Automated Fractured Bone Segmentation and Labeling from CT Images

Area of Science:

• Medical Imaging
• Radiology
• Biomedical Engineering

Background:

• Accurate segmentation of fractured bones from Computed Tomographic (CT) images is crucial for trauma analysis, diagnosis, and surgical planning.
• Current methods can be labor-intensive and may struggle with artifacts, impacting the precision of injury assessment and treatment strategies.

Purpose of the Study:

• To introduce an automated and accurate technique for segmenting fractured bones from CT images.
• To simplify the labeling process through patient-specific anatomical labeling.
• To enhance segmentation by removing unwanted artifacts like soft tissue.

Main Methods:

• A novel segmentation technique was developed, assigning unique labels based on patient-specific anatomy.
• The method incorporates artifact removal, specifically targeting unwanted elements such as flesh.
• Experiments were conducted using real-world CT data and compared against state-of-the-art techniques.

Main Results:

• The proposed segmentation technique achieved an accuracy of 95.45% on real-world data.
• Validation was performed against expert-based clinical ground-truth, confirming the method's reliability.
• The technique demonstrated superior performance compared to existing state-of-the-art methods.

Conclusions:

• The developed segmentation technique offers an accurate and efficient solution for fractured bone identification in CT images.
• This method facilitates improved severity analysis, 3D visualization, and optimal planning of recovery processes.
• A dataset of 8000 CT images will be made available upon request to support further research.