Meeting Banner
Abstract #0770

On the Reproducibility of MR-Based PET Attenuation Correction Using a Probabilistic Atlas-Based Method

Kevin T. Chen1, 2, Daniel B. Chonde, 23, Ciprian Catana4

1Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, MA, United States; 2Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States; 3Biophysics, Harvard University, Boston, MA, United States; 4Massachusetts General Hospital, Charlestown, MA, United States

We have previously implemented an MR-based PET attenuation correction (AC) method using dual-echo ultrashort echo time (DUTE) and morphological MR images to segment the tissue classes most relevant for attenuation correction (i.e. bone, soft tissue and air cavities). Attenuation maps generated from these segmented images were previously demonstrated to agree well with those generated using the sliver standard segmented CT method. In this work, the reproducibility of the probabilistic atlas-based method for generating the attenuation maps was studied by comparing the maps obtained from the MR data acquired in nine subjects at three time points.

Keywords

account acquisition agree altered analyzed assigned assigning atlas attenuation available belong bias biomedical biophysics black bone bore cavities chronological class classes classifier close coefficients comparing compatible completely compute consistently constitute correction demonstrating denotes described determine development dice dividing dual equipped even every examined exist expected finally frontal generated generation grant gray head health identified indicate indicated indicates inserted institute integrated inter intersection interval introduced investigated known lacs largest lesser linear locations longest made magnet majority maps mask materials measure medical might minimal misclassified mixture morphological morphology near nine occurred orientations original overlap overlapping particular patient plot plots previously probabilistic probability procedures produced produces proportion prototype quantitative radiology recently registered relevant representative reproducibility reproducible respectively scanned scanner scanners sciences segment segmented segmenting select selected shades similarity since sinus skull sliver soft solution source still structural subject subjects suggesting technology third though tissue trained training transverse ultrashort units visit visits volume white wise