NeckFrac
Description
Prompt detection and localization of vertebral fractures are crucial to preventing neurologic deterioration and paralysis after trauma. The objective was to train deep learning algorithms on CT images from 2,019 patients to detect potential cervical fractures and identify their locations using bounding boxes. The study also included an evaluation of the deep learning model’s performance.
Creator
UC Berkeley
Information
Pediatrics or Adult
Adult
Speciality
Orthopedics
Modality
CT scan
Training
2,019 CT scans with diagnosed cervical spine fractures and an approximately equal number of negative studies
Github
Publication
FDA