EVALUATION OF THE EDGE PROFILE SHIFTING BASED ON STATISTICAL APPROACH TO IMPROVE THE EDGE-BASED MTF MEASUREMENT

This study aims to evaluate the accuracy of edge profile shifting based on a statistical approach to improve the edge-based modulation transfer function (MTF) measurement using the slanted-edge phantom. A slanted-edge phantom was computationally developed with two materials, a water and a material of 120 HU. The spatial resolution of the phantom was varied from 0.3 to 1.25 cycle/mm and the slanted angle was varied from 0° to 10°. The statistical shifting was performed by shifting the treated ESF backward and forward by up to 10 pixels (increment of 1 pixel). At each shift, the similarity between the shifted and reference ESFs was calculated using mean squared error (MSE). The statistical edge-MTF was compared against the traditional edge-MTF and the point-MTF as the gold standard. The statistical edge-MTF approach had similar results against the point-MTF compared to the traditional edge-MTF. At a slanted angle less than 8°, the statistical edge-MTF was comparable to the point-MTF with a difference about 2%. For slanted angles more than 8°, the difference was around 5%. By contrast, the non-statistical edge-MTF approach produced larger differences as the slanted angle increased. The statistical approach for measuring edge-MTF is more robust compared to the traditional edge-MTF approach.