Automated Universal Image Quality Index Measurement vs. Automated Noise Measurement: Which Method is Better to Define CT Image Quality?

Abstract

Automatitation method in defining the quality of CT image is needed to optimize CT Scan treatment planning. So, the optimization of treatment planning can also be done automatically. There are various methods proposed to define the quality of an image. The purpose of this study was to find the simple and precision method to define CT image. We compared the performance of Automated Noise Measurement (ANM) and Automated Universal Image Quality Index (UIQI). We also compared them with the Manual noise measurement method based on the level of convergence in homogeneous images. The first step of Automated Noise Measurement was to create binary density slice using threshold values. Then, a masked image was performed by masking the original image and binary image. The standard deviation of every pixel for a certain kernel size was calculated by using a sliding window operation. The fourth step was to make a noise histogram from the noise map and determine the final noise in the image as the histogram peak. Then this calculation was normalized by the peak of the Hounsfield Unit (HU) histogram. All these steps were done with various kernel sizes for different slices in-homogenous phantom. In the Automatic UIQI method, the steps in the ANM method are carried out until the masked image stage, then UIQI is calculated for the masked image. The results show that automatic UIQI was more convergence in defining image quality than manual noise measurement and automated noise measurement by the lowest standard deviation which was only 0.00032867.