Validation of the tail replacement method in MTF calculations using the homogeneous and non-homogeneous edges of a phantom

We validated the tail replacement technique in the modulation transfer function (MTF) calculation of CT images using edges of homogeneous and non-homogeneous phantom using an automated method. The algorithm for the automated MTF calculation consists of several steps. The upper edge of the phantom was detected from its axial image and pixel values was taken to create an edge spread function (ESF). The left tail of the ESF was flipped horizontally to replace the right tail. Differentiating the ESF produced a line spread function (LSF) which was Fourier transformed to produce the MTF. To validate the result for the non-homogeneous phantom was compared with that for a homogeneous phantom. Results showed that in the homogeneous module, the MTF curve generated without tail replacement corresponded to its shape in use. However for the non-homogeneous module, the MTF obtained without tail replacement differed considerably from its shape in use. The 50% value of MTF (MTF-50) without tail replacement was 0.44 cycles/mm compared to 0.33 cycles/mm in use. Using the tail replacement, the MTFs in the homogeneous and non-homogeneous modules were comparable, with MTF-50 values of 0.33 cycles/mm for both. In summary, the MTF cannot be conventionally measured from the edge of the non-homogeneous module, but it can be resolved using the tail replacement technique on the ESF curve.