Reference 3D-tumour models from CT-data for evaluation of contouring algorithm based on the PET/CT imaging

摘要

The method of Positron Emission Tomography (PET) is commonly used in nuclear medicine for the localization of tumours. Characterization of PET imaging systems is normally done with phantoms consisting of hollow spheres, filled with radioactive solution. These phantoms are not well suited for the evaluation of algorithms used to monitor a tumour response. This is mainly, because the inactive wall leads to quantification errors in the presence of background activity and because in vivo tumour structures deviate from a perfect sphere. A method is developed for the reproducible production of F-18 (radioactive marker) labelled tumour models from alginate, which have no inactive walls and are based on real CT tumour geometries. From CT data sets tumour geometries are segmented and converted into STL-files that describe the surface geometry. Based on these data different reusable negative moulds are manufactured by a rapid prototyping process. With these moulds the final tumour model is made in specially tailored alginate mixed with the desired concentration of F-18.