2017 Winter Project Week/An open-source tool to classify TMJ OA condyles

Jump to: navigation, search
Home < 2017 Winter Project Week < An open-source tool to classify TMJ OA condyles

Key Investigators

  • Priscille de Dumast (University of Michigan)
  • Juan Carlos Prieto (University of North Carolina)
  • Beatriz Paniagua (Kitware, Inc.)

Project Description

Objective Approach and Plan Progress and Next Steps
  • Develop a Neural Network to classify 3D osteoarthritic morphological variations using 3D models
  • Develop a Slicer extension to disseminate that methodology
  • Data preprocess : Point distribution models (PDM) with 1002 correspondent points constructed using SPHARM-PDM software, features extraction from the shapes
  • Implementation of a deep learning network using Tensorflow and create classification using the features extracted before
  • Display the actual network in tensorboard
  • Combined with the web-based system from another project, the network is usable from a Slicer extension with remote computations
  • Design ideas for an standalone Slicer extension

Background and References

Features extraction source code
Classification source code