Overview

Schema Version

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

Name

Automated Inline Myocardial Segmentation of Joint T1 and T2 Mapping Using Deep Learning (mSASHA CMR dataset)

Link

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9885378/

Indexing

Keywords: myocardial segmentation, cardiac MRI, T1 mapping, T2 mapping, mSASHA, deep learning, inline analysis, Siemens Aera 1.5T

Content: CA, MR

RadLex: RID35976, RID10312

Author(s)

James P. Howard

Kelvin Chow

Liza Chacko

Mariana Fontana

Graham D. Cole

Peter Kellman

Hui Xue

Organization(s)

National Heart and Lung Institute, Imperial College London

National Amyloidosis Centre, UCL

Siemens Medical Solutions USA

National Heart, Lung, and Blood Institute, NIH

Funding

British Heart Foundation (FS/ICRF/22/26039); NIHR Imperial Biomedical Research Centre. Funders had no role in study design, analysis, or manuscript drafting.

Ethical review

Ethical approval granted by the UK Health Research Authority (IRAS 243023); written informed consent waived due to use of fully de-identified patient data.

Comments

Retrospective study of joint cardiac MR T1/T2 mapping (mSASHA) across two London hospitals with expert-labeled endocardial and epicardial contours used to train and test a CNN for automated segmentation and segment-wise mapping.

Date

Published: 2022-11-09

Created: 2020-03-01

References

[1] Howard JP, Chow K, Chacko L, Fontana M, Cole GD, Kellman P, Xue H. "Automated Inline Myocardial Segmentation of Joint T1 and T2 Mapping Using Deep Learning". Radiology: Artificial Intelligence. 2023-01-01. doi:10.1148/ryai.220050. PMID: 36721410. PMCID: PMC9885378.

Dataset

Motivation

Develop and validate an AI method for automated segmentation and segment-wise analysis of jointly acquired cardiac T1 and T2 maps, enabling inline clinical deployment.

Sampling

Consecutive adult patients undergoing CMR at two tertiary centers in London between March and November 2020.

Partitioning scheme

Consecutive adult patients from two hospitals; images from the final 3 weeks assigned to a holdout testing set. Training/validation used earlier images.

Missing information

Patient-level baseline characteristics not available under ethical restrictions.

Relationships between instances

Multiple short-axis sections (basal, mid, apical) per patient; maps acquired pre- and post-contrast when clinically indicated; multiple maps per patient.

Noise

Imaging artifacts present; 280/3426 training-assigned maps and 30/509 testing-assigned maps excluded due to artifacts; motion correction inherent to mSASHA sequence.

External data

None reported.

Confidentiality

Fully de-identified patient imaging data; consent waived.

Re-identification

Data were fully de-identified prior to analysis; low re-identification risk.

Sensitive data

Medical imaging data of cardiac patients; de-identified before analysis.