Overview

Schema Version

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

Name

Rotator cuff MRI scans for AI segmentation (232 patients)

Link

https://dx.doi.org/10.1148/ryai.220132

Indexing

Keywords: Rotator Cuff, Artificial Intelligence, Segmentation, Fat Infiltration, Muscle Volume, MRI, Shoulder

Content: MK, MR

RadLex: RID35976, RID12778, RID10312, RID10693, RID1938, RID16652, RID1905

SNOMED: 926335004

Author(s)

Lara Riem

Xue Feng

Matthew Cousins

Olivia DuCharme

Elizabeth B. Leitch

Brian C. Werner

Andrew J. Sheean

Joe Hart

Ivan J. Antosh

Silvia S. Blemker

Organization(s)

Springbok Analytics

Department of Orthopedic Surgery, University of Virginia Medical School

Brooke Army Medical Center

License

Contact

moc.scitylanakobgnirps@rekmelb.aivlis

Funding

Supported by the NIH–National Institute of Arthritis and Musculoskeletal and Skin Diseases (grant 5R41AR078720). Springbok Analytics provided financial support.

Ethical review

Retrospective study at two surgical clinics; both locations had institutional review board approval with waiver of informed consent and were compliant with HIPAA guidelines.

Date

Published: 2023-02-08

References

[1] Riem L, Feng X, Cousins M, DuCharme O, Leitch EB, Werner BC, Sheean AJ, Hart J, Antosh IJ, Blemker SS. "A Deep Learning Algorithm for Automatic 3D Segmentation of Rotator Cuff Muscle and Fat from Clinical MRI Scans". Radiology: Artificial Intelligence. 2023-02-01. doi:10.1148/ryai.220132. PMID: 37035430. PMCID: PMC10077094.

Dataset

Motivation

Develop and validate an AI algorithm for 3D segmentation of rotator cuff muscles and intramuscular fat to enable objective quantification of muscle volume and fat infiltration from clinical MRI scans.

Sampling

Retrospective collection of shoulder MRI scans from two surgical clinics (2007–2022), including controls (no diagnosed RC abnormalities) and patients with RC tears.

Partitioning scheme

Random split: training used 48 control and 154 rotator cuff tear scans; validation used 15 control and 15 rotator cuff tear scans (three per Goutallier grade).

Missing information

Variable sagittal scan coverage; relatively fewer high Goutallier grades; small validation set; no external validation.

Relationships between instances

Each MRI scan includes multiple regions of interest: humerus, scapula, clavicle, and four rotator cuff muscles (supraspinatus, infraspinatus, teres minor, subscapularis) plus their corresponding intramuscular fat.

Confidentiality

HIPAA-compliant retrospective clinical MRI dataset; IRB approval with waiver of informed consent.

Sensitive data

Clinical imaging data of patients (health data).