OrthoFrac-XR: Clinically Validated X-ray Images for Bone Fracture Detection
by Md. Arham Tabib·Updated 1mo ago
552.8 MB1files
Available on 1 platform
Sign in to view source links and access this dataset
Description
1,493 fully classified orthopedic X-ray images from approximately 1,300 distinct patients, collected from several Bangladeshi institutions. The dataset was created by Md. Arham Tabib and last updated on April 28, 2026. It includes a structured metadata file with clinical data such as patient age, gender, bone type, and fracture measurements.
Use Cases
Training fracture detection models based on high-resolution X-ray images.
Classifying fracture types (distal, proximal, post-fracture) based on the four clinically significant classes.
Conducting integrated radiographic-clinical analysis based on the structured metadata containing patient demographics and fracture measurements.
Developing models for resource-limited healthcare settings based on the dataset's focus on accessible X-ray imaging.
Strengths
1,493 images independently validated by orthopedic specialists from Dhaka Medical College for labeling reliability.
Includes a structured metadata file (bone_metadata.csv) with clinically significant data like fracture gap measurements and primary clinical observations.
Images are divided into four clinically significant classes: distal fracture (314 images), proximal fracture (254 images), post-fracture (349 images), and non-fracture (576 images).
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count for the metadata file is unknown, which may limit suitability assessment.
Data reflects a geographic bias, having been collected solely from Bangladeshi institutions.
Provenance
Source
Several Bangladeshi institutions.
Collection Method
Collected and independently validated by orthopedic specialists.
Time Range
null
Freshness
Last updated 2026-04-28 04:19:12; freshness should be verified.
Geography
Bangladesh
Data is provided in a ZIP file (552.8 MB) under a CC-BY-4.0 license.