Bone Fracture X-ray Dataset: Simple vs. Comminuted Fractures

Published: 3 December 2024| Version 3 | DOI: 10.17632/vg95gvhj3y.3
Contributors:
Fahim Faisal Talha Talha,
,

Description

Dataset Overview: This dataset has been curated to support research in bone fracture classification, focusing on simple and comminuted fractures. It includes high-quality X-ray images and a diverse set of augmented images to facilitate the development and evaluation of machine learning models in medical imaging. The dataset is ideal for image classification, segmentation, and fracture type recognition tasks. Fracture Categories: 1. Simple Fracture: Source: Exclusively sourced from hospital records. Original Images: 1,211 images. Augmented Images: 6,311 images. Total Images: 7,522 images. 2. Comminuted Fracture: Source: A mix of hospital-sourced images and web-sourced images (approximately one-tenth from web pages). Original Images: 1,173 images. Augmented Images: 7,366 images. Total Images: 8,539 images. Key Features: Number of Original Images: 2,384. Number of Augmented Images: 13,677. Total Dataset Size: 16,061 images (Original + Augmented). File Formats: JPG. Augmentation Techniques: Zoom: Randomized scaling. Rotation: ±30° rotation. Brightness and Contrast Adjustments: 80%–120% range. Flips: Vertical and horizontal flips. Update Details (Version 3): Original Submission Date: 15 Nov 2024. Update Date: 02 Dec 2024. Changes in This Version: Expanded Dataset Size: From: 6,798 total images (previous version). To: 16,061 total images. Augmented Images: Simple Fracture: Increased from 3,280 to 6,311 images. Comminuted Fracture: Increased from 2,900 to 7,366 images. Augmentation Enhancements: Additional transformations were applied to ensure varied and realistic images for training. Applications: Medical Imaging: Train and evaluate models for fracture classification and identification. Healthcare Technology: Support the development of diagnostic tools and mobile applications for real-time fracture detection. Medical Research: Aid in understanding fracture patterns and their visual indicators. Dataset Collection: Simple Fracture Images: Exclusively sourced from hospitals to ensure clinical accuracy and relevance. Comminuted Fracture Images: Sourced from a mix of hospital records and web pages, representing diverse imaging conditions and perspectives. Realistic Scenarios: The dataset simulates real-world clinical settings with varying lighting conditions, orientations, and imaging environments to provide robust training data. This dataset is a valuable resource for advancing medical imaging research and developing machine learning models tailored for fracture detection and classification tasks.

Files

Institutions

Daffodil International University

Categories

Bone Fracture

Licence