A novel method using histogram modification and Fourier domain filtering automatically tracked tag positions with an average discrepancy of 0.04 +/- 0.17 mm compared to theoretical predictions.
The proposed histogram modification and Fourier domain filtering techniques improve contrast and accurately track tag positions in cardiac magnetic resonance images.
Cardiac tagging permits non-invasive study of myocardial motion with high accuracy. Unfortunately, tagging contrast is impaired at later heart phases due to longitudinal relaxation. Histogram modification is presented as a method for improving contrast in later, faded images of a tagging series by altering these images such that their intensity histograms approximate the shape of the first, unfaded image of the series. This technique greatly improves the contrast and facilitates automatic detection of tags. Furthermore, a method is described for automatically tracking tag positions in short-axis images of the left ventricle modulated with a tagging grid. The method differs from previously reported methods in that, in one single filtering process in the Fourier domain, both the grid crossings as well as the grid centers are detected, and thus increased sampling resolution is obtained. The method was validated by applying a mathematical model of left ventricular motion to tagged images of the thigh muscle. The average discrepancy between theoretically predicted and automatically detected tag positions was 0.04 +/- 0.17 mm (mean +/- SD).
Koerkamp et al. (Sat,) conducted a other in Cardiac magnetic resonance imaging. Histogram modification and Fourier domain filtering vs. Previously reported methods was evaluated on Average discrepancy between theoretically predicted and automatically detected tag positions. A novel method using histogram modification and Fourier domain filtering automatically tracked tag positions with an average discrepancy of 0.04 +/- 0.17 mm compared to theoretical predictions.