Meeting Banner
Abstract #1207

Automated Registration of Multiple Single Breath-hold Cardiac MRI Images

Mehta T, Thomson L, Fieno D, Goyal V, Nishina H, Slomka P, Ramesh A, Gerlach J, Berman D, Germano G
Cedars-Sinai Medical Center

Cine cardiac MRI (cMRI) images are typically acquired as multiple slices with one slice position and/or orientation per breath-hold. These multiple images are often misaligned due to patient motion. We developed a novel image processing technique, which allows registering multi breath-hold short-axis (SA), 2-chamber (2CH), and 4-chamber (4CH) cMRI views into common 3D spatial coordinates. Studies from 20 consecutive patients (10 males, 10 females) undergoing cMRI exams, with 2CH, 4CH, and SA cine (20 phases), each acquired in separate breath-holds, with pixel size 1.3-1.8 mm were analyzed retrospectively. 2CH, 4CH and SA image planes were brought into common 3D coordinates by dynamic profile registration technique. The overall accuracy of the automatic algorithm was 1.00.9 mm for the 2CH to 4CH registration, and 1.11.0 mm, 1.11.2 mm, and 2.01.8 mm for SA registration in midventricular, basal and apical regions respectively. The alignment algorithm can be used as a completely automated and accurate p