Develop Automation Program of Quality Control Standard Phantom of Mammography using Digital Image Processing Technology

Abstract

The quality control using phantom in mammography is visual inspection so that the deviation can be occurred depending on the evaluator’s subjective decisions, automated detection program was developed to minimize the errors. The study subjects were 165 phantom images in 74 hospitals nationwide that passed the test of the special medical equipment by Korean Institute for Accreditation of Medical Image from May 2015 to April 2016. We analyzed the intra-rater reliability and inter-rater reliability. In the analysis outcomes to analyze reproducibility of the visual inspection, inter-rater reliability ICC was shown to be low (Fiber=0.349, Specks=0.265, Masses=0.212, Total lesions=0.378) while intra-rater reliability mean ICC was to be approximately 60%. Upon the comparison results for test objects applying the existing visual inspection method and the automated detection program based on the developed image processing technique, the latter method was evaluated to be higher utility as the new analytical method for the standardization of standard phantom quality control, detecting 0.38 unit of fiber and one unit of mass. The automated detection program can overcome the biases by manual processes, and improve the reproducibility so as to automate the quality control system consistently, simply, and correctly within the short time.