Compute the Optimal Threshold for PET Functional Volume Delineation
S Tan1*, W Chen2, W D'Souza2, W Lu2, (1) Huazhong University of Science and Technology, Wuhan, China, (2) University of Maryland School of Medicine, BALTIMORE, MDSU-E-J-111 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
To construct a method being able to compute the optimal threshold (OT) directly from PET images for functional volume delineation without any a priori information about scanners, imaging protocols and tumor size.
A mathematic model for partial volume effect (PVE) was analyzed for a spherical object with a constant activity imaged by a scanner with Gaussian point spread function (PSF). The theoretic analysis showed that information about some factors that are related to the PVE such as the PSF and tumor size had been actually embedded into PET images. We developed a new concept - the Optimal Area Ratio (OAR) that is a function of the object radius and can be derived from the model with some mathematic manipulations. With this new concept, information about PSF and tumor size can be recovered directly from PET images, and the OT can be calculated accordingly. The basic idea is to use the whole dynamic behavior (the behavior of the volume sizes due to different thresholds uniformly distributed in [0,1]) of a region-growing algorithm to find the OT. A computer simulation and experiments on phantoms with different object sizes and different noise levels were conducted to evaluate the above theoretic analysis.
For the computer simulation, the calculated OT was exactly the real OT. In the experiments on phantoms, the calculated OT led to visually satisfactory delineation results.
The new theoretic analysis showed that the OT can be calculated without any a priori information about scanners, imaging protocols and tumor size. The computer simulation and phantom experiments validated the theoretic analysis.