The Design And Implement Of Interactive Gross Tumor Volume Delineation System Based On Deformable Image Registration

This topic comes from the corporation between XXX and the Sun Yat-SenUniversity Cancer Center, Department of Radiation Oncology. In recent years,induction chemotherapy (IC) with concurrent chemoradiotherapy of IMRT(Intensity-modulated radiation therapy) has become an important issue for locallyadvanced nasopharyngeal carcinoma. Key to the successful implementation of IMRTis the proper delineation of target volume. It plays an important role in the prognosisof patients. The induction chemotherapy cause changes in tumor volume, tumorposition, and normal anatomy, making the identification and delineation of RT targetvolumes more complex and challenging. The GTV (gross tumor volume) delineationof IMRT become challenging. Currently, the doctor delineate the GTV all by hand,and their exist some problems.1) It cost the doctor a lot of time, and the delineation istedious and inefficient.2) The lack of reference makes it easy for the introduction oferror. A doctor may delineate the GTV differently at different time, and this affect theaccuracy of GTV. So, it is necessary to develop a system to help the doctor delineatethe GTV after induction chemotherapy, it can help the doctor reducing the time spendon delineation and reducing error.In this paper, we implement the interactive gross tumor volume delineationsystem which is based on deformable image registration to help the doctor delineatethe GTV after induction chemotherapy. It can help the doctor reducing the timespending on delineation and reducing error to some extent.The main work and contribution are as follows:1) We implements the interactivegross tumor volume delineation system which is based on deformable image registration. The system is developed by QT framework and based on DICOM3.0standard. The main function modules include: target delineation module, thedeformable registration module, spatial interpolation module, DICOM imagemanagement module, DICOM-RT structure set management module, imageenhancement module, MPR module.2）We design the registration module and use thedemons registration method to match the image pre-IC and post-IC. After registration,it will get the spacing transform between the image pre-IC and image post-IC. Wethen use the spacing transform to morph the pre-IC GTV to post-IC image. As a result,it will find the corresponding region of pre-IC GTV on post-IC image. The doctor canadjust the result, and complete the delineation according to the result. It can reducethe workload of doctors, and the doctor can take the result as a reference fordelineation.3) The interpolation module uses the spatial interpolation techniques todelineate the target according to the target of adjacent images. It can reduce theworkload of doctors.The system is designed according to the software engineering developmentprocess and specification. We use the objected oriented method to accomplishrequirements analysis, general design, and detailed design. Now, we haveimplemented a prototype of the interactive gross tumor volume delineation systemwhich is based on deformable image registration. The system can meet therequirements and can interchange data with the treatment planning system of hospital,and can assistant the doctor delineating the target volume. It can also be used astraining platform for new doctors to learn the target delineation, can also be used as aplatform for explore and research medical image registration algorithm.