Feasibility of non-linear image registration for radiotherapy treatment planning with gamma camera PET

Abstract

This paper describes our initial experiences of radiotherapy treatment planning with gamma camera PET, in particular the application of non-linear image registration. Preliminary investigations have involved patients with non-small cell lung cancer who first underwent 18F-FDG gamma camera PET and CT with a GE Millennium VG Hawkeye system and later received Philips AcQSim virtual simulation CT for radiotherapy treatment planning. Firstly, the Hawkeye CT is aligned to the planning CT using a novel method of automatic non-linear registration, which provides a robust pseudo-elastic deformation and incorporates a constraint that ensures the transformation between the two images is smooth and continuous. Secondly, since the gamma camera PET and Hawkeye CT examinations are carried out sequentially, PET images are well registered to the Hawkeye CT. Therefore, the Hawkeye CT to planning CT image transformation derived by the non-linear algorithm can be applied directly to the gamma camera PET to provide the necessary PET to planning CT registration. The registered PET images can be imported into virtual simulation and treatment planning systems and displayed alongside the original planning CT. Radiotherapy clinicians can then utilize the additional functional information when delineating relevant structures. Comparison with conventional affine registration demonstrates the significant contribution of the non-linear algorithm, particularly for cases with pronounced differences in patient positioning between the PET imaging and planning CT.