A line-profile based double partial fusion method for acquiring planning CT of oversized patients in radiation treatment

Abstract

True 3D CT dataset for treatment planning of an oversized patient is difficult to acquire due to the bore size and field of view (FOV) reconstruction. This project aims to provide a simple approach to reconstruct true CT data for oversize patients using CT scanner with limited FOV by acquiring double partial CT (left and right side) images. An efficient line profile-based method has been developed to minimize the difference of the CT numbers in the overlapping region between the right and left images and to generate a complete true 3D CT dataset in the natural state. New image processing modules have been developed and integrated to the Insight Segmentation & Registration Toolkit (ITK 3.6) package. For example, different modules for image cropping, line profile generation, line profile matching, and optimized partial image fusion have been developed. The algorithm has been implemented for images containing the bony structure of the spine and tested on 3D CT planning datasets from both phantom and real patients with satisfactory results in both cases. The proposed optimized line profile-based partial registration method provides a simple and accurate method for acquiring a complete true 3D CT dataset for an oversized patient using CT scanning with small bore size, that can be used for accurate treatment planning.