A multi-view imaging dataset for Garden classification of femoral neck fractures

Abstract

# A multi-modal imaging dataset for Garden classification of femoral neck fractures ## Dataset Overview This is a clinically validated, fully anonymized multi-modal imaging dataset for femoral neck fracture Garden classification, addressing the critical lack of public, expert-annotated orthopedic imaging resources. It includes three mainstream clinical imaging modalities (CT axial, coronal, and 3D MIP views), plus paired original and internal fixator occlusion-removed images. Built with strict quality control and patient-level data partitioning, it provides a standardized high-quality foundation for AI-assisted fracture diagnosis model development, benchmarking and clinical translation. ## Ethics Statement This study was approved by the Medical Ethics Committee of Southwest University and cooperating hospitals, and fully complied with the 1964 Helsinki Declaration and its amendments. All data are fully anonymized with no protected health information included. Written informed consent was waived by the ethics committee due to the retrospective and non-invasive nature of the study. ## Data Acquisition and Imaging Protocol All imaging data were collected from cooperating medical institutions in Chongqing, China, from adult patients (age ≥ 18 years) with clinically and radiologically confirmed femoral neck fractures. The raw data are standardized spiral CT volumetric data with uniform scanning parameters: - Tube voltage: 120 kV - Tube current: 200–300 mA - Slice thickness/spacing: 0.625 mm - Matrix: 512×512 - Field of view (FOV): 350×350 mm Using the open-source medical image processing software 3D Slicer, three types of 2D images were reconstructed to simulate clinical routine reading perspectives, all included in this dataset: 1. **Axial CT views**: Cross-sectional slices centered on the femoral neck midpoint, clearly displaying cross-sectional fracture details, fracture line direction and fragment alignment; 2. **Coronal CT views**: Coronal plane slices of the femoral neck longest axis, corresponding to the core perspective of clinical anteroposterior and lateral X-rays; 3. **3D projection views**: Maximum Intensity Projection (MIP) images with a 30° projection angle relative to the coronal plane and 10 mm projection thickness, displaying the overall 3D morphological characteristics of fractures. ## Dataset Composition The dataset is split at the patient level into training, validation and test sets at a 7:1.5:1.5 ratio with a fixed random seed, eliminating data leakage and ensuring balanced baseline characteristics across splits. It contains original and occlusion-removed image sets covering three imaging modalities, complete metadata for all samples, and full reproducible code for preprocessing, occlusion removal and data partitioning, with a clear hierarchical file structure for ease of use. ## Annotation Standard and Quality Control All annotations follow the classic Garden classification system, completed via a standardized blinded workflow: dual independent annotation by senior orthopedic surgeons, with discrepant cases arbitrated by a chief physician to form the gold standard. The final annotation achieved excellent inter-annotator consistency (Cohen’s Kappa = 0.95, 95% CI: 0.93-0.97). All images underwent standardized preprocessing, and a two-stage segmentation-inpainting pipeline was used for occlusion removal, with clinically validated high-quality anatomical structure restoration. Preprocessing Pipeline All images in the dataset were processed with a standardized preprocessing pipeline: 1. All reconstructed 2D images were uniformly resized to 224×224 pixels; 2. Pixel values were normalized to [0,1] using Z-score standardization to eliminate systematic differences caused by scanning parameters; 3. For images with internal fixator occlusion, a dedicated preprocessing pipeline was applied: a U-Net++ model was used for high-precision segmentation of occluded regions, followed by a PatchGAN inpainting algorithm to restore anatomical structure and fracture line details, with a segmentation accuracy of 96.8% for occluded regions. ## Citation When using this dataset, please cite the associated manuscript: Wu, M., Ma, Y., Li, Y., Zhang, X., Zhu, Y., Zhou, J., & Chen, S. (2026). A multi-modal imaging dataset for Garden classification of femoral neck fractures. ## License This dataset is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Users are free to share and adapt the dataset, provided that appropriate credit is given to the original authors, a link to the license is provided, and any changes made are clearly indicated.