Generation and performance of patient-specific forward models for breast imaging with EIT

It has now been well established that accurate geometric conformity of the forward model for EIT reconstruction has significant benefits for artefact reduction and localisation of conductivity changes within the domain. The problems of generation of patient specific forward models need to be addressed as segmentation of volumetric data from CT or MRI is inadequate for time-critical clinical use. This group has pioneered methods of generating patient-specific surface models from known landmarks and electrode positions and have used this data to warp finite element models for EIT reconstruction. This paper presents a further application of these methods to use known electrode positions for breast imaging to generate an accurate B-Spline surface model of a subject and to warp an existing finite element model to the surface using elastic deformation. Results will show that a forward model can be generated, conforming more realistically to actual subject geometry, that will further enhance the performance of the reconstruction algorithm offering significant benefits to clinical EIT breast imaging.