README.md

Back to Projects List

Automatic Landmark Identification in Cranio-Facial CBCT

Key Investigators

• Luc Anchling (UoM)
• Nathan Hutin (UoM)
• Maxime Gillot (UoM)
• Baptiste Baquero (UoM)
• Jonas Bianchi (UoM, UoP)
• Marcela Gurgel (UoM)
• Najla Al Turkestani (UoM)
• Marilia Yatabe (UoM)
• Lucia Cevidanes (UoM)
• Juan Prieto (UoNC)

Project Description

We propose a novel approach that reformulates anatomical landmark detection as a classification problem through a virtual agent placed inside a 3D Cone-Beam Computed Tomography (CBCT) scan. This agent is trained to navigate in a multi-scale volumetric space to reach the estimated landmark position. The agent movements decision relies on a combination of Densely Connected Convolutional Networks (DCCN) and fully connected layers.

Objective

1. Retrain the different models with new data
2. Do some maintenance on the previously made code

Approach and Plan

1. Use the available code to train with additional patient data for each landmarks

Progress and Next Steps

1. ALI models are currently being retrained with new data

Illustrations

Background and References