Vertebrae for Scoliosis

Published: 25 November 2025| Version 2 | DOI: 10.17632/4kby36n3ng.2
Contributors:
İrfan Harun İlkhan, Haluk Gümüşkaya, Firdevs Turgut

Description

This dataset comprises a curated, de-identified collection of posterior–anterior (PA) spine radiographs in PNG format prepared for research and educational use in scoliosis analysis. The images were obtained from Roboflow and organized under an anonymized file-naming scheme to remove personal identifiers; embedded metadata were stripped to ensure that no protected health information (PHI) is included. Where necessary for consistency across files, images were normalized and resized to a research-friendly resolution without altering diagnostically relevant morphology. The collection is suitable for common computer-vision tasks, including vertebral parsing, vertebra segmentation, and downstream Cobb-angle estimation and robustness studies. Typical use cases include method development, benchmarking, and pedagogy in medical imaging and machine learning. This resource is distributed under the CC BY 4.0 license. Users should cite this dataset and acknowledge Roboflow as the acquisition source. The dataset is intended strictly for research and education; it is not a medical device and must not be used for clinical decision-making.

Files

Categories

Computer Vision, Image Segmentation, Machine Learning, Scoliosis, X-Ray Imaging, Vertebra, Clinical Education, Deep Learning

Licence