UNLABELLED: Combining MRI morphological data with functional PET data offers significant advantages in research as well as in many clinical situations. Automatic methods are needed, however, to coregister the data from the two modalities. METHODS: Simulated PET images were created by simple and automatic segmentation of MR images followed by the assignment of different uptake values to various tissue types. The simulated PET images were registered to actual PET images using a pixel-by-pixel, PET-PET registration method. The transformation matrix was then applied to the MR images. The method was used to register MRI data to PET transmission scans and emission scans obtained with FDG, nomifensine and raclopride. Validation was performed by comparing the results to those obtained by matching internal points manually defined in both volumes. RESULTS: Emission and transmission PET images were successfully registered to MR data. Comparison to the manual method indicated a registration accuracy on the order of 1-2 mm in each direction. No difference in accuracy between the different tracers was found. The error sensitivity for the method's assumptions seemed to be sufficiently low to allow complete automation of the method. CONCLUSION: We present a rapid, robust and fully automated method to register PET and MR brain images with sufficient accuracy for most clinical applications.
UNLABELLED: Combining MRI morphological data with functional PET data offers significant advantages in research as well as in many clinical situations. Automatic methods are needed, however, to coregister the data from the two modalities. METHODS: Simulated PET images were created by simple and automatic segmentation of MR images followed by the assignment of different uptake values to various tissue types. The simulated PET images were registered to actual PET images using a pixel-by-pixel, PET-PET registration method. The transformation matrix was then applied to the MR images. The method was used to register MRI data to PET transmission scans and emission scans obtained with FDG, nomifensine and raclopride. Validation was performed by comparing the results to those obtained by matching internal points manually defined in both volumes. RESULTS: Emission and transmission PET images were successfully registered to MR data. Comparison to the manual method indicated a registration accuracy on the order of 1-2 mm in each direction. No difference in accuracy between the different tracers was found. The error sensitivity for the method's assumptions seemed to be sufficiently low to allow complete automation of the method. CONCLUSION: We present a rapid, robust and fully automated method to register PET and MR brain images with sufficient accuracy for most clinical applications.
Authors: Hannu Huhdanpaa; Darryl H Hwang; Gregory G Gasparian; Michael T Booker; Yong Cen; Alexander Lerner; Orest B Boyko; John L Go; Paul E Kim; Anandh Rajamohan; Meng Law; Mark S Shiroishi Journal: J Digit Imaging Date: 2014-06 Impact factor: 4.056
Authors: V V Rao; J L Dahlheimer; M E Bardgett; A Z Snyder; R A Finch; A C Sartorelli; D Piwnica-Worms Journal: Proc Natl Acad Sci U S A Date: 1999-03-30 Impact factor: 11.205
Authors: T Hershey; K J Black; J L Carl; L McGee-Minnich; A Z Snyder; J S Perlmutter Journal: J Neurol Neurosurg Psychiatry Date: 2003-07 Impact factor: 10.154