Cape Town - 2026 ISMRM-ISMRT Annual Meeting and Exhibition - Classification and Analysis in the Brain

Classification and Analysis in the Brain

Digital Poster

Analysis Methods

Monday, 11 May 2026

Digital Posters Row D

16:10 - 17:05

Session Number: 363-05

No CME/CE Credit

Similar Sessions

This session focusses on image analysis and tools using data processing and AI methods to classify the brain.

		363-05-001. Neuroimaging-Based Analysis: Structural Brain Abnormalities and Diagnostic Classification of Ménière's Disease Wenliang Fan Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China Impact: MD patients showed substantial alteration in gray matter volume and surface indexes across frontal and cingulate gyri.our classification model, incorporating a comprehensive array of neuroimaging features, achieved an impressive accuracy of 84% in distinguishing MD patients from HC.
		363-05-002. Comparative Analysis of Radiologist, Radiomics, and Multidimensional Deep Learning for Preoperative Molecular Classification Xin Ge, Yuhui Xiong, Jing Zhang, Wen Wang Tangdu Hospital, Fourth Military Medical University, Xi'an, China Impact: Standardized quantitative MRI and head-to-head benchmarking identify a deployable 2.5D deep-learning ensemble that improves IDH/1p/19q classification and independent survival stratification, informing preoperative decisions and enabling multicenter, vendor-agnostic trials, prospective utility studies, and AI-assisted treatment selection in neuro-oncology.
		363-05-003. Automated Motion Artifact Check for MRI (AutoMAC-MRI): Explainable severity grading of motion artifacts in MR images Antony Jerald, Dattesh Dayanand Shanbhag, SUDHANYA Chatterjee GE HealthCare, Bengaluru, India Impact: Automated, explainable MRI motion artifact grading accelerates workflows, reduces patient recalls, and improves diagnostic confidence. It empowers clinicians with actionable insights, enhances patient comfort, and opens new research directions in explainable motion detection and severity grading.
		363-05-004. Classification of tumor treatment response from multi-time point multisequence MRI scans Ewunate Kassaw, Satyajit Maurya, Amit Mehndiratta, Anup Singh Indian Institute of Technology, Delhi, India Impact: This study presents an objective and automated approach for evaluating the progress of brain tumor treatment response from multiparametric longitudinal MR scans. This can contribute to the non-invasive and timely diagnosis and treatment of patients.
		363-05-005. Premature brain volume reduction and its dynamics in the early course of multiple sclerosis Hagen Kitzler, Thomas Pollinger, Caroline Koehler, Tjalf Ziemssen, Paul Kuntke Technical University of Dresden, Dresden, Germany Impact: This work provides in vivo evidence of significant differences in age-corrected brain volumes between patients in the early MS compared to healthy controls. Our approach of population-based atrophy categorization and dynamics provides early identification of risk groups with accelerated neurodegeneration.
		363-05-006. SynthScore: predicting reference-based motion artifact scores without a reference image Tamsin Edwards Lambourne, Yuh-Shin Chang, Brooks Applewhite, Malte Hoffmann, Robert Frost Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown, United States of America Impact: We present a deep-learning model where our predicted motion artifact scores have a similar correlation level with radiologist assessment as reference-based metrics, demonstrating applicability for settings where reference images are unavailable, and expert assessment is impractical.
		363-05-007. Impact of MRI Acceleration using Compressed SENSE on Brain Cortical Thickness Measurements Laura Leukert, Fabian Bongratz, Benita Schmitz-Koep, Michael Dieckmeyer, Jan Kirschke, Dennis M Hedderich TUM School of Medicine and Health, Munich, Germany Impact: Differences in MRI acceleration, image quality and software implementation measurably influence brain cortical thickness estimates. These factors must be considered while interpreting morphometric studies to avoid misclassification of neurodegenerative change and to ensure reliable clinical decision-making in longitudinal patient care.
		363-05-008. Deep Learning with MRI and Clinical Data for Differentiation of Radiation Injury from Tumor Recurrence in Brain Metastases Yuhan Liang, Zelong Chen, Menglin Ge, Yulin Wang Chinese PLA General Hospital, Beijing, China Impact: This deep learning tool noninvasively distinguishes radiation injury from tumor recurrence in brain metastases with over 91% accuracy, potentially reducing unnecessary biopsies and optimizing post-radiotherapy management.
		363-05-009. Highly Accelerated Brain MRI Matches Routine Scans for AI Image Analysis Task Performance Moona Mazher, Miguel Rosa-Grilo, Haroon Chughtai, David Thomas, Millie Beament, Frederik Barkhof, Catherine J. Mummery, Nick C. Fox, Geoff Parker, Daniel Alexander University College London, London, United Kingdom Impact: Ultra-fast MRI scans analysed with a foundation model achieve comparable performance to routine scans, supporting their clinical adoption. This enables rapid, reliable neuroimaging, improving patient throughput, accessibility, and scalable workflows without sacrificing diagnostic accuracy.
		363-05-010. Deep learning MRI model for transcription factor-based classification of pituitary neuroendocrine tumor subtypes Elizabeth Ndimulunde, Bing-Fong Lin, Dao-Chen Lin, Chia-Feng Lu National Yang Ming Chiao Tung University, Taipei, Taiwan Impact: This study demonstrates the potential of MRI-based deep learning for comprehensive, non-invasive lineage classification for pituitary neuroendocrine tumors, reducing dependence on postoperative immunohistochemistry and enabling improved preoperative diagnostic accuracy and personalized treatment planning.
		363-05-011. Improving ME-ICA for task fMRI in susceptibility-affected brain regions at 7 Tesla Toshiki Okumura, Ikuhiro Kida Center for Information and Neural Networks, National Institute of Information and Communications Technology, Osaka, Japan Impact: Majority vote increased ME-ICA classification robustness and improved fMRI analysis (GLM and decoding analyses), particularly when including brain regions affected by susceptibility artifacts.
		363-05-012. Exploring Brain Iron Deposition in Thalassemia Using Sub-voxel QSM and Machine Learning Models for Accurate Diagnosis Mingrui Yang, Peng Peng, Huang Yugui, Chunxia Zhu, Jiatong Liang, Guowei Chen, Kong Rong, Cheng Tang, Hui Zhang The First Affiliated Hospital of Guangxi Medical University,, Nanning, China Impact: This study applies the innovative sub-voxel QSM (Chi-separation) technique with machine learning and SHAP analysis to accurately diagnose brain iron deposition in thalassemia, providing new insights for early diagnosis and monitoring.
		363-05-013. Geometric properties of caudate and putamen mark the progression of Huntington’s disease Dorian Pustina, Haikel Bogale, Sandhitsu Das, Brian Avants, Swati Sathe, Cristina Sampaio, Andrew Wood CHDI Management, United States of America Impact: The striatum’s geometry holds untapped biomarker potential. The striatum does not simply shrink, it changes shape. Even using imperfect segmentations, geometric descriptors reveal disease-related signatures that complement volumetric loss.
		363-05-014. Do Perivascular Space Abnormalities in the Brain Differ Among Various Disease States? Thais Bezerra, Zhibin Chen, Lucas Scárdua Silva, Brunno Machado de Campos, Gabriel de Deus Vieira, Thiago Rezende, Alfredo Damasceno, Simone Appenzeller, Marcio Balthazar, Marcondes Cavalcante França Junior, Leonilda Santos, Felipe Von Glehn Silva, Patrick Kwan, Terence O'Brien, Fernando Cendes, Meng Law, Clarissa Yasuda, Ben Sinclair University of Campinas, Campinas, Brazil Impact: Here we provide evidence that there are distinct distributions of PVS abnormalities between different diseases, indicating that PVS may be more than just a general marker of brain health, but are affected differently in different diseases.
		363-05-015. Random matrix theory denoising for preclinical diffusion MRI Ricardo Coronado-Leija, Yoko Bekku, Jiangyang Zhang, James Salzer, Els Fieremans, Dmitry Novikov Center for Advanced Imaging Innovation and Research (CAI²R), New York University Grossman School of Medicine, New York, United States of America Impact: Preclinical scanners achieve very strong diffusion weightings, but these are typically not used due to low signal-to-noise ratio and noise floor constraints. Here, we demonstrate how random matrix theory denoising before image reconstruction improves preclinical high $b$-value diffusion MRI.
		363-05-016. T2-Weighted Contrasts and Manual Editing Significantly Affect FreeSurfer Cortical Morphometry and Clinical Associations Haley Wiskoski, Summan Zahra, Juan Arias, Scott French, Kevin Johnson, Simeon Smith, Caleb Boehler, Raza Mushtaq, Maria Altbach, Gene Alexander, Theodore Trouard, Craig Weinkauf University of Arizona, Tucson, United States of America Impact: This study demonstrates that adding T2-weighted images to FreeSurfer systematically alters cortical morphometry and weakens biological associations. Establishing that T1-only pipelines, with manual editing, yield more reliable results supports methodological standardization and enhances reproducibility across neuroimaging and Alzheimer’s disease research.