Virtual Bone Shape Aging

Francesco Caliva, Alejandro Morales Martinez, Sharmila Majumdar, Valentina Pedoia

Keywords: Bone Shape, Osteoarthritis, Spherical Encoding

TL;DR: We use deep learning to age knee bone surfaces four years.

Abstract: We use deep learning to age knee bone surfaces four years. We propose to encode an MRI-based bone surface in a spherical coordinate format, and use these spherical maps to predict shape changes in a 48 months time frame, in subjects with and without osteoarthritis. The experiments show that a 2D V-Net can predict bone surface shape with a mean absolute error of about 1 mm. Our code is available at https://github.com/fcaliva/Bone_Shape_Virtual_Aging.

Paper Type: both

Primary Subject Area: Detection and Diagnosis

Secondary Subject Area: Application: Radiology

Paper Status: original work, not submitted yet

Source Code Url: The code is available at https://github.com/fcaliva/Bone_Shape_Virtual_Aging/

Data Set Url: The Osteoarthritis Initiative (OAI) is a multi-center, ten-year observational study of men and women, sponsored by the National Institutes of Health (part of the Department of Health and Human Services). The goals of the OAI are to provide resources to enable a better understanding of prevention and treatment of knee osteoarthritis, one of the most common causes of disability in adults. https://nda.nih.gov/oai/

Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.