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

RSNA Cervical Spine Fracture Detection (RSNA-CSF) Dataset

https://mira.rsna.org/dataset/4

1.0

informatics@rsna.org

Largest publicly available, multi-institutional and multinational expert-labeled dataset of cervical spine fracture CT images for AI research. Used in the RSNA 2022 AI Challenge. Includes patient-level, vertebra-level, bounding box, and segmentation annotations. Data from 12 institutions across nine countries and six continents. The dataset included 3112 CT scans; 1445 studies positive for fracture (954 men, 491 women; mean age, 56.78 years ± 21.97), supplemented with 1667 studies negative for fracture (1022 men, 645 women; mean age, 50.61 years ± 21.29).

Cervical spine injuries are a common form of traumatic injury, affecting more than 3 million patients per year in North America. Cervical fractures can lead to substantial disability. There is a need for fast and accurate diagnosis, providing an excellent clinical use case for AI algorithms. The lack of publicly available, expertly annotated cervical spine fracture datasets hinders further improvements in model performance.

Curated and created expert annotation of a large high-quality cervical spine fracture CT dataset from 12 institutions from six different continents. Unconventional annotation methods by means of prelabeling were used, with labels provided by contributing sites. Hybrid schema chosen: study-level annotation detailing each cervical spine level fractured or no fracture in control dataset. Smaller subset (approx. 16% of positive fracture cases) assigned image-level annotations, including bounding boxes. Additional subset of cases containing segmentation masks of the vertebrae.

Data distribution used for the Kaggle competition, with 2019 cases in the training set, 304 cases in the public test set, and 789 cases in the private test set. Great care was also taken to ensure that data were distributed equally with respect to sex, age, contributing site, and fracture level across the training, validation, and test sets.

The strict inclusion criteria of axial noncontrast 1-mm-thick section images may limit its application to practices that have different section acquisitions or reformat their CT cervical spine scans from a postcontrast acquisition. The dataset treats acute and chronic fractures the same. This dataset excluded patients who underwent prior surgery because of the challenges of streak artifacts and altered anatomy. Some fractures were visualized that were not accounted for in the radiologist’s report, with the radiologist’s report chosen as ground truth.

A single study may contain multiple fractures.

Images are commonly confounded by superimposed degenerative disease and osteoporosis, making fracture detection more complex.