Fracture Check

Fracture Check

Screens hand, arm, leg, hip and shoulder X-rays for possible fractures, shows a heatmap of where the model looked, and marks the likely break. A student research project for learning, not for medical decisions.

Made by akinchanJain7 on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: Fracture Check screens X-rays of the hand, arm, leg, hip or shoulder for signs of a fracture. It estimates whether a break is present, draws a heatmap of where the model looked, and marks the most likely spot. The same page holds notes on how the output is read and a general care guide for fractures.

Who it suits: It suits students and hobbyists who want to see how a medical imaging model works from upload to result. Anyone learning about explainable image models can use it as a worked example.

Industry | Who uses it | How they use it | Example
Education | Computer science students | Study how a classifier, a heatmap and a detector fit together | Compare the heatmap with the marked box on one X-ray
Research | Medical imaging researchers | Look at a small public-dataset baseline | Read the test results next to the model card
Teaching | Instructors | Show an explainable model live in a browser | Upload an X-ray during a class demo
Health education | Curious readers | Read the care guide and the reading help | Learn what the result screen means

Anyone who needs a real diagnosis should skip it, since only a clinician reading the film counts.

Where to use it: You can run it in two places.

• A web browser, on the Space's page at Hugging Face
• Windows, macOS or Linux, as a local copy run with Python 3.12

Use case | Who | What they get back
Fracture estimate | Students | An estimate of whether the X-ray shows a fracture
Heatmap | Learners | A picture of the area the model focused on
Break marker | Researchers | A box placed on the most likely break
Care guide | Readers | General notes on fracture care and on reading the result

How to set it up: The online page needs nothing installed. A local copy needs the following.

• Python 3.12
• Windows, macOS or Linux
• The Python packages listed in requirements.txt

1. Install Python 3.12.
2. Install the Hugging Face hub package and download a copy of the Space.
3. Move into the new folder and install Gradio 6.31.0 and the packages in requirements.txt.
4. Start the app with python app.py.

pip install -U huggingface_hub
hf download akinchanJain7/fracture-check --repo-type space --local-dir fracture-check
cd fracture-check
pip install gradio==6.31.0
pip install -r requirements.txt
python app.py

How to use it: Upload one X-ray showing a hand, an arm, a leg, a hip or a shoulder, then read the estimate, the heatmap and the marked box. On a local copy, this command starts the page.

python app.py

Tool | What it does
ConvNeXt-Tiny classifier | Estimates whether the X-ray shows a fracture
Grad-CAM heatmap | Shows where the classifier looked
Faster R-CNN detector | Marks where the break most likely is
Care guide | Explains how to read the result and gives general fracture care notes

Example: You upload an X-ray of a hand. The page answers with the classifier's estimate of whether a fracture is present, a heatmap of the area it focused on, and a box on the likely break.

For newcomers and experts: If you are new to this, open the Space in a browser and upload one X-ray. Look at the heatmap first, then at the box, and see whether they point to the same area.

If you know the field, the stack is a ConvNeXt-Tiny classifier with Grad-CAM for explanation and a Faster R-CNN detector for localization, trained on the FracAtlas dataset. In a one-time test on 589 X-rays, 102 of them fractured, the classifier found 94 of the fractures. It also cleared 71% of the X-rays with no fracture. In 75% of the fractured X-rays, the detector's first box sat on the fracture.

Upload an X-ray
ConvNeXt-Tiny estimates a fracture
Grad-CAM draws a heatmap
Faster R-CNN marks the likely break
Page explains the result

Good to know: It is a student research project and not a medical device. Do not use it to decide whether to seek care.

• License: MIT
• Cost to run: none named
• Data it reaches: the X-ray you upload, and at start-up app.py fetches its code from GitHub and its models from the model repo
• Not for: deciding whether to seek care, or replacing a doctor's reading of an X-ray
• Dataset: FracAtlas, shared under CC BY 4.0

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: gradio
• License: mit

Links

• Space: https://huggingface.co/spaces/akinchanJain7/fracture-check
• Files: https://huggingface.co/spaces/akinchanJain7/fracture-check/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.