Acceptance Testing and QC of Digital Radiography Units Douglas Pfeiffer, MS, DABR Boulder Community Hospital Outline Status of Task Groups Acceptance testing Quality Control Technologist Physicist Relevant Task Groups TG 150 “Acceptance Testing and Quality Control of Digital Radiographic Imaging Systems” Physicist stuff TG 151 “Radiographic System Quality Control” Technologist stuff TG 116 “DR Exposure Index Task Group” Task Group Charges TG 150: This group will outline a set of tests to be used in the Acceptance Testing and Quality Control of Digital Radiographic Imaging Systems. TG 151: Task Group to recommend consistency tests (one or more) designed to be performed by a Medical Physicist, or a radiologic technologist under the direction of a medical physicist, to identify problems with an imaging system that need further evaluation by a medical physicist. TG 116: To identify a method of providing feedback, in the form of a standard index, to operators of DR systems, which reflects the adequacy of the exposure that has reached the image receptor (IR) after every exposure event. TG 150 schedule Should be competed in about 2 years May have educational session RSNA 2008 AAPM 2009 Preliminary guidelines may be provided via MedPhys TG 150 Working in concert with TG 151 Have sent to vendors a survey regarding the QC recommendations for their systems Comprehensive list of product specifications is being developed (living) Concern regarding access to “for processing” or earlier images TG 150 Determining what corrections are allowed in a “for processing” image (IEC 62220-1) Flat field Geometric distortion Reference list being compiled TG 151 Pretty much the same schedule as TG150… Compiling a summary of QC tools/programs provided by manufacturers of radiographic systems or by third parties TG 116 Draft written 11/2007 The goal of TG 150 is to 0% 81% 6% 13% 0% 1. 2. 3. 4. 5. write code for the automated evaluation of digital QC images establish standardized tests for digital imaging equipment compel digital imaging system manufacturers to implement standardized QC in their systems validate the QC programs currently provided by digital imaging system manufacturers define a technologist-level quality control program for digital imaging equipment The goal of TG 150 is to 1. 2. 3. 4. 5. write code for the automated evaluation of digital QC images establish standardized tests for digital imaging equipment compel digital imaging system manufacturers to implement standardized QC in their systems validate the QC programs currently provided by digital imaging system manufacturers define a technologist-level quality control program for digital imaging equipment Charge of AAPM TG 150, http://aapm.org/org/structure/default.asp?committee_code=TG150, and Ranger, private communication. Acceptance Testing Inventory: did you get what you ordered (features, accessories) Installation: is it correctly placed in the room Function: does it work as advertised Clinical use: are the images acceptable Inventory Hardware Grids (40”, 72”) Cassettes Test tools Software Special processing QA tools Installation According to plan Any unforeseen impediments Shielding (x-ray, after all) Function Get performance specifications (tee hee!) Laser printer Workstation calibration System interfaces Beam characterization / Generator calibration Exposure index calibration Exposure index linearity Limiting resolution / MTF Noise and low contrast Geometric accuracy Throughput Plate erasure (CR) Automatic exposure control calibration Artifacts Laser Printer Values may be specified by modality manufacturer LUT curve is modality-specific End result should look like workstation Typical values and ranges Dmin: 0.05 ± 0.03 Low density: 0.45 ± 0.07 Mid density: 1.20 ± 0.15 High density: 2.20 ± 0.15 Workstation Calibration AAPM Task Group 18 Calibrated to DICOM Gray Scale Display Funtion Luminance and illuminance measurements are required TG-18QC Test Pattern best for overall check System Interfaces Verify PACS, RIS communications NOW, prior to clinical use No worky later – you have VERY grumpy staff! Standard Radiographic Testing Light field / radiation field Electronic Radiochromic film Image receptor / radiation field Direct: beam centering Bead at center of receptor Collimate smaller than detector area CR: same as screen film Standard Radiographic Testing Generator calibration Output mR/mAs linearity HVL kVp accuracy Reproducibility PBL function (CR) Light field intensity SID Beam Characterization Exposure index calibrations are exposure dependent Agfa CR: 75 kVp, 1.5 mm Cu Fuji CR: 80 kVp, no added filtration Kodak CR: 80 kVp, 0.5 mm Cu + 1 mm Al TG 116: 80 kVp, 1 mm Al + 1 mm Cu + 1 mm Al HVL requirements 0.1, 1.0, 10 mR Tight tolerances: 10.0 mR ± 0.2 @ 80 kVp ± 1 SID (heel effect) mAs 60 kVp, 5 mR, unfiltered From: TG116_v9c Exposure Index Calibration TG 116 Avoid “speed class” – different world now (“acts sort of like a 200 speed system”) KIND – Indicated Equivalent Air Kerma: “an indicator of the x-ray beam air kerma, expressed in µGy, that is incident of the digital detector and used to create the radiographic image” DI – Deviation Index: the relative deviation from the value targeted (KTGT) by the system for a particular body part and view K DI = 10 × Log10 [ IND K TGT ( b ,ν ) ] Exposure Indexes From: TG116_v9c Exposure Index Linearity Should match manufacturer equation over at least 3 orders of magnitude Recommend 0.1 mR, 1 mR, 10 mR Wait 10 minutes before processing CR plates Manufacturer-specific tolerances Kodak: nominal EI ± 100 The exposure index 94% 1. Varies from manufacturer to manufacturer 6% 2. Is required for FDA approval of a system 3. Is automatically flagged when an exposure is out of range 4. Allows a patient to see what his dose was 5. Is prominent in the DICOM header 0% 0% 0% The exposure index 1. 2. 3. 4. 5. Varies from manufacturer to manufacturer Is required for FDA approval of a system Is automatically flagged when an exposure is out of range Allows a patient to see what his dose was Is prominent in the DICOM header Williams, et al, Digital Radiography Image Quality: Image Acquisition, J Am Coll Radiol 2007;4:371-388. Limiting Resolution / MTF MTF is preferable Difficult in the field Need manufacturer software 60 kVp, 5 mR Patterns up to 5 lp/mm, slight angle (2-5°) No edge enhancement Visual performance within 10% of manufacturer specification Verify at center and periphery mammography contact test tool Measure for all matrix/plate sizes Limiting Resolution Noise and Low Contrast Low contrast is easiest field measurement “UAB”, CIRS Model 903 Noise and Low Contrast Low contrast performance should improve with increased exposure Kodak CR950 SN 18 x 24 GP 9101061800 24 x 30 GP 9102064558 35 x 43 GP 9104078405 Minimum detectable contrast 0.1 mR 1.0 mR 10 mR 2.0% 1.5% 1.5% 1.1% 1.1% 1.1% 0.6% 0.6% 0.6% Noise and Low Contrast Noise should vary predictably for the system being tested Use Linearity images (0.1 mR, 1 mR, 10 mR) Can extend to improve fit Plot STDEV vs Exposure (or some function thereof) For example: Fuji CR 5000 plot (stdev)2 vs. 1/exposure least squares fit slope = 1.8 – 2.0 r2 > 0.98 Noise and Low Contrast Kodak CR 975 Pixel value ∝ log(E) SD(ROI) ∝ E-0.5 Noise vs Input Exposure 100 -0.3311 Noise (SD) y = 9.2215x 2 R = 0.9853 10 1 0.1 1 Exposure (mR) 10 Geometric Accuracy Image should not be morphed in any way Two radiopaque objects of known length (or one 2D object of known dimensions) Measured should be accurate to within 1% (Never found a problem) Nominal = 152.4 mm Throughput System may be bogged down due to faulty controlling computer, hardware or electronic malfunction Multiple readouts in rapid succession 10 mR Direct: 5 exposures Single plate CR: 5 plates Multiplate CT: max feed loading Throughput Measure time from beginning of first readout to end of last readout Direct: end of first exposure to able to make a sixth exposure CR: first plate into reader to last plate out of reader Throughput (readouts/hour) = 60 x N / t Should be within 10% of manufacturer specification Plate Erasure (CR) Exposure IP of each size to 10 mR Erase Reprocess Compare exposure index to manufacturer specifications Kodak: EI < 100 AEC Calibration CR response is different from S-F Energy dependence Determines typical exposure index and patient dose Proper gain setting may take some time Lowest possible for NOISE tolerance AEC Calibration Uniform phantom 2” – 12” equivalent PMMA (oofdah!) 60 – 120 kVp Cell selection Cell balance Artifacts CR Screens Dust, etc in optical path Mechanical motion of slow, fast scan mechanisms Image processing glitches Direct Flat fielding errors Detector faults Image processing glitches Use large SID to limit heel effect Artifacts Ghost! Ghost image was left after one hearty exposure Imagine the result after generator testing Always cover the detector with at least 0.5 mm lead (more is better) Artifacts Screen Cleaner! Recommend against Kodak Premoistened wipes Clean only when needed Use lint-free cloth, barely damp with water (at least for Kodak) Artifacts EI = 250 EI = 430 2 days post-erasure 4 days post-erasure Background Fog! Natural background radiation is ~0.5 mR/day (Plates not sensitive to much of this) Typical diagnostic plate exposure ~ 1 mR Recommend using or erasing cassettes every 2-3 days Clinical Phantom image quality DOSE Exposure index ranges Will take some time to develop Exam-specific Can use range of about factor of 1.75 (sort of equivalent to the S-F ± 0.3 OD variation) Phantom Image Quality System performance Clinical-esque Anthropomorphic Use your favorite Clinical technique Regarding comprehensive acceptance testing of digital imaging systems, all of the following are true EXEPT: 1. 75% 19% 2. 3. 0% 4. 6% 5. 0% It is not needed due to the automatic calibration and self-monitoring of modern systems It verifies correct installation and calibration It sets baseline values to be used for future testing It provides an opportunity to train technologists in appropriate QC It is the time when radiation shielding should be verified Regarding comprehensive acceptance testing of digital imaging systems, all of the following are true EXEPT: 1. 2. 3. 4. 5. It is not needed due to the automatic calibration and self-monitoring of modern systems It verifies correct installation and calibration It sets baseline values to be used for future testing It provides an opportunity to train technologists in appropriate QC It is the time when radiation shielding should be verified Williams, et al, Digital Radiography Image Quality: Image Acquisition, J Am Coll Radiol 2007;4:371-388. Quality Control Technologist TG 151 Physicist TG 150 Technologist QC (TG 151) Daily Look for artifacts Monthly Laser printer sensitometry Self-calibration module may fail! Quarterly Workstation QC (monthly?) TG-18QC, SMPTE Repeat analysis Per manufacturer Test object Flat field Repeat Analysis Use ranges established at acceptance period Most repeats should be positioning, motion, etc. “Odd” images can often be saved with window width / level or reprocessing Check with radiologist if Exposure Index too low NEVER repeat for Exposure Index too high (unless saturated) one facility had it as a policy! Physicist (TG 150) Laser printer – review QC Workstation calibration Beam characterization Exposure index calibration Exposure index linearity Limiting resolution / MTF Noise and low contrast Physicist (TG 150) Plate erasure (CR) Artifacts Phantom Exposure Index review Generator calibration? Dose Creep Not the guy who monitors the doses! Recall that images just get better with higher dose “I’ll just up my mAs a bit to be sure” AEC systems calibration check regularly Monitor the indexes! Generator Calibration? Self-calibrating during the exposure HVL change?? Anything funky will show up in dose IMAGE QUALITY DOSE Dose creep, the gradual increase in patient doses, 17% 8% 75% 0% 0% 1. 2. 3. Was a problem with screen-film imaging also Exists only in computed radiography departments Is due to the decrease in noise with increased detector exposure 4. Is easily recognized by the radiologist 5. Is not impacted by the presence of technique charts Dose creep, the gradual increase in patient doses, 1. 2. 3. 4. 5. Was a problem with screen-film imaging also Exists only in computed radiography departments Is due to the decrease in noise with increased detector exposure Is easily recognized by the radiologist Is not impacted by the presence of technique charts Williams, et al, Digital Radiography Image Quality: Image Acquisition, J Am Coll Radiol 2007;4:371-388. Helpful Tools Some manufacturers provide proprietary tools to assist… Fuji GE Flat field images taken prior to other image quality tests, as driven by the software. GE Flat Field for processing pretty darn nice! GE GE Kodak Kodak Kodak References AAPM Monograph 20: Specification, Acceptance Testing and Quality Control of Diagnostic X-ray Imaging Equipment (1991) AAPM Report 74: Quality Control in Diagnostic Radiology (2002) AAPM Monograph No. 30: Specifications, Performance Evaluations and Quality Assurance of Radiographic and Fluoroscopic Systems in the Digital Era (2004) IPEM Report 91 Recommended Standards for the Routine Performance Testing of Diagnostic X-Ray Imaging Systems (2005) Frey, Changes in Equipment Acceptance Testing, J Am Coll Radiol; 2005; 2:639-641 AAPM Report 93 CR Acceptance Testing and QC (2007) Williams, et al, Digital Radiography Image Quality: Image Acquisition, J Am Coll Radiol 2007;4:371-388. Exposure Indicators The Digital Imaging Chain Diagnostic Exposure Detection (Lat. Image) mAs too high Capture (Sampl./Quant.) Processing Output/Display (Hardcopy/Softcopy) Signal CV Image Processing mAs optimal CV logE mAs too low Imaging System Code Values Code Values Code Values Film-Screen World Adequate film optical density guaranteed adequate dose: Papa Bear: mAs too high, dose too high, film too dark. Mama Bear: mAs too low, dose too low, film too light. Baby Bear: mAs OK, film OD OK, dose OK. The film-screen characteristic (H&D) curve defined the proper exposure range. What’s the Fuss About? Since 1897, we have gotten by just fine in film-screen without a universal dose parameter for every image. Why is this all of the sudden a problem for digital imaging? In The Digital World: The good news: Digital systems produce diagnostically useful images over a very wide exposure range. Images always have adequate brightness. The bad news: Digital systems produce diagnostically useful images over a very wide exposure range. Images always have adequate brightness. In the digital world: Since the always-acceptable brightness masks the dose (mAs) used, the chance for producing images with too little or too high dose is increased. Docs will either: suffer too much noise (i.e. missed diagnosis if dose too low), or will be very pleased with absence of noise (but patients suffer high dose). Film/Screen Above optimal exposure Optimal exposure Below optimal exposure Digital Effects of Dose on Digital Image Quality Image brightness does not change with dose: Double the mAs still gives same brightness. What DOES change with dose are the noise properties of the image. 0.1 mR 0.3 mR 1.0 mR Not Your Daddy’s S-F Image brightness is always adequate, regardless of dose. Image brightness does not depend on mAs (an mAs range over 100x greater than the acceptable mAs range for film/screen). Adequacy of dose manifests as adequacy of image quality – dose does not manifest as brightness. The Answer My Friend Film/screen was able to get by without any dose parameter to accompany each image, because film/screen was a self-regulating system. Digital is not self-regulating, and the chance of missed diagnosis, or excessive dose, is too great without having a dose parameter accompanying each image. The Way We Were Each hospital selected a film-screen system based on their desired operating speed class Speed = 200 / (mR required for OD=1) if 0.5 mR radiation exposes film to OD=1: Speed = 200/(0.5) = 400 if 1 mR radiation exposes film to OD=1: Speed = 200/(1) = 200 if 2 mR radiation exposes film to OD=1: Speed = 200/(2) = 100 New World Symphony In digital: Each hospital decides on desired speed class for each exam type (AEC calibration / technique chart) Noise Dose The manufacturer calibrates so that digital exposure value relates to the incident exposure If vendor does not use a radiation meter to calibrate, then dose indicator is worthless. DR Patient Exposure Indicators Canon Reached Exposure Value REX GE Dose Area Product DAP, dGy*cm2 & skin dose mGy Hologic DAP & Exam Factor, Center of Mass of log E Histogram Imaging Dynamics Corporation, IDC Dose parameter, median exposure histogram f # ~ dose/target Philips/Siemens/Thompson (Trixel) Dose Area Product DAP, dGy*cm2 also Exposure Indicator or EI, ~ Speed Class SwissRay mA, sec, field size, kVp, no exposure indicator What is Dose Area Product (DAP)? DAP = (entrance dose R, Gy) * (field size, cm2) If you double the patient’s skin entrance dose, e.g. double the mAs, the DAP also doubles. If you double the x-ray field size (e.g. open the collimators), the DAP will also double. Measured by detector inside collimator, so DAP could be part of patient’s image record in DR. But defining a proper value of DAP is problematic because DAP depends on collimation. CR Exposure Indicators •Agfa •LgM, Logarithm of the Median of the histogram •+0.3 LgM = 2x exposure, –0.3 LgM = 1/2x exposure •Fuji •S number, Sensitivity Number •200/S proportional to exposure •Kodak •EI, Exposure Index •+300 EI = 2x exposure, –300 EI = 1/2x exposure •Konica •S value, similar to Fuji Appropriate Values for Digital Dose Indicators (for a 200 speed class exam) Exposure (mR) Fuji (S) Kodak (EI) Agfa (LgM) IDC/DR f# Low 0.5x optimal 0.5 400 1700 1.9 -1 Optimal 1 200 2000 2.2 0 High 2x Optimal 2 100 2300 2.5 +1 AEC Calibration 200 speed appropriate? Can modify for diagnostic requirements Energy response is different from screenfilm Each manufacturer has specific calibration methods Quality Control Three levels of system performance quality control 1. Routine: Technologist - no radiation measurements 2. Full inspection: Physicist - radiation measurements and non-invasive adjustments 3. System adjustment: Vendor service - hardware and software maintenance Periodic Quality Control (RT) Daily Inspect CR system interfaces Erase image receptors Weekly / Biweekly Calibrate monitors (SMPTE) QC phantom test image performance Check / clean PSP screens and cassettes Quarterly Review image retakes and exposures Update QC log. Review out-of-tolerance issues Periodic Quality Control Annually (Physicist) Perform linearity / sensitivity / uniformity Inspect / evaluate image quality Re-establish baseline values Review retakes, service records What is needed? Computer friendly phantoms Objective quantitative analysis methods System performance tracking and database logs Exposure monitoring tools with database Quality Control Tracking the indexes Not “Too dark or too light?” “Is the index appropriate?” May shift techniques from what you’re used to overall higher (200ish vs. 400ish)? increase kVp? Quality Control Control Limits Know what is appropriate for your facility work with radiologists, RT’s, manufacturer Know what a change in your index means a change of 300 for Kodak is a factor of 2 in exposure Set appropriate limits what’s a reasonable, acceptable variation? what impacts the number (collimation, etc.) DO NOT REPEAT IF THE DOSE IS TOO HIGH!!! Quality Control Repeat analysis DO NOT REPEAT FOR HIGH DOSE (no dark films) Light films Low (or high) exposure index Collimation Positioning Artifacts (junk on screen, mainly) Frequency Components sin(x) sin(3x) sin(5x) sin(7x) sin(9x) sin(x)+sin(3x) +sin(5x) +sin(7x) +sin(9x) MTF Comparison Image Quality Descriptors “Resolution” MTF Noise DQE “Resolution” Pixel size? Misleading MTF? Better, but incomplete Resolution and digital sampling MTF of pixel (sampling) aperture Modulation 1 0.9 0.8 0.7 0.6 0.5 50 µ m 100 µ m 0.4 0.3 0.2 0.1 200 µ m 0 0 2 4 6 8 10 12 14 16 18 20 22 Frequency (lp/mm) Detector pitch Detector aperture Cutoff frequency = 1 / ∆x MTF of sampling aperture Nyquist (max usable) frequency = 1/2∆x, when pitch = aperture Line Pair Resolution MTF: object signal MTF at 4.5 cm above stand Scan Perfect resolution: 50 µm pixel a-Se 70 µm pixel Subscan CsI 100 µm pixel Microcalcifications?? Noise mAs Electronic Putting It Together Detective Quantum Efficiency almost universally regarded as the best overall indicator of the image quality of digital radiography systems NPS(f) = Noise Power Spectrum: noise is frequency dependent G = a magic number