New approach based on tetrahedral-mesh geometry for accurate 4D Monte Carlo patient-dose calculation

• Published in: Physics in medicine & biology 2015-02, Vol.60 (4), p.1601-1612
• Main Authors: Han, Min Cheol, Yeom, Yeon Soo, Kim, Chan Hyeong, Kim, Seonghoon, Sohn, Jason W
• Format: Article
• Language: English
• Subjects: 4D Monte Carlo; 4D tetrahedral phantom; Algorithms; deformation vector fields; dose calculation; Four-Dimensional Computed Tomography - methods; Monte Carlo Method; Phantoms, Imaging; Radiation Monitoring - methods; Radiotherapy Planning, Computer-Assisted - methods; tetrahedral patient model
• ISSN: 0031-9155; 1361-6560
• DOI: 10.1088/0031-9155/60/4/1601

In the present study, to achieve accurate 4D Monte Carlo dose calculation in radiation therapy, we devised a new approach that combines (1) modeling of the patient body using tetrahedral-mesh geometry based on the patient's 4D CT data, (2) continuous movement/deformation of the tetrahedral patient model by interpolation of deformation vector fields acquired through deformable image registration, and (3) direct transportation of radiation particles during the movement and deformation of the tetrahedral patient model. The results of our feasibility study show that it is certainly possible to construct 4D patient models (= phantoms) with sufficient accuracy using the tetrahedral-mesh geometry and to directly transport radiation particles during continuous movement and deformation of the tetrahedral patient model. This new approach not only produces more accurate dose distribution in the patient but also replaces the current practice of using multiple 3D voxel phantoms and combining multiple dose distributions after Monte Carlo simulations. For routine clinical application of our new approach, the use of fast automatic segmentation algorithms is a must. In order to achieve, simultaneously, both dose accuracy and computation speed, the number of tetrahedrons for the lungs should be optimized. Although the current computation speed of our new 4D Monte Carlo simulation approach is slow (i.e. ~40 times slower than that of the conventional dose accumulation approach), this problem is resolvable by developing, in Geant4, a dedicated navigation class optimized for particle transportation in tetrahedral-mesh geometry.