Intermodality, retrospective image registration in the thorax.

Abstract

The purpose of this study was to develop an accurate, retrospectively applicable procedure for registering thoracic studies from different modalities in a short amount of time and with minimal operator intervention. CT and PET studies were acquired from six patients. The pleural surfaces in both image sets were determined by segmenting based on 50% of the maximum soft-tissue value in the study. These surfaces were converted into three-dimensional volumes and used to register the CT and PET studies in three dimensions using a sum of least squares fitting approach. The registered PET study was then displayed in a hot metal scale overlayed on top of the gray scale CT study. The accuracy of the fit was evaluated through a phantom study and preliminary clinical evaluation. A phantom study was performed to determine the limits of this technique. The accuracy was determined to be less than 2.3 mm in the x and y direction and 3 mm in the z direction. Preliminary clinical evaluation was also performed with encouraging results. This technique accurately registers PET and CT images of the thorax, retrospectively, without the need for external fiducial markers or other a priori action.