Volume-Preserving Deformable Registration Method for Temporal Change Analysis on Medical Images

Abstract

Lung cancer screening for identifying early lung cancer has been achieved by follow-up computed tomography (CT) techniques. In lung cancer screening,benign and malignant nodules can be classified through nodule growth assessment by measuring temporal changes in the volume and intensity of nodules in succeding follow-up CT scans. Various automatic or semi-automatic nodule segmentation methods have been developed for nodule volumetry and growth assessment for solid nodules. However, ground glass opacity (GGO)nodules are very elusive to automatically segment due to their inhomogeneous interior. The nodule growth assessment of GGO nodules can be achieved by the subtraction after registration between follow-up CT scans. During the registration, the volume of nodule regions in the floating image should be preserved, whereas the volume of other regions in the floating image should be aligned to that in the reference image. Several constraints for preserving the nodule volume have been proposed in the analysis of nodule changes that occur between reference and floating images. However, they are not appropriate for GGO nodules due to the lung respiratory motion and the difficulty of nodule segmentation. In this dissertation, we propose an accurate and fast deformable registration method. It applies the volume-preserving constraint to candidate regions of GGO nodules, which are automatically detected by gray-level cooccurrence matrix (GLCM) texture analysis. Considering that GGO nodules can be characterized by their inner inhomogeneity and high intensity, we identify the candidate regions of GGO nodules based on the homogeneity values calculated by the GLCM and the intensity values. Furthermore, we accelerate our deformable registration by using Compute Unified Device Architecture (CUDA). In the deformable registration process, the computationally expensive procedures of the floating image transformation and the cost function calculation are accelerated by using CUDA. The experimental results demonstrated that our method almost perfectly preserves the volume of GGO nodules in the floating image as well as effectively aligns the lung between the reference and floating images. Regarding the computational performance, our CUDA-based method delivers about 45.9× faster registration than the conventional method. Our method can be successfully applied to a GGO nodule follow-up study and can be extended to the volume-preserving registration and subtraction of specific diseases in other organs.