10/1 Reconstruction of 3D bone shape from 2D fluoroscopy images using statistical shape models

		Estimation of 3D shape from 2D fluoroscopy images.

3D models routinely used to provide image guidance and enhanced visualisation are typically extracted from 3D imagery like CT or MRI. To avoid the high radiation dose and costs associated with such scans, image free approaches are becoming popular especially in orthopaedic surgery. Without 3D imagery, building a 3D model that is specific to the patient anatomy is quite challenging. In our prior work from Co-Me phase 1 we proposed two novel methods based on manually digitised points, which are used to estimate the complete 3D bone shape. In this project, we aim at estimating bone shape from X-ray images. Currently there are no methods that solve the anatomical shape estimation from X-ray or fluoroscopic images with appropriate accuracy.

This project builds on the successful research in shape estimation in project 7 of Co-Me phase 1. Model-based shape estimation from 3D data has been successfully solved and the shape estimation from 2D data is the logical continuation of this work. Currently, routinely acquired X-ray images remain unused in the shape estimation and the estimation has thus to be computed intra-operatively. Shape estimation from 2D has the potential to be pre-operatively available, more efficient and less invasive. The methods developed in this project will be applied to hip (estimation of proximal femur and pelvis) and spine (vertebro/kyphoplasty) interventions.