wagstyl, Sophie Adler, Sniya Sudhakar, Pritika Gaur, Kshitij Mankad, Asthik Biswas, Ulrike Loebel, aswin chari, Martin Tisdall, Felice D'Arco Great Ormond Street Hospital for Children NHS Foundation Trust, London, United Kingdom Impact: MELD-GRAPH has been trained on FLAIR and MPRAGE MRI images for FCD detection. Some groups found that using FLAIR increases false positives. We show that using MP2RAGE-Uni gives similar/slightly better sensitivity than the standard input. MPRAGE-only results in reduced sensitivity.
		562-04-015. Super-Resolution with Noisy References for High-Resolution Pediatric Brain MRI Muye Zhang, Hongjia Yang, Juncheng Zhu, Jiaxin Xiao, Yuhang He, Jialan Zheng, Zihan Li, Ting Yin, Wei Liu, Ziyu Li, Yi Liao, Haibo Qu, Qiyuan Tian Tsinghua University, Beijing, China Impact: Our method reduced the requirement for high-SNR training data in super-resolution tasks which are difficult to obtain. It improves the feasibility of deep learning-based super-resolution for sedation-free pediatric imaging, increasing clinical accessibility for challenging populations.
		562-04-016. Evaluating the Impact of Deep-Learning-Acceleration on Anatomical Imaging in Pediatrics Bryce Geeraert, Xucheng Zhu, Dan Rettmann, Marc Lebel, Catherine Lebel University of Calgary, Calgary, Canada Impact: Deep-learning-accelerated imaging can produce accurate depictions of brain anatomy in half the time of a standard imaging sequence. Nuanced differences are observed between imaging modalities, but head motion seems significantly reduced. Further investigation is required to aid interpretation.