Evaluation of Low-Contrast Resolution for Computed Tomography by Method of Optimized CNR and Experts' Subjective Evaluation

Abstract

This study aimed to optimize the evaluation procedure of contrast to noise ratio (CNR) on low-contrast resolution (LCR) phantom images of computed tomography (CT). Axial CT images of phantom Catphan 600 which contained the low-contrast module CTP515 and the uniformity module CTP486 were acquired from five 64-slices spiral CT scanners with the same scanning parameters. Through conjoint analysis of images of CTP515 and CTP486, two steps of optimization were added to the evaluation procedure to get the optimized CNR results. Meanwhile, four experts were invited to perform the subjective evaluation of the same low-contrast images. With the method of normalization, results gained by optimized CNR and the experts' subjective evaluation were compared to investigate the consistency of the two evaluation methods. According to the comparison of the normalized values, the optimized CNR gained by the method of two-stage average could be used as basis to assist tester achieve similar result compared to that from experts' subjective evaluation.