Determining acceptance levels for automatic daily image quality control in magnetic resonance imaging

Transcription

1 Determining acceptance levels for automatic daily image quality control in magnetic resonance imaging Poster No.: C-1125 Congress: ECR 2016 Type: Authors: Keywords: DOI: Scientific Exhibit J. I. Peltonen, T. Mäkelä, A. Sofiev, E. Salli; Helsinki/FI Quality assurance, Physics, Equipment, Computer Applications- General, MR, MR physics /ecr2016/C-1125 Any information contained in this pdf file is automatically generated from digital material submitted to EPOS by third parties in the form of scientific presentations. References to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not in any way constitute or imply ECR's endorsement, sponsorship or recommendation of the third party, information, product or service. ECR is not responsible for the content of these pages and does not make any representations regarding the content or accuracy of material in this file. As per copyright regulations, any unauthorised use of the material or parts thereof as well as commercial reproduction or multiple distribution by any traditional or electronically based reproduction/publication method ist strictly prohibited. You agree to defend, indemnify, and hold ECR harmless from and against any and all claims, damages, costs, and expenses, including attorneys' fees, arising from or related to your use of these pages. Please note: Links to movies, ppt slideshows and any other multimedia files are not available in the pdf version of presentations. Page 1 of 14

2 Aims and objectives The quality control (QC) process of magnetic resonance imaging (MRI) equipment can be considered as two folded process. The absolute performance of the equipment is measured typically annually. Between these comprehensive evaluations, the image quality is monitored with more lightweight checks. The checks are typically done daily and are labeled respectively. [1-5] The daily quality assurance is usually seen as a quick control before the first patient of the day to test the image quality of the equipment and to see if any error messages emerge. If the checkup is repeated daily with identical imaging parameters and phantom setup, the images can be used to derive long term data from the MRI scanner performance. Available variables are depending on the used phantom which is also effecting the required time and complexity of the imaging. [6-7] If the number of scanners is great, the required workload to keep up with the daily quality control image flow can become overwhelming when resting solely on manual labor. The speed and repeatability of the QA process can be improved greatly if automatic image processing is used [8-9]. The automatic workflow can be developed even further to derive device specific acceptance levels for the parameters to detect hardware faults. In this study we have developed a method of determining acceptance levels for signalto-noise ratio, image uniformity and image ghosting measured from daily QC data. The presented method is independent of used scanner-phantom setup and it can be easily used to deliver alerts on hardware faults to QC personnel. Page 2 of 14

3 Methods and materials Currently, the range of scanners consists of 14 scanners which of 3 are mobile and 1 is an extremity scanner. Imaging sites are located geographically sparsely and each of them contains from one to three scanner. The initial QC image scanning is done daily before the first patient study of the day similarly with all scanners. The image is taken using a standard head coil and a cylindrical, disc or sphere shaped phantom provided by the manufacturer of the machine. With the extremity scanner, a coil with minimal diameter able to house the phantom is used. The exact combination of the used coil and phantom model is varying between imaging sites. Scanned QA images are sent to DICOM server dedicated to image processing. The process has been presented as a flowchart in Figure 1. The imaging sequence used for the daily QA image is a standard spin echo (SE) sequence with parameters presented in the Table 1. The resulting image includes variable sized round signal area in the middle and void background area around representing the noise as presented in Figure 2. The exact daily QC sequence parameters are presented in Table 1. The variables extracted from the QC images are signal-to-noise ratio, image uniformity, image ghosting. The parameters are calculated according to IEC [10] and NEMA [11-12] recommendations. The normality of the results in each parameter is studied by using one-sample Kolmogorov-Smirnov test (p < 0.05). The normality of each parameter is an essential requirement to be able to use statistical measures in defining detection intervals for parameters. The detection intervals for the parameters are set by using sliding window spanning on last 30 days. From the window the mean and standard deviation of each parameter is calculated. The initial detection interval is set to three standard deviations above and below the mean value. After that the values in the window, which are outside the calculated limits, are removed and the calculation repeated to obtain final detection intervals. The used method is tested by calculating resulting detection intervals for two years of operation on one magnet for each parameter. The magnet is a 1.5T statically installed scanner. During the period, the operation of the scanner was mostly stable with some minor technical problems. Page 3 of 14

4 Images for this section: Fig. 1: The flowchart of the used automatic MRI QA process. Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 4 of 14

5 Table 1: The daily QA sequence parameters. Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 5 of 14

6 Fig. 2: An example of a daily QA image. Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 6 of 14

7 Results The signal-to-noise ratio was determined as normally distributed in all MRI systems. Both image uniformity and image ghosting were determined as normally distributed in 11 MRI systems. Thus, the acceptance levels for each parameter can be determined as a difference from the mean value of the existing results in the multiples of the standard deviation. The effect of the used multiple of standard deviation is illustrated in Figure 3. The example of resulting detection intervals in one scanner is presented in Figures 4-6. The longer time span of abnormal signal-to-noise ratio is representing a randomly occurring problem in the signal path of the scanner. Single outliners in each measure are representing misaligned phantom or similar human error. Page 7 of 14

8 Images for this section: Fig. 3: For example, by setting two sided acceptance levels three standard deviations (3#) above and below the mean value of the results, we will get 0.26% rate for false error indications in the automatic image and results processing. In this example these acceptance levels are applied to a SNR data of one magnet. Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 8 of 14

9 Fig. 4: An example of retrospective adaptable detection intervals in signal to noise measurement Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Fig. 5: An example of retrospective adaptable detection intervals in uniformity percentage measurement Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 9 of 14

10 Fig. 6: An example of retrospective adaptable detection intervals in ghosting percentage measurement Radiology, HUS Helsinki Medical Imaging Center - Helsinki/FI Page 10 of 14

11 Conclusion We have found that acceptance levels for key image quality parameters in automatic daily MRI QC can be derived based on mean value and the standard deviation of the earlier data. The presented method enables a tool for detecting abnormal data from the daily QC image flow. The method is adapting to any scanner-phantom combination and requiring minimum amount of manual labor after the initial setup. If the system is integrated as a part of daily QC data gathering, the abnormal operation of scanner equipment can be detected without delays before any major effect on clinical image quality or scanner breakdown. Page 11 of 14

12 Personal information In case of any questions, comments or ideas feel free to contact the presenting author. Authors of this poster would be more than happy to explain any of unclear issues, exchange experiences and recieve a valuable feedback. Juha Peltonen juha.peltonen@hus.fi J. I. Peltonen, T. Mäkelä, A. Sofiev, E. Salli HUS Medical Imaging Center, Radiology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland Page 12 of 14

13 References 1. Firbank MJ, Harrison RM, Williams ED & Coulthard A. Quality assurance for MRI: practical experience. Br J Radiol 2000:73: McRobbie D & Quest R. Effectiveness and relevance of MR acceptance testing: results of an 8 year audit. Br J Radiol 2002:75: Ihalainen T, Sipilä O & Savolainen S. MRI quality control: six imagers studied using eleven unified image quality parameters. Eur Radiol 2004:14: Ihalainen TM, Lönnroth NT, Peltonen JI, et al. MRI quality assurance using the ACR phantom in a multi-unit imaging center. Acta Oncol 2011:50: Koller C, Eatough J, Mountford P & Frain G. A survey of MRI quality assurance programmes. Br J Radiol Gardner EA, Ellis JH, Hyde RJ, Aisen AM, Quint DJ & Carson PL. Detection of degradation of magnetic resonance (MR) images: comparison of an automated MR image-quality analysis system with trained human observers. Acad Radiol 1995:2: Gunter JL, Bernstein MA, Borowski BJ, et al. Measurement of MRI scanner performance with the ADNI phantom. Med Phys 2009:36: Reiner BI. Automating quality assurance for digital radiography. Journal of the American College of Radiology 2009:6: Bourel P, Gibon D, Coste E, Daanen V & Rousseau J. Automatic quality assessment protocol for MRI equipment. Med Phys 1999:26: International Engineering Consortium. IEC Magnetic Resonance Equipment for Medical Imaging-Part 1: Determination of Essential Image Quality Parameters National Electrical Manufacturers Association. NEMA Standards Publication MS Determination of Signal-to-Noise Ratio (SNR) in Diagnostic Magnetic Resonance Imaging Page 13 of 14

14 12. National Electrical Manufacturers Association. NEMA Standards Publication MS Determination of Image Uniformity in Diagnostic Magnetic Resonance Images Page 14 of 14