Cape Town - 2026 ISMRM-ISMRT Annual Meeting and Exhibition

Machine Learning in Neuroimaging

Digital Poster

Neuro B

Wednesday, 13 May 2026

Digital Posters Row C

13:40 - 14:35

Session Number: 562-03

No CME/CE Credit

Similar Sessions

This session brings together work where machine learning is applied to neuroimaging, whether at the acquisition, analysis, or decision-making level.

		562-03-001. Clinical validation of deep learning accelerated 3D MPRAGE at 1.5T and 3T: repeatability and volumetric analyses Atefeh Zeinoddini, Eugene Milshteyn, MARY THOMAS, Dan Rettmann, Susie Huang, Ivan Jambor Massachusetts General Hospital, Boston, United States of America Impact: DL Speed MPRAGE provides accurate volumetric data and good short-term repeatability at significantly reduced scan times compared with convectional MPRAGE.
		562-03-002. Subject-specific microstructure integration in virtual brain models: advancing Brain Digital Twin technologies Eleonora Lupi, Fulvia Palesi, Anita Monteverdi, Marta Gaviraghi, Carolyn McNabb, Pedro Luque Laguna, Eirini Messaritaki, Ilaria Gabusi, Alessandro Daducci, Marco Palombo, Mara Cercignani, Egidio D’Angelo, Claudia Gandini Wheeler-Kingshott University of Pavia, Pavia, Italy Impact: By incorporating MRI-derived, subject-specific microstructural and neural conduction features, our approach enables personalised simulations of brain dynamics in The Virtual Brain, marking a significant step toward precision neuro-modelling and the use of Brain Digital Twins in personalised medicine.
		562-03-003. Slice-wise Vision Transformer with Max-Aggregation for Multi-label Quality Assessment of Ultra-Low-Field Pediatric Brain MRI Jiaqi Dou, Song Tian Tsinghua University, Beijing, China Impact: This study introduces a slice-wise Vision Transformer with max-aggregation that achieves state-of-the-art performance in ultra-low-field pediatric MRI quality assurance, providing objective and reproducible image quality control that supports confident interpretation and ensures reliable data for downstream analysis.
		562-03-004. A Multimodal MRI–Clinical Deep Learning Model for Stratifying CSF Tap-Test Responders in Normal-Pressure Hydrocephalus Haoyue Guan, Yuwei Dai, Yuli Wang, Farzad Maroufi, Jon Socha, John Theodroe, Sevil Yasar, Danica Cecil, Vora Maulik, Xue Feng, Mark Luciano, Harrison Bai Johns Hopkins University, Baltimore, United States of America Impact: The multi-modality deep learning model integrating radiological image features and clinical variables demonstrated potential for identifying NPH patients most likely to benefit from CSF tap test.
		562-03-005. Pre-contrast 3D-MAGIC Multiparametric Mapping in Contrast-Enhancing Gliomas and Brain Metastases Noemi Sgambelluri, Dirk H. J. Poot, Frans Vos, Gonzalo Mosquera Rojas, Juancito van Leeuwen, Florian Wiesinger, Marion Smits, Juan Hernandez-Tamames Erasmus MC, Rotterdam, Netherlands Impact: Using pre-contrast quantitative MRI maps acquired in patients, we observe pre-contrast T1 and T2 differences which are promising to predict T1-weighted contrast enhancement, reducing gadolinium use and enabling safer, data-driven imaging for brain tumors.
		562-03-006. Assessment of deep learning–based image reconstruction in orbital MRI Aurore Sajust de Bergues de Escalup, Augustin Lecler, Émilie Poirion, Caroline Papeix, Romain Deschamps, Dan Milea, Julien Savatovsky, Loïc Duron, Emma O’Shaughnessy Fondation Rothschild, Paris, France Impact: Deep learning–based reconstruction improves orbital MRI quality across quantitative and qualitative parameters without altering diagnostic findings. Integrating AI-driven denoising into clinical protocols could enhance image interpretability and reader confidence in assessing optic nerve and orbital disorders.
		562-03-007. Vascular Atrophy and Lesion Segmentation in Stroke: Extending Multi-U-Net LST-AI S Senthil Kumaran, Himanshu Singh, Yug Kaushik, Chirag Sharma, Vishnu V. Y., Leve Joseph Devarajan Sebastian, Ajay Garg All India Institute of Medical Sciences, New Delhi, India Impact: This study presents a vascular-aware lesion segmentation framework that effectively distinguishes stroke lesions from WMH across different scanners. This approach enhances reproducibility and anatomical precision, thereby advancing neurovascular modelling, longitudinal stroke assessment, and large-scale lesion–connectome research.
		562-03-008. Rectified Flow for Missing MRI Modality Reconstruction in Glioma Qinqin Xie, Haifeng Wang, yuxia liang, Yu Shang, Jin wang, Ming Zhang, chen niu The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, China Impact: Applying the RF model to glioma modality completion enables more stable cross-modal mapping, improving the spatiotemporal consistency and physiological plausibility of generated results, thereby providing more reliable support for clinical multimodal image analysis.
		562-03-009. Superficial Siderosis Detection and Localization Using Simulation and Guided Backpropagation Mohammadreza Amirian, Lucas Burget, Bianca Mazini, Silvia Pistocchi, Giulia Bommarito, Vincent Dunet, Gilles Allali, Hagen Kitzler, Bénédicte Maréchal, Jonathan Disselhorst Siemens Healthineers International AG, Lausanne, Switzerland Impact: We proposed a technique to train the localization of superficial siderosis without segmentation labels. We further showed that the resulting model can detect the presence of real superficial siderosis with promising performance.
		562-03-010. Human Brain Age Prediction using 0.055T Ultra-low-field MRI Bowen Qiu, Vick Lau, Xuehong Lin, Junhao Zhang, Xiang Li, Gilberto K. K. Leung, Yujiao Zhao, Alex T. L. Leong, Ed X Wu The University of Hong Kong, Hong Kong, China Impact: Brain age prediction can be applied on 0.055T ultra-low-field MRI with deep learning partial-Fourier reconstruction and super-resolution for image enhancement. This potentially empowers accessible brain health monitoring.
		562-03-011. Automatic White Matter Segmentation on Ultra-High Contrast dSIR Images Using nnU-Net Samuel Porter, Maryam Tayebi, Jiantao Shen, Paul Condron, Mark Bydder, Gil Newburn, Ben Bristow, Eryn Kwon, Daniel Cornfeld, Miriam Scadeng, Tracy Melzer, Samantha Holdsworth, Graeme Bydder University of Canterbury, Christchurch, New Zealand Impact: UHC dSIR combined with a self-configuring 3D nnU-Net delivers accurate, scalable white-matter segmentation from minimal labels, enabling reproducible quantification of WM signal changes. This lowers manual burden, supports multi-site standardisation, and accelerates clinical translation of UHC as a sensitive biomarker.
		562-03-012. Differentiating MDD and PTSD through Dynamic Functional Connectivity and Machine Learning Yiwen Ou, Jing Jiang, Lei Li, Qiyong Gong State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College. Address: No.1, Shuaifuyuan, Dongcheng District, Beijing 100730, China Impact: Identifying distinct dynamic functional connectivity patterns in MDD and PTSD enables improved neurobiological differentiation, informs targeted therapeutic strategies, and opens avenues for network-based biomarkers guiding diagnosis and personalized intervention.
		562-03-013. Impact of simulated MRI artifacts on deep learning-based brain age prediction Janine Hendriks, Michelle Jansen, Richard Joules, Óscar Peña-Nogales, Femke Elsen, Anya Povolotskaya, Paulo Rodrigues, Frederik Barkhof, Anouk Schrantee, Henk Mutsaerts Amsterdam University Medical Center, University of Amsterdam, Amsterdam, Netherlands Impact: This study demonstrates that common MRI artifacts, especially motion and ghosting, substantially distort brain age predictions from deep learning-based algorithms. Quantifying these effects highlights the need for artifact-aware model development to improve the reliability of MRI-based brain age biomarkers.
		562-03-014. Machine Learning for Multiple Sclerosis: Multimodal Classification, Phenotype Differentiation and Disability Prediction Paola Valsasina, Loredana Storelli, Giulia Mazzetti, Nicolò Tedone, Patrizia Pantano, Silvia Tommasin, Antonio Gallo, Alvino Bisecco, Nicola De Stefano, Alessia Bianchi, Elisabetta Pagani, Maria A. Rocca, Massimo Filippi IRCCS San Raffaele Scientific Institute, Milan, Italy Impact: Integrating demographic, clinical, and MRI features within machine learning frameworks effectively distinguished multiple sclerosis patients from controls, identified disease phenotypes, and predicted disability. Grey matter volumetry emerged as a key neuroimaging biomarker driving model performances.
		562-03-015. Automated Detection of Neuromelanin Neurodegeneration Using Deep Learning in a Latin American Population: MEX-PD MRI Study Shreya Shrivastava, Juan Manuel Esquivias Farias, Stéphane Lehéricy, Yamil Matuk-Pérez, Erick Humberto Pasaye, César Domínguez-Frausto, Miguel E. Rentería, Alejandra Evelyn Ruiz-Contreras, Alejandra Medina-Rivera, Sarael Alcauter, Rahul Gaurav Paris Brain Institute - ICM, Inserm U1127, CNRS UMR 7225, Sorbonne Université, UMR S 1127, Paris, France Impact: Parkinson's Disease (PD) is shaped by both genetic and environmental factors. However, most of the understanding of PD comes from European populations and non-European populations remain underrepresented. Henceforth, we explored nigral neuromelanin MRI changes using underrepresented Latin American population.