Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Automated Image Registration of LGE‐MR Imaging and Tc‐99m SPECT Myocardial Perfusion for Validation of Scar Quantification Marina Piccinelli, PhD1, Weihua Zhou, PhD1, Ji Chen, PhD1, C. David Cooke, MSEE1, John N. Oshinski, PhD1, Jonathan D. Suever, MS2, Arshed A. Quyyumi, MD3, Ernest V. Garcia, PhD1 ASNC, September 26-29, 2013 Chicago, IL 1Department of Radiology and Imaging Sciences, Emory University School of Medicine, Atlanta, GA, USA BACKGROUND AND OBJECTIVES Both LGE‐MR and SPECT MPI have been clinically used to assess myocardial infarction. We developed a procedure for the automated co‐registration of the 3D SPECT LV perfusion distribution onto the 3D LGE‐MR distribution to more accurately validate our new MPI‐based scar quantification using registered LGE‐MR as the reference standard. 2Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA Quantitative SPECT and Scar Quantification Tool Quantitative scar analysis of the 3D LV SPECT images was performed with the ECTb using perfusion, viability and thickening criteria. METHODS The registration procedure was applied to the whole cohort allowing for a co‐localization of the infarcted tissue as identified by the ECTb and the MR areas with highest scar percentage. Image resolution still limits scar quantification at the apex and the base. Display of ECTb tool for MPI infarct size quantification Twenty patients who had both LGE‐MR and resting Tc‐99m SPECT MPI at 6 months post‐MI have already been acquired for this preliminary validation. Extraction of Anatomical Features from LGE‐MR Binary masks were manually created from the LGE‐MR: one identifying the biventricular contour and one delineating the LV endo & epicardium. Shape analysis tools allowed the rotation of the MR images to match the conventional orientation used for MPI. A 3D triangulated surface of the LV epicardium was extracted. The scar was identified on MR slices and quantified radially slice by slice as a % of local myocardium thickness. Original Image RESULTS 3Division of Cardiology, Emory University School of Medicine, Atlanta, GA, USA Example cases 1 and 2 (from top to bottom) 3D reconstruction of MPI‐derived LV mid‐ wall surface with infarct area displayed. Epicardial wall also shown Scar Identification LGE‐MR / MPI‐SPECT Registration A rigid surface registration using the Iterative Closest Point algorithm was performed to register the two LV epicardial surfaces. The calculated transformation matrix was then successively used to align the two volumetric images. MR and SPECT image fusion MPI‐derived mid‐wall surface: in red blackout infarct area Percentage of scar on myocardium thickness represented on a few contiguous mid‐wall profiles (same view as center figure) CONCLUSIONS The MR/SPECT registration in combination with our newly developed scar quantitative tools allowed the co‐registration of LGE‐MR and SPECT and the direct comparison of scar extent for future validation using our 20 patient cohort. Disclosures and Acknowledgements Myocardium 3D reconstruction Segmented scar 3D distribution MR and SPECT image fusion 3D rigid registration of SPECT surface and image on MR This study was supported by NIH grant R01‐HL‐085417. Some of the authors (ECG, JC, CDC) receives royalties from the sale of the ECTbTM related to this research. The terms of this arrangement have been reviewed and approved by Emory University in accordance with its conflicts of interest practice.