Type of Document	Master's Thesis
Author	Manoharan, Rajesh
Author's Email Address	rmanoh1@lsu.edu
URN	etd-11052004-143204
Title	Validation of PET/CT Dataset for Radiation Treatment Planning
Degree	Master of Science (M.S.)
Department	Physics & Astronomy
Advisory Committee	Advisor Name Title Kenneth L. Matthews II Committee Chair Steven Bujenovic Committee Co-Chair Erno Sajo Committee Member Juhan Frank Committee Member Oscar Hidalgo-Salvatierra Committee Member
Keywords	image registration image fusion PET/CT
Date of Defense	2004-09-02
Availability	unrestricted
Abstract PET/CT scans are frequently used for radiation treatment planning (RTP). Our work demonstrates a practical approach for validating the PET/CT dataset for RTP. We tested this QA process on a Reveal HD PET/CT scanner. The phantom used is a TGM2 ISIS QA phantom, a 14 cm acrylic cube with a central bore for object inserts. It has four different built-in inserts for electron density verification. 22Na seeds are inserted into the pinholes at the side of the cube. PET/CT images of the phantom with 22Na seeds are acquired and fused in the scanner Syngo fusion software. Registration of the PET/CT dataset is visualized by raising the lower threshold of the PET images to reduce the 22Na point sources to a few pixels and comparing it with the CT images of 22Na seeds. Geometric scaling accuracy of the pixels is verified by measuring the dimension of the cube in x, y and z axes. The HU values of four electron density verification inserts are measured and compared with manufacturer specified HU values. These QA tests are repeated in the RTP software after importing the PET/CT dataset. A quantitative analysis of registration error and geometric scaling accuracy of pixels are verified independently using MATHEMATICA. The resolution of the PET scanner was determined by measuring the FWHM of capillary tube sources inserted in a Styrofoam block based on the NEMA-2 protocol. Minor misalignment of the fused images was detected in the scanner (~1 mm) while the imported dataset in the RTP system showed a major misalignment (~6 mm) when fused by auto fusion software. The maximum geometric scaling errors of object sizes were observed in the z direction (5.2% decrease) in the scanner and the scaling errors were less in the RTP software (2.9% decrease). The greatest HU errors in the CT image compared with expected HU values were observed in the bone density insert (28% increase) in the scanner and all HU values for different inserts were shifted up by a constant value in the RTP system. The resolution of the PET scanner was comparable to the manufacturer’s specification.
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    Manoharan_thesis.pdf 1.32 Mb 00:06:05 00:03:08 00:02:44 00:01:22 00:00:07