Session S62.6

A Semi-Automatic Software Package for Analysis of Dynamic Contrast-Enhanced MRI Myocardial Perfusion Studies

NA Pack*, S Vijayakumar, TH Kim, CJ McGann, EVR DiBella

University of Utah
Salt Lake City, UT, USA

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is used for the detection and assessment of coronary artery disease. Typically with DCE-MRI, 3-5 short-axis slices of the left ventricle (LV) are imaged every heartbeat during adenosine stress, to follow the uptake of a gadolinium-based contrast agent. While ischemia may be detected visually from the MR images by trained cardiologists or radiologists, it is likely that semi-quantitative or quantitative analysis of the dynamic images can improve the accuracy of diagnoses. Such analyses have yet to be standardized and can be limited by the availability and capability of existing software. This work aims to provide a modular Matlab-based software solution that can improve the analysis of DCE cardiac perfusion images.
The software, MPI2D (Myocardial Perfusion Imaging 2D), can be run from the command line or with a graphical user interface and takes DICOM perfusion images as input. The software has stages for registration, segmentation, fitting, and display of results. The registration stage is used to compensate for respiratory motion and can be manual or semi-automatic. Once the images are registered, the user manually segments the endocardial and epicardial borders of the LV from a single image frame. The segmented LV is automatically divided into a user-defined number of radial and circumferential segments. The change in signal intensity curves are fit to a pharmacokinetic model. Currently implemented choices include a 2-compartment model, the Fermi function model, model-independent analysis, Patlak plot analysis, and maximum upslope analysis.
MPI2D has been used to analyze 14 adenosine stress and rest studies acquired at the University of Utah. The perfusion estimates were similar to published literature values and were not significantly different between the different models, except for the Fermi model at stress. One observer processed four of these low-dose datasets with MPI2D on separate occasions to assess the intra-observer variability of perfusion estimates in AHA-defined regions. There was good correlation between perfusion estimates (y=1.00x-0.02, r=0.98). MPI2D is freely available by request for research purposes. This myocardial perfusion software analysis package allows for semi-automatic processing of the data to estimate kinetic parameters with relatively low intra-observer variability.
This project was supported by NIH R01 EB000177.

(Abstract Control Number: 205)