Breast MRI Accreditation Program Clinical Image Quality Guide

Transcription

1 Breast MRI Accreditation Program Clinical Image Quality Guide Introduction This document provides guidance on breast MRI clinical image quality and describes the criteria used by the ACR Breast MRI Accreditation Program s clinical image reviewers in reviewing images. All breast MRI interpreting physicians and technologists should carefully read this document before selecting and submitting accreditation cases to the ACR. The breast MRI facility s lead interpreting physician is responsible for ensuring high image quality and implementing an effective quality assurance (QA) program. Successful facilities are often those in which the staff s commitment to high quality reflects the interests of the lead interpreting physician. Therefore the lead interpreting physician must routinely review and critique the image quality produced at the facility. Furthermore, the lead interpreting physician must understand the requirements of accreditation and demonstrate an interest in the results. Basic Assumptions The ACR clinical image reviewers assume that the images submitted for accreditation are representative of the facility s best work. The facility s lead interpreting physician for breast MRI must review and approve the selected cases and completed forms and acknowledge this by signing the Test Image Data form. (Incorrect information may result in accreditation failure.) The ACR Committee on Breast MRI Accreditation understands that images obtained during all breast MRI examinations may not meet these criteria. Consequently, sufficient time (45 days) is allowed to select cases that are examples of best work. If you have difficulty finding examples of your best work during this time period, please contact the ACR to request an extension. In addition, ACR reviewers recognize that some cases may be of adequate quality for diagnosis even though they are less than optimal. Also, some specific features of pulse sequences that are used clinically for breast MRI examinations are variable because of the physician s personal preferences as well as due to the varying capabilities of different MRI systems. Despite this variability, experienced interpreting physicians are able to agree on what constitutes acceptable and unacceptable diagnostic exams based on both objective and subjective criteria. The Page 1 of 9

2 accreditation process also provides guidance and recommendations for achieving optimal image quality beyond that which is merely acceptable. The facility must submit 1 case with a known, enhancing, biopsy-proven carcinoma clearly visible in the breast parenchyma. Indicate its laterality and location on the form. The cases must meet the following criteria: The lead interpreting physician must review and approve the clinical images and forms. (Incorrect clinical images or incorrect information listed on the Test Image Data form may result in accreditation failure.) Cases should not be older than 6 months from the date on the Testing Memorandum. Cases must be from actual patients (not models or volunteers) and must have been formally interpreted. All images of a case must be from the same patient. The case must be a bilateral exam of native breasts (i.e., no TRAM or autologous tissue reconstructions). The MRI must be performed prior to any surgery on the breast with cancer (e.g., excisional biopsy or lumpectomy). Do not submit exams from mastectomy patients. This is important for several reasons: - Unilateral cases do not allow ACR reviewers to evaluate your facility s ability to acquire bilateral exams (i.e., they aren't necessarily representative of the timing or spatial resolution of a bilateral exam). - The reviewers will not be able to determine if both sides of the breast coil are working properly without both breasts present. - It eliminates symmetry that clinical readers and reviewers depend on. Each case must include localizer or scout sequences. Each case must include the following 4 sequences. (Aurora systems acquire a precontrast T1-weighted series that is also a T2-weighted/bright fluid series. For those cases, only 3 sequences need be submitted.) - T2-weighted/bright fluid series - Multi-phase T1-weighted series: Pre-contrast T1 Early phase (first) post-contrast T1 Delayed phase (last) post-contrast T1 If possible, only submit the required sequences. If incomplete or incorrect sequences are submitted, the unit fails accreditation. The T2-weighted/bright fluid series may be run as a single series on both breasts or as 2 separate series, 1 on each breast. In the latter case, 2 separate series numbers and data should be entered in the Test Image Data form. Page 2 of 9

3 The multi-phase T1-weighted series: - Must be run bilaterally, with both the left and right breasts in the same series. - May be in 3 separate series, in 2 series (i.e., 1 for pre-contrast, the rest for postcontrast), or in a single series (i.e., pre-contrast and post-contrast). - All 3 multi-phase series (with the exception of Aurora systems) should match in terms of spatial and temporal parameters. Some small deviations may exist in the parameters listed in the series DICOM header files (e.g., in TR and TE values) between the pre- and post-contrast series. As long as these differences are small and do not affect the ability to subtract pre- from post-contrast series, such small differences are acceptable. - If chemical-shift (i.e., frequency-selective) fat suppression is not used or is not evident, then subtraction of pre-contrast from post-contrast series may be used to eliminate the bright signal from fat. If this is done, then both the unsubtracted (source) series and the subtracted series (i.e., pre-contrast subtracted from postcontrast, slice by slice) must be included for both the early and delayed phases. - IV contrast must be evident in the 2 post-contrast T1-weighted series. If possible, each sequence should be presented separately and not as stacked or interleaved sequences. (Contact your MRI manufacturer representative for assistance). All sequences must demonstrate adequate breast positioning and include the entire breast, including the axillary tail. All sequences must demonstrate sufficient signal-to-noise ratio (SNR) and not appear too grainy. If artifacts are present, they must not compromise the diagnostic value of the images. Facilities must submit images in a DICOM format on CD or DVD with an embedded viewer. Each case on the discs submitted for accreditation must be tested by the facility on a separate computer (independent of the acquisition or workstation) to ensure that they open within 2 minutes. Breast MRI Evaluation Categories ACR clinical image reviewers evaluate 5 critical categories when scoring examinations submitted for accreditation. Each is described in detail below. Evaluation Categories A. Pulse Sequences and Image Contrast B. Positioning and Anatomic Coverage C. Artifacts D. Spatial and Temporal Resolution E. Exam Identification Page 3 of 9

4 Category A: Pulse Sequences and Image Contrast The selection of appropriate pulse sequences is critical. They are the major determinant of image contrast and, thus, determine the appearance of both normal tissues and breast pathology. In order to depict and characterize abnormalities, a comprehensive breast MRI exam should include pulse sequences that provide more than one type of image contrast. The specific type of pulse sequences (e.g., conventional SE, fast SE, turbo SE, gradient-echo) and the precise imaging parameters (e.g., TR, TE, flip angle, echo train length) are not specified and are left to the discretion of the imaging facility. The major categories of image contrast that will be assessed in the evaluation process are: 1. T2-weighted/bright fluid series Term used to indicate an image where most of the contrast between tissues or disease states is due to differences in the tissue T2 (or T2*), with fluid such as blood or cystic fluid appearing bright. The term T2-weighted may be misleading in that spin density differences and T1 differences also may contribute to image contrast. The T2-weighted/bright fluid sequence should demonstrate fluid as sufficiently bright to be considered a true bright fluid sequence. The case will fail if bright fluid is absent or indistinguishable from background tissues, such that the deficit could lead to misdiagnosis or is inadequate to reliably distinguish cysts from solid masses. If the bright fluid contrast is present, but is faint or highly variable, the case could fail if other image problems exist. A STIR image with TI set to suppress fat signal may be considered a T2-weighted/bright fluid series if it successfully shows fluid to be bright. A T2*-weighted or a non-spoiled (steady state) T1-weighted gradient echo pulse sequence also may be satisfactory as a bright fluid sequence if it shows fluid as being adequately bright (i.e., brighter than all other tissues in the breast). 2. Multi-phase T1-weighted series a) Pre-contrast T1-weighted without or with fat suppression Term used to indicate an image where most of the contrast between tissues and disease states is due to differences in T1. A T1-weighted image is achieved by imaging with a short TR relative to the longest tissue T1 of interest and a short TE relative to the tissue T2 (to reduce T2 contributions to image contrast). Short TR/short TE sequences are a necessary component of the required examination. This sequence must have sufficient dark fluid contrast to be considered a true T1-weighted sequence. Fat suppression may be used for this sequence, along with the post-contrast series that follow. If fat suppression is not used, subtraction images with these pre-contrast images subtracted from post-contrast images must be reconstructed and submitted. b) Post-contrast T1-weighted with fat suppression or subtraction (early and delayed phases) The intent of early phase and delayed phase post-contrast imaging is to capture information on lesion enhancement in the early and late phases of postcontrast enhancement. These pulse sequences must be identical to the pre-contrast T1-weighted series described above in terms of spatial and temporal parameters. (There are 2 acceptable exceptions to this requirement: 1) small deviations between pre- and post-contrast series may appear in the DICOM header files of some series, even though identical acquisition parameters were selected, and 2) Aurora EDGE Page 4 of 9

5 software uses a distinctly different acquisition time for the pre- contrast series compared with that of the post-contrast series.) If fat suppression is used for the precontrast T1, it should also be used for this sequence. If fat suppression is not used, subtraction images with pre-contrast images subtracted from these post-contrast images must be reconstructed and submitted, along with the unsubtracted source postcontrast T1-weighted images. At least 2 phases of post-contrast images must be submitted: the earliest and the latest phase post-contrast series. All sequences must demonstrate sufficient signal to noise (SNR) and not appear too grainy. The details of the pulse sequence, along with other selected parameters such as imaging matrix and number of signal acquisitions (per phase-encoding step) acquired, will determine the image acquisition time and SNR. In general, short acquisition times are desirable to limit patient motion and discomfort and provide flexibility such that additional sequences may be obtained if desired. However, short acquisition times should not be employed if they can only be obtained at the expense of overall image quality and diagnostic value. IV contrast must be evident in the 2 post-contrast T1-weighted series. Please see the documents on MR Contrast Agents and Contrast Media at the ACR Radiology Safety webpage at: for contrast safety information. Category B: Positioning and Anatomic Coverage Proper breast positioning and anatomic coverage are important components of breast MRI exams. The field-of-view (FOV) and breast coverage must be adequate for accurate diagnosis of breast cancer. Specific image acquisition planes are not specified and are left to the discretion of the facility. The following aspects of positioning will be evaluated: Adequate breast tissue inside the coil Proper positioning of the breast within the coil Properly positioned nipple Coverage of the entire breast, from the axillary tail to the inframammary fold Absence or minimization of skin folds Appropriate FOV Category C: Artifacts Excessive artifacts on any image may interfere with image interpretation. Although some artifacts may be unavoidable on certain images, others may be indicators of errors in pulse sequence selection details, inadequate equipment, or lack of preventive maintenance at an MRI facility. The artifacts listed below are among the most common. All required images should be assessed to determine if any of these artifacts are present, especially if they could potentially compromise the diagnostic value of the images. Images will be reviewed for excessive artifacts that may Page 5 of 9

6 interfere with image quality. Reviewers will not downgrade images for visible markers used to identify nipples or lesions or the presence of clip or metallic artifacts, unless they obscure the cancer or an entire breast. Motion/ghosting These artifacts appear as periodic replication or partial replications of bright structures along the phase-encoding direction. These ghosting artifacts could be due to patient motion, fluid pulsation including cardiac or vascular pulsation, unstable gradients, or other causes. Non-uniform/heterogeneous fat suppression These artifacts appear as uneven darkening of the fat signal in different portions of the image set. This may be due to either a heterogeneous magnetic field or a heterogeneous radiofrequency (RF) field. Aliasing/wrap artifacts The image appears wrapped around onto itself. This is due to signal-producing tissue outside the selected FOV wrapping back into the displayed FOV on the opposite side of the image. If image wrap occurs, it is usually along the phaseencoding direction. Increasing the FOV or applying phase oversampling are the two most effective ways of eliminating or minimizing wrap artifacts. Truncation/ringing artifacts (edge ringing) These artifacts appear as periodic parallel lines or ringing adjacent to borders or tissue discontinuities, in either the phase-encoding or frequency-encoding directions. This is due to selection of too small a matrix in one or both in-plane directions, occurring most commonly in the phase-encoding direction. Non-uniform/heterogeneous signal within breasts This is due to RF heterogeneity, receiver coil non-uniformities, non-functioning coil elements, improper patient positioning, or metal in the magnet or on the patient. Susceptibility These artifacts appear as localized field distortion or non-uniformities produced by differing tissue magnetic susceptibilities (especially at air-tissue interfaces). Chemical shift This artifact occurs along the frequency-encoding direction at fat/water soft tissue interfaces as a thin intense band of high or low signal. Geometric distortion This condition is when size, orientation or shape is not accurately represented on the image. Filtering This condition results when excessive software smoothing is used to reduce apparent noise in the image. Excessive filtering or smoothing obscures true anatomical structure through image blur and can reduce image contrast. RF leak (zipper artifact) These artifacts appear as linear hyperintense or variable intensity lines parallel to the phase-encoding direction. They are often caused by unwanted sources of RF signals originating within (e.g., light bulbs or other electronic equipment) or from outside the scanner room (e.g., RF signals penetrating inside the scan room because the scanner room door is open, the RF seal between the door and frame is damaged, or RF shielding is inadequate). Misregistration of subtracted images On subtracted images, incomplete subtraction of background tissue signals occurs because pre- and post-contrast images do not register properly, usually due to patient motion. Page 6 of 9

7 Category D: Spatial and Temporal Resolution For the multiphase, T1-weighted series, ACR reviewers will evaluate the following parameters obtained from the DICOM header: Slice thickness Interslice gap In-plane pixel size (frequency-encoding direction) In-plane pixel size (phase-encoding direction) Temporal resolution Spatial resolution There are 5 determinants of voxel dimensions in an MRI examination: 1. Slice thickness (ST) 2. Field of view along the phase-encoding direction (FOV p ) 3. Field of view along the frequency-encoding direction (FOV f ) 4. Number of phase encoding steps (N p ) 5. Number of frequency encoding steps (N f ) The voxel volume is the pixel area multiplied by the slice thickness. Pixel Size and Voxel Volume Calculations In-plane pixel size (phase) = (FOV p /N p ) In-plane pixel size (frequency) = (FOV f /N f ) Pixel area = (FOV p /N p ) x (FOV f /N f ) Voxel volume = (Pixel area) x (ST) Alterations in any of these 5 parameters will change the voxel volume, the SNR of the image, and the amount of partial volume averaging exhibited in each image. Alterations in the number of phase-encoding steps (N p ) affects scan time, while alterations in the number of frequencyencoding samples (N f ) may affect the maximum number of slices as well as the minimum possible TE for the imaging sequence. Voxels that are too large for the anatomic structures being examined cause the viewer to observe what is perceived as blurriness in the image. While decreasing the size of the voxels has the potential to increase the sharpness of the image, there is a concomitant decrease in SNR. When the SNR is too low for the anatomic structures being examined, the viewer observes what is perceived as image "graininess." Increasing the SNR of the image by increasing the voxel volume, increases the smoothness of the image, but at the expense of increased image blur, increased partial volume averaging, or both. The Breast MRI Accreditation Program s requirements for spatial resolution for the T1-weighted multi-phase series are as follows: The acquired (not interpolated) slice thickness must be 3.0 mm. Cases with slice thicknesses >4.0 mm will fail; cases with slice thicknesses >3.0 but 4.0 mm may fail if the reviewers note deficiencies in other categories. Page 7 of 9

8 The in-plane pixel resolution must be 1.0 mm (for both phase and frequency). Cases with in-plane pixel resolution >1.2 mm will fail; cases with in-plane pixel resolution >1.0 but 1.2 mm may fail if the reviewers note deficiencies in other categories. The interslice gap must be 0 mm. Cases with gaps >0 mm will fail. Facilities should check the image DICOM header information and use the formulas listed above to determine if their examinations meet the ACR s spatial resolution requirements. Temporal resolution The total time between the contrast injection completion and the end of the early phase post-contrast T1 series is evaluated to assess temporal resolution. This is determined by adding the time delay between the end of the contrast injection and the start of the 1st post-contrast series and the acquisition time for the early phase (i.e., 1st) post-contrast T1 series. For example: The difference between the time at the end of contrast injection and the time that the early phase post-contrast T1 series begins = 40 seconds = Time Delay The difference in time between the time that the early phase post-contrast T1 series begins and the time that the early phase post-contrast T1 series ends = 1 minute (i.e., 60 seconds) = Acquisition Time The Total Time = Time Delay + Acquisition Time = 40 seconds + 60 seconds = 100 seconds (i.e., 1.67 minutes) The first post-contrast sequence must have been completed in 4.0 minutes of completion of contrast injection. Cases with a total time >5.0 min will fail; cases with a total time >1.0 but 5.0 min may fail if the reviewers note deficiencies in other categories. Category E: Exam Identification MR images are important medical records. Standardized identification of these images is important to assure that they are not lost or misinterpreted. If displayed on the image, identification should be easily readable and located so it does not overlap relevant anatomy on the image. If not displayed, the identification must be easily accessed through the DICOM header for each image provided on the CD or DVD. The following exam identification information should be available with each image of each case: Page 8 of 9

9 Patient s first and last names Patient age or date of birth Patient identification number Facility name Examination date Laterality, left or right of midline section Interslice gap All patient information will be kept confidential by the ACR, as stated in the Practice Site Survey Agreement. If laterality is absent or incorrect, the case will fail accreditation. The Testing Instructions explain how to label the discs and disc cases. All accreditation labels provided by the ACR must be placed on the disc case (not on the disc). The disc must be clearly labeled with the CD# and the BMRAP ID# using a permanent marker or a disc label. The ACR will return all accreditation application materials if the discs and cases are not properly labeled. ACR Clinical Image Evaluation Criteria Summary Facilities not submitting the required sequences or submitting cases that do not meet the following criteria will fail accreditation. Required Sequences T2- Weighted/ Bright Fluid Series Pre- Contrast T1 Early Phase Post- Contrast T1 Delayed Phase Post- Contrast T1 Category A: Pulse Sequences and Image Contrast Adequate SNR/not too grainy Sufficient bright fluid contrast Adequate SNR/not too grainy Adequate SNR/not too grainy If fat suppression is not evident, subtracted images also must be provided Technical factors match precontrast T1 IV contrast is evident Image Quality Criteria Category B: Positioning and Anatomic Coverage Adequate breast tissue inside coil Breast properly positioned within coil Properly positioned nipple Image set covers both breasts, from axillary tails to inframammary folds Minimal or no skin folds Category D: Spatial and Temporal Resolution NA Slice Thickness 3 mm Gap 0 mm In-plane pixel (phase) 1 mm In-plane pixel (frequency) 1 mm Early phase postcontrast T1-weighted series completed within 4 minutes of completion of injection Page 9 of 9