Model Card: YOLOv8-Bone-Fracture-Detection

Model Description:

Detect and recognize bone fractures, implants, and other abnormalities in X-ray images with bounding box localization and label output.

Fine-tuned by: Musa Yilmaz / Open Institue of Technology

Model type: Object Detection & Recognition

Language(s):

Python: 99.3%

Other: 0.7%

License: GNU AFFERO GENERAL PUBLIC LICENSE Version 3, 19 November 2007

Finetuned from model: YOLOv8

Repository: https://github.com/yilmazmusa08/YOLOv8-Bone-Fracture-Detection

Uses

Intended Use:

Assist radiologists and orthopedic surgeons in the detection and diagnosis of bone fractures and abnormalities in X-ray images.

Aid in treatment planning by providing accurate localization and identification of fractures, implants, and other pathologies.

Educational purposes to train medical students and residents in the interpretation of X-ray images.

Out-of-Scope Use:

Unauthorized surveillance or invasion of privacy.

Non-medical purposes.

Ethical Considerations:

Data Bias: Model performance might vary depending on the diversity and representativeness of the training data.

Privacy Concerns: X-ray images might contain sensitive patient information.

Misuse Potential: Risk of misuse if the model is used for unauthorized purposes.

How to Get Started with the Model

Go to /App directory, use command:

python streamlit run app.py

Training Details:

The training data originates from the "Bone Fracture v2" dataset available on Roboflow, accessible through the link provided.

Dataset link: https://universe.roboflow.com/capjamesg/bone-fracture-v2/dataset/3

Train Set: 1630 Images (71%)

Validation Set: 440 Images (19%)

Test Set: 220 Images (10%)

Evaluation

Performance Metrics:

Accuracy (Detection): 80%

Accuracy (Recognition): 65%

Citation:

BibTeX:

@software{jocher2023ultralytics,

author = {Jocher, Glenn and Chaurasia, Ayush and Qiu, Jing},

title = {Ultralytics YOLO},

version = {8.0.0},

date = {2023-1-10},

url = {https://github.com/ultralytics/ultralytics}}

APA:

Jocher, G., Chaurasia, A., & Qiu, J. (2023). Ultralytics YOLO (Version 8.0.0) [Software]. Retrieved from https://github.com/ultralytics/ultralytics