New algorithm to simulate organ movement and deformation for four-dimensional dose calculation based on a three-dimensional CT and fluoroscopy of the thorax

Purpose: The aim of this study was to develop a 4D-modeling algorithm, designated "3D+," to simulate organ movement and deformation for 4D dose calculation without the need for 4D-CT or deformable image registration and to assess the validity of this algorithm. Methods: This 3D+ algorithm virtually creates 4D-CT images by deforming static 3D-CT data according to a typical motion model and motion data at multiple observation points collected via fluoroscopy. A typical motion model intended for patients with lung tumors immobilized with a vacuum pillow inside a stereotactic body frame was constructed. The geometric accuracy of virtual 4D-CT images created using this 3D+ algorithm was evaluated in eight patients by comparing the simulated results with actual 4D-CT images in terms of visual assessment, landmark analysis, and comparison of the radial distance from the tumor centroid to the body or lung surface. Results: The average accuracy for all patients, as determined via landmark analysis, was 2.8 +/- 1.8 mm, very similar to results obtained through 4D-CT and deformable image registrations. Error in the radial distance from the tumor centroid to the body or lung surface was generally within 1.0 or 2.0 mm, respectively, in virtual versus actual 4D-CT images. Therefore, it is assumed that these geometric errors would have only negligible effects on dose calculation. Conclusions: 4D modeling of the thorax utilizing the 3D+ algorithm shows acceptable accuracy and is more suited for routine clinical use in terms of processing time than conventional 4D-CT and deformable image registration. The 3D+ algorithm may be useful for simulating dose distribution for advanced beam delivery techniques, such as real-time tumor tracking irradiation and adaptive radiation therapy. (C) 2009 American Association of Physicists in Medicine. [DOI: 10.1118/1.3213083]