Fiche: Personalized 3D Bone Segmentation/

The promise of deep learning in medical image analysis is big. In order to prove the power of deep learning in medicine, Scyfer has developed an online service for bone segmentation of the hip and femur in 3D CT scans.

3D bone segmentation deep learning in medicine

The challenge

Firstly, 3D CT hip scans require manual processing in order to segment an accurate 3D representation of the bone structure. However, the tools currently available cannot deliver the required accuracy in the critical areas of the bone. Moreover, creating a useful segmentation requires manual correction of 50-200 images of a single 3D CT scan. This results in:

  • Tedious and time-consuming process
  • Non-reproducible outcomes (i.e. manual corrections differ each time)
  • Skilled resources required
  • Expensive solutions

Our solution

With this challenge in mind, Scyfer created an online bone segmentation service to segment and annotate the hip and femur bone. This then comes in use for medical research and post-processing of 3D CT scans. You just upload your DICOM file; we deliver a file where the different bones are segmented.

  • Provided as online service via API
  • Available for use in advanced simulation services
  • Available as post-processing service in scan devices
  • Also supports research projects

3D bone segmentation v2 deep learning in medicine

 

The benefits

Thanks to automatic classification, the benefits of 3D bone segmentation are 100% convenience and 100% service:

  • More accurate
  • Reproducible outcomes
  • Saves tedious work and time
  • Saves costs

The trained solution focuses on the area of interest: the intersection between the hip and femur. This is where we explicitly trained the algorithm to be highly accurate (90%-95%).

3D bone segmentation v3

Questions?

Then contact us.
Are you interested in our medical and research activities? We’re open for more projects on deep learning in medicine.
www.bonesegmentation.com.

EU flagThe work leading to these project results has received funding from the European Union Seventh Framework Programme FP7/2007-2013 under grant agreement no. 632913, is part of FICHe (Future Internet Challenge eHealth), and is realized with FIWARE software components.