Image Fusion Based on Estimation Theory: Applied to PET/CT for Radiotherapy
Jinzhong Yang, Rick S. Blum, Peter Balter and Laurence E. CourtAffiliation:
1515 Holcombe Blvd, Unit 94, Houston, TX 77030, USA.
AbstractThis paper reviewed three state-of-the-art image fusion methods that were developed based on the estimation theory and evaluated these methods in the fusion of PET/CT images for radiotherapy applications. These fusion methods were developed in a framework of maximum likelihood estimate and firstly introduced the expectation-maximization algorithm to image fusion at either pixel-level or feature-level. Some recent patents on similar image fusion approaches have been discussed. The estimation theory based methods were previously evaluated for the fusion of visual and infrared images, however, they have not been tested for fusion of medical images such as PET/CT images. In this study we demonstrated through experiments the potential applicability of the pixel-level fusion and region-level fusion approaches based on the EM algorithm for PET and CT image fusions. We have shown that the fused image might be useful for tumor target delineation and image-guided radiotherapy.
Estimation theory, expectation-maximization algorithm, hidden Markov model, image fusion, image-guided radiotherapy, PET/CT.
Download Free Order Reprints Order Eprints Rights and Permissions