Fully Automatic Segmentation of the Myocardium in Cardiac Perfusion MRI

Reduced myocardial perfusion is an important indicator for coronary artery diseases. To perform early diagnosis the myocardial perfusion needs to be quantified. For this the myocardium has to be segmented in all time steps of the image sequence. Manually identification of myocardial boundaries is a very time consuming and tedious task. Existing automatic methods ignore motion artifacts, assumes that the heart is located at the center of the images, or requires user interactions to select a region of interest (ROI) covering the heart. Therefore, a fully automatic segmentation method is introduced in this work. This method uses spatial and temporal information and deals with motion artifacts. The ROI covering the heart is located by a hierarchical pattern matching algorithm. Motion artifacts are minimized by image registration using mutual information. Estimation of characteristic intensity-time curves followed by pixel-wise classification and boundary extraction deliver the final segmentation. The proposed method was successfully evaluated for 10 out of 11 data sets.

[1]  S. Dymarkowski,et al.  Clinical cardiac MRI , 2005 .

[2]  S. Sotthivirat,et al.  Active Contours With Automatic Initialization For Myocardial Perfusion Analysis , 2005, 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference.

[3]  D A Bluemke,et al.  Myocardial First-Pass Perfusion Magnetic Resonance Imaging: A Multicenter Dose-Ranging Study , 2004, Circulation.

[4]  Jinn-Yi Yeh,et al.  Myocardial border detection by branch-and-bound dynamic programming in magnetic resonance images , 2005, Comput. Methods Programs Biomed..

[5]  H. Oswald,et al.  Noninvasive detection of myocardial ischemia from perfusion reserve based on cardiovascular magnetic resonance. , 2000, Circulation.

[6]  M. Breeuwer,et al.  Automatic correction of myocardial boundaries in MR cardio perfusion analysis , 2002 .

[7]  Hildur Ólafsdóttir,et al.  Nonrigid registration of myocardial perfusion MRI , 2005 .

[8]  E. Wu,et al.  Image Registration in Myocardial Perfusion MRI , 2005, 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference.

[9]  R. Wilson,et al.  Myocardial perfusion reserve: assessment with multisection, quantitative, first-pass MR imaging. , 1997, Radiology.

[10]  M. Breeuwer,et al.  Optimal myocardial boundary estimation for MR cardio perfusion measurements using sensitivity analysis , 2002, Computers in Cardiology.

[11]  S. Globits,et al.  Myokardiale First-Pass-Perfusionsdiagnostik mittels Magnetresonanztomographie , 2003 .

[12]  José M. F. Moura,et al.  STACS: new active contour scheme for cardiac MR image segmentation , 2005, IEEE Transactions on Medical Imaging.

[13]  Vincent Vaerman,et al.  Automatic heart localization from a 4D MRI data set , 1997, Medical Imaging.

[14]  Ross T. Whitaker,et al.  Automatic segmentation of cardiac short axis slices in perfusion , 2006, 3rd IEEE International Symposium on Biomedical Imaging: Nano to Macro, 2006..

[15]  David T. Gering,et al.  Automatic Segmentation of Cardiac MRI , 2003, MICCAI.

[16]  José M. F. Moura,et al.  Contrast-Invariant Registration of Cardiac and Renal MR Perfusion Images , 2004, MICCAI.

[17]  Hildur Ólafsdóttir,et al.  Unsupervised motion-compensation of multi-slice cardiac perfusion MRI , 2005, Medical Image Anal..

[18]  Mikkel B. Stegmann,et al.  Motion-Compensation of Cardiac Perfusion MRI Using a Statistical Texture Ensemble , 2003, FIMH.

[19]  Katja Bühler,et al.  Fast and Robust Localization of the Heart in Cardiac MRI Series - a Cascade of Operations for Automatically Detecting the Heart in Cine MRI Series , 2008, VISAPP.

[20]  Boudewijn P. F. Lelieveldt,et al.  Fully Automated Registration of First-Pass Myocardial Perfusion MRI Using Independent Component Analysis , 2007, IPMI.

[21]  Paul A. Viola,et al.  Alignment by Maximization of Mutual Information , 1997, International Journal of Computer Vision.