Registration Technology Based On Multimodality Image Of Bone Tumor And Employment

Bone tumor is a kind of common tumor disease which happens in bone or bone attachments(such as blood vessel,nerve,medulla).Bone tumor can be divided into two classes: benign tumor and malign tumor.Benign bone tumor can be completely cured and has a good prognosis. Malign tumor deteriorates rapidly ,has a poor prognosis and often causes death to the patient.Removal of tumors is the best cure method.The remains of tumor often cause recrudescence of tumor and make bone tumor deteriorate.To discover bone tumor early and confirm the location and bondary of bone tumor is very important to treatment.The object of our research subject is bone tumor locating on vertebra. X-ray picture has diffculty in display of clavicle, scapula,vertebra, basilar skull for their complex structure and ovelap of tissues.We have a registration of Positron Emission Tomography(PET)/ Computer Tomography(CT)/ Nuclear Magnetic Resonance Imaging(MRI) images of bone tumor locating on vertebra to diagnose it. PET has a reflection of biological metabolizability of body, represents the biochemical, physiological and pathological conditions, and displays the boundary and bulk of bone tumor. CT has advantage over X-ray picture in display symptom of bone tumor for its transsect imaging and high resolution power and can display the boundary of tumor in bone structure.MRI can image any slice and has a high resolution power in soft tissue.MRI can check if tumor invades the neighbor tissues such as fat, muscle and other viscera or not and confirms the boundary of tumor in soft tissue.It is convenient to discover bone tumor and confirm the location and boundary of it by combination of information of PET/CT / MRI in diagnosis of bone tumor.We take a registration of CT and MRI images first and have a fusion of PET and the registration results then. We choose B spline as transform method for its high performance as a nonrigid transform method and local control character.Because CT and MRI are not the same modality, we choose mutual information method basing on probability as measurement criterion.The objective evaluation of registration results with joint entropy and the"double blind"evaluation by senior doctors lead to a conclusion that registration results are good. We develop a platform for registration mainly by integrating ITK and VTK in C#. The registration of PET/CT/MRI images for fusion assisting diagnosis has not been covered yet.