Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/dataset.json

Name

UKHD–UKB Longitudinal MS MRI for New Lesion Classification (internal/external cohorts used in Brugnara et al., 2024)

Link

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11605143/

Indexing

Keywords: multiple sclerosis, FLAIR, MRI, new lesion detection, longitudinal, external validation, synthetic data augmentation, GAN, attention-based CNN

Content: NR, MR, RS

RadLex: RID28801, RID10312, RID35806, RID50239

SNOMED: 24700007

Author(s)

Gianluca Brugnara

Chandrakanth Jayachandran Preetha

Katerina Deike

Robert Haase

Thomas Pinetz

Martha Foltyn-Dumitru

Mustafa A. Mahmutoglu

Brigitte Wildemann

Ricarda Diem

Wolfgang Wick

Alexander Radbruch

Martin Bendszus

Hagen Meredig

Aditya Rastogi

Philipp Vollmuth

Organization(s)

Heidelberg University Hospital

Bonn University Hospital

License

Text: Not publicly shared; imaging data cannot be shared due to privacy regulations.

Contact

Corresponding author: P. Vollmuth, email as provided in article: ed.nnobku@htumllov.ppilihp

Funding

Acknowledgments list support including Else Kröner Fresenius Foundation (2022_EKCS.17) and OASIS-3 grant support (NIH P30 AG066444, P50 AG00561, P30 NS09857781, P01 AG026276, P01 AG003991, R01 AG043434, UL1 TR000448, R01 EB009352).

Ethical review

Institutional review boards at Heidelberg University Hospital and Bonn University Hospital approved the study and waived informed consent.

Comments

Retrospective internal cohort from Heidelberg University Hospital (UKHD) and external test cohort from Bonn University Hospital (UKB) used to train and evaluate AI models for detection of new FLAIR lesions in multiple sclerosis during longitudinal follow-up. Imaging data cannot be shared due to privacy regulations.

Date

Published: 2024-10-16

References

[1] Brugnara G, Jayachandran Preetha C, Deike K, et al.. "Addressing the Generalizability of AI in Radiology Using a Novel Data Augmentation Framework with Synthetic Patient Image Data: Proof-of-Concept and External Validation for Classification Tasks in Multiple Sclerosis". Radiology: Artificial Intelligence. 2024-10-16. doi:10.1148/ryai.230514. PMID: 39412405. PMCID: PMC11605143.

Dataset

Motivation

Evaluate whether GAN-based synthetic patient data augmentation improves generalizability for detecting new MS lesions on follow-up MRI.

Sampling

Retrospective single-center internal cohort (UKHD, 3.0-T Siemens) from Jan 2010–Jun 2022; external cohort (UKB, 1.5-T or 3.0-T Philips and GE) from Jan 2009–Apr 2023.

Partitioning scheme

Patients split 4:1 into cross-validation (fivefold) and internal test sets at UKHD; external testing performed on an independent UKB cohort.

Missing information

Ethnicity was available only for OASIS-3; detailed demographic breakdowns and imaging file formats not reported.

Relationships between instances

Longitudinal exam pairs per patient (baseline and follow-up). Multiple pairs per patient exist in internal cohort; bootstrapping used to account for repeated measures.

External data

Synthetic training augmentation used T1-weighted data from OASIS-3 combined with internal lesion/segmentation information to generate synthetic FLAIR images.

Confidentiality

Imaging data cannot be shared due to privacy regulations.

Re-identification

Not discussed beyond IRB approval and privacy restrictions; data not shared to protect privacy.

Sensitive data

Clinical neuroimaging of patients with MS; subject to privacy regulations.