Robert A. Bell, PhD

On the preference scale, conducting on-site MRI quality control (QC) appears to rank just slightly higher than a colonoscopy examination and an Internal Revenue Service audit. Although most facility managers know that QC can be helpful, their excuses for failing to have a program orfor those with QCnot demanding a conscientious effort from their staff, are varied and colorful.

“My service contract covers QC.”
“It takes too much time away from my patient scanning.”
“The phantom costs too much.”
“A QC program is too difficult for my staff to administer.”
“QC doesn’t provide me with enough useful information to justify the effort.”

A 16-year career in independent MRI performance testing has led me to the conclusion that none of these arguments has any merit. During that time, I have conducted testing on more than 450 MRI systems ranging in field strength from 0.8T to 3T and spanning virtually every MRI vendor.  The large majority of these studies has been on newly commissioned MRI systems, for which I have been asked to be part of initial acceptance testing. One should note that each of these units had been certified for first-patient use prior to my testing.

In about 70% of these cases, a problem with the system was identified. Although some problems were of a minor nature (small geometric distortion errors or patient landmark light misalignment), also encountered were bad head and body coils, faulty gradient amplifiers, leaky RF rooms, large internal landmark errors, substantial unmarked areas experiencing greater than 5 gauss, major image artifacts, intermittent computers, and a host of other serious failures. These problems were all seen after the vendors had certified the systems as ready for patient scanning.


This article is an attempt to persuade those who care about image quality to care about scanning efficiency and understand that, bottom line, MRI quality control is worth the effort. The information it provides will advise of potential future problems and help to focus efforts when difficulties arise. To demonstrate how, each excuse will be explored.

“My service contract covers QC.” There is no question that MRI systems have become more robust with the advent of digital spectrometer control and advanced gradient and RF systems. However, it is also clear that vendors now require each of their service engineers to be the first responder for many more MRI units than in past years. The typical service engineer now has prime responsibility for six to 10 MRI units and some cover more than a dozen. While the skill and dedication of these professionals are to be admired, a large fraction of their work is reactive in nature. They usually respond to problems and spend far less time analyzing otherwise “healthy” units. Although some vendors offer interactive analysis via the web, these programs do not preclude the appearance of some image artifacts or catch all equipment malfunctions.

“QC takes too much time away from my patient scanning.” A properly designed and implemented on-site QC program should not require more than about 7 minutes per day for a daily program or 10 minutes per week if this is all the time that can be  afforded. If the technologist on the last shift places the phantom in the head coil before leaving, QC can be the first scan of the next day and little effort is wasted in setup. In deciding between daily and weekly, just remember that the latter will decrease the likelihood of spotting a problem by a factor of seven. It also seems that faults occur just after the last quality control check.

“The phantom costs too much.” Phantom cost is certainly a reasonable concern.  When the American College of Radiology (ACR) was developing the procedures for MRI accreditation, one of the goals was to produce a phantom that could be sold for less than $1,000. Although the price increased last year, the ACR MRI phantom, can still be purchased for about $780, the equivalent of about 2 hours of service outside contracted hours. This can be used in the vast majority of vendors’ head coils and provides a wide range of tests (vide infra).

Figure 1. The ACR MRI phantom can be purchased for about $780.

“A QC program is too difficult for my staff to administer.” Setting up an in-house QC program can be a major undertaking, requiring the skills of someone who has a profound background in MRI. Fortunately, the ACR has produced such a program and has excellent materials to guide initiates. The QC guide, published in 2001, is a collective effort of many highly knowledgeable contributors who have spent countless hours on MRI test procedures. The guide contains lucid descriptions of each test and has pictures of the expected results.

Figure 2. The ACR’s MRI Quality Control Manual contains all necessary materials for establishing a QC program for MRI.

“QC does not provide me with enough useful information to justify the effort.” The technologists spend the most time on the system. Their eyes and experience can help to detect instrument problems often before catastrophe strikes. QC provides them with tools to focus on selected operational characteristics and with a periodic review of system performance. One should remember that a single hour of downtime usually costs about $600 to $1,000 in lost revenue at the average MRI site.

When one of your physicians complains about image quality, how do you determine if it was due to lack of patient cooperation, operator error, poor choice of protocols, or machine malfunction? Without QC, one major source of error cannot be eliminated. QC programs also tend to help the technologists to understand more about MRI physics and to become more skilled in the unit’s proper operation. They can often foresee problems and suggest solutions before scanning is finished. They are also more effective in calming patient fears about the examination since they can speak with more authority about its fundamentals.


So how does one start an MRI QC program? What tests should it cover? Who should be responsible for the various aspects of the program? What data should be recorded? How often should QC be conducted?

If you have access to an MRI scientist or a medical physicist who has training in MRI, discuss your needs with them and follow their advice. Large hospitals or university medical centers with MRI research programs will undoubtedly have such expertise available. Smaller institutions or freestanding imaging centers may not enjoy these services. For them, there is a ready solution at hand.

A cost-effective and time-efficient QC plan has been developed by the ACR as a part of its MRI accreditation program.By completing accreditation, a site demonstrates compliance with a recognized standard. Indeed, some payors now require ACR accreditation before they will reimburse a site for MRI examinations. Approximately 3,200 sites and more than 4,000 magnets have now been accredited or are in the process. On application, the ACR will send the MRI QC manual and an order form for the phantom as a part of its standard package.

The ACR MRI QC program specifies at least weekly measurement of the following system parameters:

  • General condition of the system
  • Center frequency and transmitter gain
  • Geometric distortion (3 axes)
  • High-contrast spatial resolution
  • Low-density contrast detectability
  • Image artifact assessment
  • Hard-copy image QC

Checking these parameters requires only two scans for a total imaging time of 3 minutes. Analysis takes only 3 to 4 minutes so the entire process can be completed while the first patient of the day is filling out their paperwork or while the coffee is brewing. The weekly review of film accuracy adds another 3 minutes. A review of these items covers a wide range of potential system malfunctions and can alert the operator to problems in time to get service on-site pronto.

General Condition of the System. This is a brief but thorough overview of table motion, console function, RF door seal, room temperature, patient monitors, cryogen levels, and other aspects of the imaging environment. Substandard performance can be noted and the appropriate response requested usually without going “hard down.”

Center Frequency and Transmitter Gain. Center frequency is a direct function of the magnet’s field. Drift in one means drift in the other. For a standard superconducting magnet, drift should be less than about 0.05 parts per million (ppm) per hour. Faster drift indicates the magnet is not as superconducting as it should be (usually due to inadequate welds in the superconducting wire of the magnet). Transmitter gain is a measure of the RF power needed to produce a 90-degree flip of the patient’s magnetic field. For studies on the same phantom, it should be highly consistent. In a sense, it is similar to taking a patient’s temperature, a sensitive test but not highly specific. Changes usually indicate possible problems in the RF transmitter or receiver circuitry (includes coils).

Geometric Distortion. Sizes and shapes in MRI depend on the accurate calibration of the spatial gradients. These can change and need adjustment periodically. We check the length of known distances in three dimensions to assess the accuracy of the gradients and, hence, the accuracy of tumor size measurements, etc, taken from the images.

Figure 3. Sizes and shapes in MRI depend on accurate calibration of the spatial gradients.

High-Contrast Spatial Resolution. A series of holes of known sizes are used to determine if one can properly resolve objects based on the field of view and the acquisition matrix. Failure can be due to the use of overly aggressive spatial filters, vibration in the system (motion blurring), inadequate gradient function, or other causes.

Figure 4. Failure to resolve objects based on the field of view and the acquisition matrix can be due to the use of overly aggressive spatial filters.

Low-Contrast Object Detectability. This test uses a series of thin plates with holes in them to assess the ability to discern size at low contrast levels. It also serves as a visual evaluation of signal-to-noise ratio (SNR) because decreases in SNR reduce the number of holes that are seen. This is a quick and easy way to see if your system SNR is changing.

Figure 5. Test serves as a visual evaluation of SNR ratio because decreases in SNR reduce the number of holes that are seen.

Image Artifact Assessment. Any artifacts should be noted and reported to service. These can often presage equipment failure. They are also an excellent way to monitor the condition of the RF screen room since leakage usually produces dotted line artifacts across the images perpendicular to the frequency-encoding direction.

Hard-Copy Image QC. How do you know that the image on the system monitor is being faithfully transferred to film? This can be verified by filming the Society of Motion Picture and Television Engineers (SMPTE) test pattern from the console and checking the optical densities of the grayscale patches within it. Be sure to use the correct window and level settings as specified by your vendor. The evaluation requires a transmission densitometer, but such devices are usually part of the imaging department (standard in mammography QC) or can be obtained as used devices for a few hundred dollars.

Figure 6. Optical densities in the SMPTE test pattern verify that the image on the system monitor is being transferred to film.

The ACR program also specifies the evaluation of other system characteristics on an annual basis. These include:

  • Magnetic field homogeneity
  • Slice position accuracy
  • Slice thickness accuracy
  • RF coil checks for all coils used clinically
  • Interslice RF interference
  • Soft-copy display accuracy

Typically, these are beyond the training of local staff and are usually done by an MRI scientist or a medical physicist with MRI experience. ACR refers to such an examination as the annual benchmark study. These tests serve as vital checks that the parts of the system are functioning properly (homogeneity and RF coils) and that fundamental system parameters such as slice thickness and location are accurate. Photometry evaluation of the system monitors can detect dimming over time and shading that may not be apparent to the user. Both problems limit the viewer’s ability to detect subtle image information.


A number of additional tests above and beyond what is required for compliance with the ACR accreditation program can also help a facility maintain a magnet uptime. These tests are described below.

SNR Consistency. Acceptance testing or benchmark examinations usually take about 8 to 12 hours on-site. During this time, it is advisable to conduct a number of SNR tests on the head coil that are separated by a few hours. This gives some indication of the repeatability of head coil performance. Variations of more than a few percentage points often reflect intermittent system problems or periodic noise sources.

Quadrature Phase Error. If you have an older MRI system that does not utilize digital phase discrimination, it may be prudent to have your MRI scientist examine it for quadrature phase artifacts and to check the phase setting.

Magnetic Fringe Field. For system testing, I carry a gaussmeter and verify the location of the fringe field 5 gauss line per MRI site plan. Surprisingly, I have found at least a dozen locations in which a substantial amount of magnetism (>5 gauss) was present in uncontrolled areas or in space that was not properly designated by warning signs. This is a violation of FDA directives and is a possible source of liability if someone with a pacemaker or other magnetically sensitive device is adversely affected.

Staff Training Review. When conducting a benchmark study or acceptance testing, it is also advisable to review the on-site QC program. This is an excellent opportunity to verify that staff members are conducting QC properly and are recording the correct information. It is also a good time to address any questions and to reinforce the importance of QC to overall system performance and image quality.

The Role of the Service Engineer. Local service personnel sometimes view QC efforts as a lack of confidence in their work. It is important that they be involved in the process and not treat it with suspicion. Help them to understand that QC is a group effortin which they are a key factorto maintain superb image quality and system performance. Ask for their comments on your QC plans, and invite local service to be present during system testing. They are a wealth of information and can be extremely helpful.


Quality control is worth the effort. It will provide greater confidence in your system performance, decrease downtime, improve your technologists’ understanding of MRI, and encourage local service to be responsive when you spot a problem. QC is best conducted on a daily basis but if all you can commit to is once a week, it is better than no QC at all. And it is not nearly as bad as a colonoscopy examination.

Author’s Note: Contact the ACR for information on the MRI Accreditation Program at The American College of Radiology, 1891 Preston White Drive, Reston, VA 22091, (800) 770-0145; .

Robert A. Bell, PhD, is president of RA Bell and Associates, an independent consulting firm specializing in the technical and operational aspects of advanced imaging modalities. He has personally tested more than 450 MRIs for clients in most of the United States, Canada, and Australia. He welcomes questions and comments and can be contacted at (858) 759-0150.