4/14/2009. The Big Picture of Quality. MRI Quality Assurance and ACR MRI Accreditation Program. Basic Elements for Image Quality.

Transcription

1 The Big Picture of Quality MRI Quality Assurance and ACR MRI Accreditation Program Chen Lin, PhD Indiana University School of Medicine & Clarian Health Partners Diagnosis accuracy Image quality Knowledge and expertise of radiologists Effective communication Workflow and efficiency Safety Satisfaction of patients and referring physicians Basic Elements for Image Quality Other Contributions to Image Quality 2D SE 60 2D T1 FLAIR 2D FLASH 3D FLASH 3D MPRAGE IQ=4 IQ=6 IQ=4 IQ=5 IQ=5 Signal, Spatial Resolution, Contrast 1. Signal 2. Spatial Resolution 3. Contrast 4. Artifacts 5. Noise 6. Temporal resolution 7. Scan time 8. Temporal Stability or Reproducibility 9. MRS line-width and baseline Common MRI Artifacts Noise in MRI Magnetic field (B 0 ) 1. Main field inhomogeneity 2. Susceptibility artifacts RF Field (B 1 ) 1. Cross-talk (Imperfect slice profile) 2. Zipper (RF leak) 3. RF feed through 4. RF noise/interference 5. Inhomogeneity (intensity and suppression) Data Acquisition and Image Reconstruction 1. Aliasing 2. Truncation (Gibbs Ringing) 3. Blurring 4. Chemical shift 5. Partial volume Physiology / Patient 1. Bulk Motion 2. Flow Originated from the patient / object. Also introduced by the hardware components in the receiving chain/path. Thermal/random noise has the same characteristic over a wide frequency range, i.e. White noise Noise measurement: Standard deviation of pixel intensity in noise region. Signal Intensity Noise Level Gradients (G) 1. Ghosting 2. Distortion or Non-linearity rbw frequency 1

2 Signal to Noise Ratio (SNR) Signal: Average pixel intensity for a specific region/tissue SNR = Signal / Noise SNR = f(seq) B 0 PD x y z ( N PE NEX / RBW ) 1/2 Contrast to Noise Ratio (CNR) Contrast: Signal Tissue1 Signal Tissue2 CNR = Contrast/Noise Better measure of conspicuity SNR 8 SNR 32 = 0.5 SNR 8 SNR 128 = 0.25 SNR 8 Use consistent imaging parameters for SNR measurement and comparison. Artifact to Noise Ratio Artifact: Average of pixel intensity in an artifact region Major Influences to Image Quality 1. Scanner performance, imaging technique & protocol MRCP w. 3D SPACE & Var FA MRCP w. 3D SPACE & Const FA 2. Patient management 3. Exam planning (Slice position, orientation and coverage; Acquisition timing) 4. Image presentation (Film and Display Quality) Image Quality Assurance Scanner Performance Monitoring Acceptance Testing after Installation, Upgrade and Service Scheduled Daily/Weekly QA Scheduled Quarterly/Annually QA Test all coils, RF room integrity, PMU, Display, Injector, etc. Preventive Maintenance Testing for Advanced Applications Protocol Optimization and Standardization Tech Education & Training Image Quality/Artifacts Reporting and Troubleshooting (Continuous Improvement Process) 2/3/2006 ACR QA Program Personnel Qualification Preventive Maintenance and Safety Weekly QA by MR Technologists: Phantom scans, visual checks, film test Acceptance and Annual Testing by Physicist Re-accreditation Every Three Years: Attestation of weekly and annual QA Submit expanded phantom scans Submit clinical cases ACR Weekly QA Tests Visual Checks Patient table RF shield Patient monitoring Safety Equipment Film QA Base + Fog Speed Contrast 2

3 Features in ACR Phantom Appropriate size, shape, loading and relaxation time. Stable for over time An uniform region Grid for geometric accuracy Wedges for slice position measurement Ramps for slice thickness measurement Shallow holes for low contrast detectability measurement Small holes for spatial resolution measurement ACR Phantom Appropriate for most head coils Phantom Setup and slice positioning Straight and in the center of the coil. Use a customized holder or the same support. Line up slice #1 and #11 to the cross of wedges ACR Weekly QA Protocol CSE sequence, TR/TE = 500ms/20ms Acq. matrix = 256x256 and FOV = 250mm 11 slice of 5mm w. 100% gap What about other imaging options? None, except image intensity correction if phase array coil is used. (for accreditation) Normalize or Pre-scan Norm (Siemens) SCIC or PURE (GE) 1. Central Frequency (CF) Typically set to the resonance frequency of water protons. Proportional to the field strength (B 0 ). Search for the largest water signal using spectroscopy sequence with 10Hz increment. Affects the quality of fat saturation, the severity of off-resonance artifacts, etc. Finding Center Frequency Display as one of the calibration / pre-scan parameters. Can also be found in DICOM header: DICOM TAG (0018, 0084) Imaging Frequency ACR recommended criteria: < +/-1.5ppm different between weekly tests Possible cause of failure: Phantom not at iso-center Magnet drift RF (synthesizer) instability 2. Transmitter Gain/Voltage Reflects RF power required to achieve certain flip angle Depends on RF coil (the coupling), phantom (coil load), RF pulse type, etc. Should remain constant over time if nothing in the pulse sequence or hardware has changed. 3

4 184.6 mm Accurate Image 4/14/2009 Finding Transmitter Gain/Voltage Criteria and Causes of Failure GE: displayed on screen (db) Siemens: options >> adjustments Transmitter amplitude (temp) (V) Philips: under system performance parameters rf_act_drivescale Toshiba: acquisition window Center frequency (MHz) RF level Recommended criteria: 10% increase from baseline. Possible cause of failure: Incorrect positioning of phantom. Defect in RF transmitting coil (such as improper quadrature phase shift). RF (amplifier) instability. 3. Geometric Accuracy Gradient Nonlinearity Directly related to the image distortion, affect the accuracy of distance and area measurement. Especially important for applications such as volumetric imaging, image co-registration, imagebased stereotactic procedures. Frequency Linear Gradient Non-linear Gradient X (cm) Object Geometric Accuracy Measurement Criteria and Causes of Failure mm mm ACR Recommended Criteria: +/- 2mm in all directions Possible cause of failure: Gradient non-linearity and miss-calibration Field inhomogeneity and low receiver bandwidth Eddy currents Inappropriate phantom positioning Measure distances along main axes of phantom Compare with known values 4

5 4. Spatial Resolution Measurement Ability to detect small structures. Depends on FOV and acquisition matrix size But also depends on: pulse sequence type relaxation times k-space sampling filter settings etc. Spatial Resolution Measurement Resolution Grid Method UL LR Evaluate the separation of holes arranged in two square arrays. The resolution is the smallest size of holes can be separated. Prevent Partial Volume Error Resolution Holes Image Matrix Criteria and Causes of Failure ACR Recommended Criteria: <= 1.0 mm Possible cause of failure: Blurred by image filter. Interference from artifacts Poor SNR Increased eddy currents Poor gradient calibration Poor B 0 uniformity Reduced stability of system 5. Low Contrast Detectability (LCD) The ability to distinguish objects with subtle difference in signal intensity. CNR = S1 S2 /(St. Dev. of Noise) LCD: A visual test. Related to SNR, but more subjective. Imaging Slice LCD Measurement Holes of different size Plastic Sheet of different thickness 5

6 Total Number of Spokes 4/14/2009 The Relationship between SNR and LCD Signal-to-Noise Ratio Criteria and Cause of Failure ACR Recommend Criteria: More than 9 for 1.5T or lower More than 37 for 3.0T Reduction of 4 spokes from the baseline Possible Cause of Failure: Inaccurate slice position Poor SNR Artifacts Expanded Phantom Scans 1. ACR Sag Loc 2. ACR Ax T1 3. ACR Ax PD/T2 4. Site Brain Ax T1 5. Site Brain Ax T2 Weekly Phantom Scans 11 slices of 5 skip 5 Complete the Phantom Site Scanning Data Form (Large Phantom) online. Additional Phantom Tests Image Uniformity Use imaging intensity correction option with a phased array coil. Slice thickness and slice offset Be careful with your phantom position Ghosting artifact Checked in a recent PM or annual testing Ask your physicist to check the phantom images before ACR submission. ACR MRI Clinical Modules Head Spine MSK Body MRA Cardiac Choice of Basic and Specialty exams Each scanner must apply for all modules routinely performed ACR MRI Evaluation Categories Pulse Sequences and Image Contrast Spatial and Temporal Resolution Exam Identification: Missing Information Anatomic Coverage and Imaging Planes Artifacts Filming Technique (for hard copy film submissions only) Need good planning and team work. 6

7 ACR Accreditation Preparation ACR Online Submission Refill ACR phantom Perform PM and annual tests. Review your clinical protocols and make sure they meet ACR requirements (i.e. the resolution and scan time.) Register online and choose the type clinical cases carefully according to your patient population. Timeline for Clinical cases and phantom scans Phantom Scans 1 Notified by ACR Clinical cases (45 days) Phantom Scans 2 Material Due (45 days) Extension I (30 days) Extension II (30 days) Application Due (60 days) Tips and Suggestions Discuss with ACR if you are not clear about what they ask for. Acquisition matrix: frequency encoding and phase encoding used Reconstruction matrix: number of pixels in each dimension of the image Display matrix: number of pixels in each dimension of the post-processed image Tips and Suggestions QA for Advanced Applications Pay attention to anatomical coverage. Check image quality and present your best case. Submit all the required series including the localizer with cross-reference, but only what s required. Readable, but too much noise fmri QA Spectroscopy QA Flow Quantification QA Ask your radiologist to review the clinical images before ACR submission. 7

8 AU 4/14/2009 Challenges of fmri fmri QA Measurements Based on signal difference between rest and active states. Such difference (i.e. BOLD effect) is only a few %. While problems such as baseline drift can be corrected in postprocessing, it should still be kept to minimum Time Run extended fmri protocol with low and high excitation flip angles. Measure : SNR (RF check) SGR (Signal to Ghost Ratio) (Gradient & eddy current check) Signal intensity drifts and fluctuations. (System stability check) * A. Simmons, et al, MRM 41: (1999) fmri QA Example fmri QA Results of Two GE 3T Scanner Variation (%) Ghosting (%) Noise SNR SB w. PA < SB w. PA & IFIS SB w. BC LB w. PA LB w. BC < < < < SB=Short Bore; LB=Long Bore; PA=Phased Array; BC=Bird Cage Possible Cause of Failure Instability of RF amplifier and receiver Changes of eddy currents due to temperature change Summary Contributing factors in MR image quality. Principles behind the ACR weekly tests. ACR accreditation process. fmri QA 8

9 Thank You! 9