Physically Based Nonrigid Registration Using Smoothed Particle Hydrodynamics: Application to Hepatic Metastasis Volume-Preserving Registration

Recent advances in computing hardware have enabled the application of physically based simulation techniques to various research fields for improved accuracy. In this paper, we present a novel physically based nonrigid registration method using smoothed particle hydrodynamics for hepatic metastasis volume-preserving registration between follow-up liver CT images. Our method models the liver and hepatic metastasis as a set of particles carrying their own physical properties. Based on the fact that the hepatic metastasis is stiffer than other normal cells in the liver parenchyma, the candidate regions of hepatic metastasis are modeled with particles of higher stiffness compared to the liver parenchyma. Particles placed in the liver and candidate regions of hepatic metastasis in the source image are transformed along a gradient vector flow-based force field calculated in the target image. In this transformation, the particles are physically interacted and deformed by a novel deformable particle method which is proposed to preserve the hepatic metastasis to the best. In experimental results using ten clinical datasets, our method matches the liver effectively between follow-up CT images as well as preserves the volume of hepatic metastasis almost completely, enabling the accurate assessment of the volume change of the hepatic metastasis. These results demonstrated a potential of the proposed method that it can deliver a substantial aid in measuring the size change of index lesion (i.e., hepatic metastasis) after the chemotheraphy of metastasis patients in radiation oncology.