An Open Benchmark Challenge for Motion Correction of Myocardial Perfusion MRI

摘要: Cardiac magnetic resonance perfusion examinations enable noninvasive quantification of myocardial blood flow. However, motion between frames due to breathing must be corrected for quantitative analysis. Although several methods have been proposed, there is a lack widely available benchmarks compare different algorithms. We sought many algorithms from groups in an open benchmark challenge. Nine clinical studies two centers comprising normal and diseased myocardium at both rest stress were made this study. The primary validation measure was regional flow based on the transfer coefficient $(K^{\rm{trans}})$ , which computed using compartment model reserve (MPR) index. ground truth calculated contours drawn manually all by single observer, visually inspected second observer. Six participated 19 correction compared. Each method used one three models: rigid, global affine, or local deformation. similarity metric also varied with employing either sum-of-squared differences, mutual information, cross correlation. There no significant differences $K^{\rm{trans}}$ MPR compared across models metrics. Compared truth, only metric, deformation models, had bias. In conclusion, enabled evaluation indices over wide range methods. particular, benefit nonrigid registration techniques other evaluated data results are Atlas Project ( www.cardiacatlas.org ).