Program Information

A Software Tool for Automated Analysis of CT Quality Assurance Phantoms

J Grimes

J Grimes*, S Leng , C McCollough , Mayo Clinic, Rochester, MN

Presentations

TH-C-18A-7 Thursday 10:15AM - 12:15PM Room: 18A

Purpose: Automated software can save time, reduce cost and minimize human error in the analysis of quality-assurance phantom images. The goal of this work was to develop a software tool to facilitate an efficient, enterprise-wide, CT quality-assurance program and to perform the daily water QC tests recommended by the ACR.

Methods: The QC software was developed in Matlab and includes tools for automating analysis of the daily water phantom and all four modules of the ACR CT phantom. The daily water phantom analysis follows guidelines from the ACR’s 2012 CT quality control manual, which requires ROI analysis in the center and leading/trailing image of an image stack of both axial and helical acquisitions. A DICOM receiver on a dedicated workstation was configured to run the software automatically after receiving phantom images from a scanner. Results are added to a Matlab structure that serves as a database for tracking data trends, and a standardized report is automatically generated. Performance of the automated ACR phantom analysis was compared to that of trained observers for two tasks requiring subjective user decisions (slice thickness and high-contrast spatial resolution tasks). Images with multiple slice thicknesses and reconstruction kernels were evaluated by two observers and their results compared to the automated analyses.

Results: As part of an ongoing pilot study, preliminary tests of the software have been performed on two scanners. Results are available within seconds of scanning the QC phantom. In the human observer study, the software-estimated slice thickness was within +/- 0.5 mm and the highest frequency bar pattern was within +/- 1 line pairs/cm.

Conclusion: The developed software provides a robust tool for efficiently managing quality assurance of a large number of scanners. The workflow features an efficient web-based design that automatically analyzes QC images sent from the scanner and generates user-customizable reports.


Contact Email: