Using fuzzy logics to determine optimal oversampling factor for voxelizing 3D surfaces in radiation therapy

Voxelizing three-dimensional surfaces into binary image volumes is a frequently performed operation in medical applications. In radiation therapy (RT), dose-volume histograms (DVHs) calculated within such surfaces are used to assess the quality of an RT treatment plan in both clinical and research settings. To calculate a DVH, the 3D surfaces need to be voxelized into binary volumes. The voxelization parameters may considerably influence the output DVH. An effective way to improve the quality of the voxelized volume (i.e., increasing similarity between that and the original structure) is to apply oversampling to increase the resolution of the output binary volume. However, increasing the oversampling factor raises computational and storage demand. This paper introduces a fuzzy inference system that determines an optimal oversampling factor based on relative structure size and complexity, finding the balance between voxelization accuracy and computation time. The proposed algorithm was used to automatically calculate oversampling factor in four RT studies: two phantoms and two real patients. The results show that the method is able to find the optimal oversampling factor in most cases, and the calculated DVHs show good match to those calculated using manual overall oversampling of two. The algorithm can potentially be adopted by RT treatment planning systems based on the open-source implementation to maintain high DVH quality, enabling the planning system to find the optimal treatment plan faster and more reliably.