RSNA Abdominal Traumatic Injury CT - RSNA/AI-Challenge-Data GitHub Wiki
Blunt force abdominal trauma is among the most common types of traumatic injury, with the most frequent cause being motor vehicle accidents. Abdominal trauma may result in damage and internal bleeding of the internal organs, including the liver, spleen, kidneys, and bowel. Detection and classification of injuries are key to effective treatment and favorable outcomes. A large proportion of patients with abdominal trauma require urgent surgery. Abdominal trauma often cannot be diagnosed clinically by physical exam, patient symptoms, or laboratory tests. Prompt diagnosis of abdominal trauma using medical imaging is thus critical to patient care. AI tools that assist and expedite diagnosis of abdominal trauma have the potential to substantially improve patient care and health outcomes in the emergency setting. To create the ground truth dataset, RSNA collected imaging data sourced from 23 sites in 14 countries on six continents, including more than 4,000 CT exams with various abdominal injuries and a roughly equal number of cases without injury.
Description
The dataset is contained in a Zip archive that includes both DICOM image files (.dcm) and tabular annotation files (.csv). A detailed description of the dataset is provided in Rudie JD et al. "Radiological Society of North America - Abdominal Traumatic Injury CT (RATIC) Dataset," arXiv:2405.19595 (https://doi.org/10.48550/arXiv.2405.19595).
License
You may access and use these de-identified imaging datasets and annotations (“the data”) for non-commercial purposes only, including academic research and education, as long as you agree to abide by the following provisions: Not to make any attempt to identify or contact any individual(s) who may be the subjects of the data. If you share or re-distribute the data in any form, include a citation to the “RSNA Abdominal Traumatic Injury CT (RATIC) Dataset, Copyright RSNA, 2023” as follows: Rudie JD, et al. The RSNA Abdominal Traumatic Injury CT (RATIC) Dataset [10.48550/arXiv.2405.19595]. arXiv:2405.19595."
Tutorial
Files
train.csv Target labels for the train set. Note that patients labeled healthy may still have other medical issues, such as cancer or broken bones, that don't happen to be covered by the competition labels.
- patient_id - A unique ID code for each patient.
- [bowel/extravasation]_[healthy/injury] - The two injury types with binary targets.
- [kidney/liver/spleen]_[healthy/low/high] - The three injury types with three target levels.
- any_injury - Whether the patient had any injury at all.
[train/test]_images/[patient_id]/[series_id]/[image_instance_number].dcm The CT scan data, in DICOM format. Scans from dozens of different CT machines have been reprocessed to use the run length encoded lossless compression format but retain other differences such as the number of bits per pixel, pixel range, and pixel representation. Expect to see roughly 1,100 patients in the test set.
[train/test]_series_meta.csv Each patient may have been scanned once or twice. Each scan contains a series of images.
- patient_id - A unique ID code for each patient.
- series_id - A unique ID code for each scan.
- aortic_hu - The volume of the aorta in hounsfield units. This acts as a reliable proxy for when the scan was. For a multiphasic CT scan, the higher value indicates the late arterial phase.
- incomplete_organ - True if one or more organs wasn't fully covered by the scan. This label is only provided for the train set.
sample_submission.csv A valid sample submission. Only the first few rows are available for download.
image_level_labels.csv Train only. Identifies specific images that contain either bowel or extravasation injuries.
- patient_id - A unique ID code for each patient.
- series_id - A unique ID code for each scan.
- instance_number - The image number within the scan. The lowest instance number for many series is above zero as the original scans were cropped to the abdomen.
- injury_name - The type of injury visible in the frame.
segmentations/ Model generated pixel-level annotations of the relevant organs and some major bones for a subset of the scans in the training set. This data is provided in the nifti file format. The filenames are series IDs. You can find a description of the source model (total segmentator) here and the data used to train that model here.
Note that the NIFTI files and DICOM files are not in the same orientation. Use the NIFTI header information along with DICOM metadata to determine the appropriate orientation.
[train/test]_dicom_tags.parquet DICOM tags from every image, extracted with Pydicom. Provided for convenience.