Efficient Interpolations for Fusing Short-Axis and Long-Axis MR Images To Enhance 3-D Cardiac Reconstruction

Abstract

A new method is proposed for fusing short-axis and long-axis MR images using efficient interpolations. It is suitable for the enhancement of 3-D cardiac reconstruction. Short-axis cardiac images are similar to regular MR images, which provide information about the cross-section of a heart, and long-axis cardiac images contribute supplementary information in a parallel direction to the long-axis. Our method can produce higher quality interpolated images compared to other methods. It includes five steps: arranging long-axis images in a circular form, finding a cubic spline curve, circular interpolation, fast parametric shape-based interpolation of short-axis images, and fusion of short-axis and long-axis images. From our experiments, the processing time improvement is fifty-three times faster than the gray-level shape-based method with at least 20% less mean square errors. Compared to the linear interpolation, our method has improved a minimal of 34% in mean square errors. This approach can help doctors make their diagnoses faster, easier, and more precise.