Meeting Banner
Abstract #1338

Assessing the Effects of Vessel Segmentation Boundary Size on Flow Quantification Error and Comparing Multiple Automatic Segmentation Algorithms

Paul Kokeny1, Jing Jiang2, 3, Ewart Mark Haacke2, 3

1Biomedical Engineering, Wayne State University, Detroit, MI, United States; 2Radiology, Wayne State University, Detroit, MI, United States; 3MRI Institute for Biomedical Research, Detroit, MI, United States


A mathematical analysis of the sources of error present in PC-MRI was performed in order to study the effects that vessel segmentation boundary size has on flow quantification error. The analysis concluded that an undersized boundary, missing vessel pixels, will result in greater flow error than an oversized boundary, with partial volume error. Four automatic vessel segmentation algorithms were tested and compared using cine PC-MRI simulations that mimic the cardiac cycle. The results from this testing supported the theory presented and demonstrated that one of the algorithms stood out as a fast and accurate way to quantify flow.