In this paper, we proposed a new quantitative evaluation method of low-contrast detectability for computed tomography(CT)images. It is well known that the contrast-noise-ratio is not enough for CT images to evaluate the low-contrast detectability as image quality index corresponding to various images with different frequency properties. In order to overcome this problem, we adopted a signal-to-noise ratio based on the matched filter model using low-contrast rod images (disc images) as the input signal. The low-contrast detectability indices obtained from this analysis method indicated good agreement with results of perceptual detectability studies. Even for the images obtained by various rod sizes, tube current settings and reconstruction kernels, this method derived appropriate index values matching the results of perceptual studies. The method proposed in this paper is very useful for CT to evaluate quantitative low-contrast image quality.
