Bringing Your Own X-Ray DDH DICOMs

This guide will show you how to analyze your own X-Ray DICOMs using Retuve.

Install Retuve's AI Plugins

As of 2025-06-01, Retuve does not include default AI plugins. You will need to install an external plugin. The following plugins are available:

Note: All plugins are under their own licenses. Please check the license of each plugin before using it.

• retuve-yolo-plugin

For this guide, we will use the retuve-yolo-plugin, which has the version 2 x-ray system.

We highly recommend reading how version 2 was trained and validated here: https://github.com/radoss-org/retuve-yolo-plugin?tab=readme-ov-file#update---x-ray-version-2

Install it with:

pip install git+https://github.com/radoss-org/retuve-yolo-plugin

Creating a Minimal Script

You only need to make a few changes to the example script to run the AI on your own data.

1. Import the X-Ray Plugin and pydicom

from retuve_yolo_plugin.xray_v2 import yolo_predict_dcm_xray
import pydicom

2. Load Your Own DICOM File

Replace the example code that loads a sample image and labels:

img_file, labels_json = download_case(Cases.XRAY_JPG)
img_raw = Image.open(img_file)
labels = json.load(open(labels_json))

with:

dcm_file = "path/to/your/dicom/file.dcm"
dcm = pydicom.dcmread(dcm_file)

You can remove all references to labels_json—it is not needed for the AI plugin.

3. Update the analyse_hip_xray_2D Function Call

Make sure you use the DICOM data and the plugin function:

hip_data, img, dev_metrics = analyse_hip_xray_2D(
    dcm,
    keyphrase=default_xray,
    modes_func=yolo_predict_dcm_xray,
    modes_func_kwargs_dict={},
)

Example Scripts

Here is a complete example script for analyzing a single DICOM file:

import pydicom
from retuve_yolo_plugin.xray_v2 import yolo_predict_dcm_xray
from retuve.defaults.hip_configs import default_xray
from retuve.funcs import analyse_hip_xray_2D

# Example usage
dcm_file = "path/to/your/dicom/file.dcm"
dcm = pydicom.dcmread(dcm_file)

hip_data, img, dev_metrics = analyse_hip_xray_2D(
    dcm,
    keyphrase=default_xray,
    modes_func=yolo_predict_dcm_xray,
    modes_func_kwargs_dict={},
)

img.save("xray.jpg")

metrics = hip_data.json_dump(default_xray, dev_metrics)
print(metrics)

To process multiple DICOM files, use a loop:

import os
import pydicom
from retuve_yolo_plugin.xray_v2 import yolo_predict_dcm_xray
from retuve.defaults.hip_configs import default_xray
from retuve.funcs import analyse_hip_xray_2D

dicom_files = [
    "path/to/your/dicom/file1.dcm",
    "path/to/your/dicom/file2.dcm",
    # Add more files as needed
]

for dcm_file in dicom_files:
    dcm = pydicom.dcmread(dcm_file)

    hip_data, img, dev_metrics = analyse_hip_xray_2D(
        dcm,
        keyphrase=default_xray,
        modes_func=yolo_predict_dcm_xray,
        modes_func_kwargs_dict={},
    )

    img.save(os.path.splitext(dcm_file)[0] + "_xray.jpg")

    metrics = hip_data.json_dump(default_xray, dev_metrics)
    print(metrics)

Troubleshooting

• Plugin not found: Make sure you have installed the plugin as described above.
• DICOM files not detected: Ensure your files are in the correct location and are valid DICOM files.

Further Reading

• Retuve Documentation
• YOLO Plugin for Retuve

Note: All plugins are under their own licenses. Please check the license of each plugin before using it.

If you need to process individual files or want more control, see the minimal script guide or the example in the other part of this codebase.