1、Designation: E 1935 97 (Reapproved 2003)Standard Test Method forCalibrating and Measuring CT Density1This standard is issued under the fixed designation E 1935; the number immediately following the designation indicates the year oforiginal adoption or, in the case of revision, the year of last revis
2、ion. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This test method covers instruction for determining thedensity calibration of X- and g-ray computed tomography (CT)systems and
3、for using this information to measure materialdensities from CT images. The calibration is based on anexamination of the CT image of a disk of material withembedded specimens of known composition and density. Themeasured mean CT values of the known standards are deter-mined from an analysis of the i
4、mage, and their linear attenu-ation coefficients are determined by multiplying their measuredphysical density by their published mass attenuation coeffi-cient. The density calibration is performed by applying a linearregression to the data. Once calibrated, the linear attenuationcoefficient of an un
5、known feature in an image can be measuredfrom a determination of its mean CT value. Its density can thenbe extracted from a knowledge of its mass attenuation coeffi-cient, or one representative of the feature.1.2 CT provides an excellent method of nondestructivelymeasuring density variations, which
6、would be very difficult toquantify otherwise. Density is inherently a volumetric propertyof matter. As the measurement volume shrinks, local materialinhomogeneities become more important; and measured valueswill begin to vary about the bulk density value of the material.1.3 All values are stated in
7、SI units.1.4 This standard does not purport to address the safetyconcerns, if any, associated with its use. It is the responsibilityof the user of this standard to establish appropriate safety andhealth practices and determine the applicability of regulatorylimitations prior to use.2. Referenced Doc
8、uments2.1 ASTM Standards:E 1316 Terminology for Nondestructive Examinations2E 1441 Guide for Computed Tomography (CT) Imaging2E 1570 Practice for Computed Tomographic (CT) Exami-nation23. Terminology3.1 Definitions:3.1.1 The definitions of terms relating to CT, that appear inTerminology E 1316 and G
9、uide E 1441, shall apply to theterms used in this test method.3.2 Definitions of Terms Specific to This Standard:3.2.1 density calibrationcalibration of a CT system foraccurate representation of material densities in examinationobjects.3.2.2 effective energythe equivalent monoenergetic en-ergy for a
10、 polyenergetic CT system. Thus, the actual, polyen-ergetic CT system yields the same measured attenuationcoefficient for an examination object as a theoretical, monoen-ergetic CT system at the effective energy.3.2.3 phantoma part or item being used to calibrate CTdensity.3.2.4 examination objecta pa
11、rt or specimen being sub-jected to CT examination.4. Basis of Application4.1 The procedure is generic and requires mutual agreementbetween purchaser and supplier on many points.5. Significance and Use5.1 This test method allows specification of the densitycalibration procedures to be used to calibra
12、te and performmaterial density measurements using CT image data. Suchmeasurements can be used to evaluate parts, characterize aparticular system, or compare different systems, provided thatobserved variations are dominated by true changes in objectdensity rather than by image artifacts. The specifie
13、d proceduremay also be used to determine the effective X-ray energy of aCT system.5.2 The recommended test method is more accurate and lesssusceptible to errors than alternative CT-based approaches,because it takes into account the effective energy of the CTsystem and the energy-dependent effects of
14、 the X-ray attenu-ation process.5.3 This (or any) test method for measuring density is validonly to the extent that observed CT-number variations arereflective of true changes in object density rather than imageartifacts. Artifacts are always present at some level and canmasquerade as density variat
15、ions. Beam hardening artifacts are1This test method is under the jurisdiction of ASTM Committee E07 onNondestructive Testing and is the direct responsibility of Subcommittee E07.01 onRadiology (X and Gamma) Method.Current edition approved March 10, 2003. Published May 2003. Originallyapproved in 199
16、7. Last previous edition approved in 1997 as E 1935 - 97.2Annual Book of ASTM Standards, Vol 03.03.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.particularly detrimental. It is the responsibility of the user todetermine or establis
17、h, or both, the validity of the densitymeasurements; that is, they are performed in regions of theimage which are not overly influenced by artifacts.5.4 Linear attenuation and mass attenuation may be mea-sured in various ways. For a discussion of attenuation andattenuation measurement, see Guide E 1
18、441 and PracticeE 1570.6. Apparatus6.1 Unless otherwise agreed upon between the purchaserand supplier, the density calibration phantom shall be con-structed as follows (see Fig. 1):6.1.1 A selection of density standards bracketing the rangeof densities of interest shall be chosen. For best results,
19、thematerials should have known composition and should bephysically homogeneous on a scale comparable to the spatialresolution of the CT system. It is a good idea to radiographi-cally verify homogeneity and to independently verify chemicalcomposition. All materials should be manufactured to repro-duc
20、ible standards. Solids should be readily machinable and notsusceptible to surface damage.6.1.2 One or more cylinders of each density standard shallbe machined or prepared, or both. Selecting cylinders overrectangles reduces the uncertainties and streaks that sharpcorners have on volumetric determina
21、tion and verificationmethods. The cylinders should be large enough that the meanCT number corresponding to each standard can be computedover a hundred or more uncorrupted (see 8.1.3) pixels but smallenough relative to the dimensions of the host disk that radialeffects are minimal.6.1.3 The physical
22、density of each density standard shall bedetermined empirically by weighing and measuring the speci-mens as accurately as possible. It is a good idea to indepen-dently verify the measured densities using volumetric displace-ment methods.6.1.4 The mass attenuation coefficient, /r, at the effectiveene
23、rgy of the system (see 8.3) shall be determined from areference table. For compounds, /r can be obtained by takingthe weighted sum of its constituents, in accordance with thefollowing equation:m5 /r5(iwi/r!i(1)where:wi= the weight fraction of the ith elemental component.6.1.5 For each density standa
24、rd, the measured density, r,shall be multiplied by its corresponding mass attenuationcoefficient, /r, as determined in 6.1.4. The linear attenuationcoefficient, , thus obtained shall be permanently recorded foreach density calibration standard.6.1.6 A host disk to hold the density standards shall be
25、fabricated. The opacity of the disk should approximate theattenuation range of the examination objects. If possible, thehost disk should be of the same material as the examinationobjects, but other requirements take precedence and maydictate the selection of another material.6.2 In general, it is ve
26、ry difficult to find acceptable materialsfor density standards. Published density data are generally notreliable enough for calibration purposes. Homogeneity oftenvaries on a local scale and negatively influences the calibrationprocedure. Machine damage can increase the density at thesurface of a sa
27、mple, making it difficult to determine the densityof the interior material crucial to the calibration process.Lot-to-lot variations in composition or alloy fraction can makeit difficult to compute mass attenuation coefficients. For theseand other reasons, development of a good density calibrationpha
28、ntom takes effort, resources and a willingness to iterate theselection and production of standards until acceptable resultsare obtained.6.2.1 Liquids make the best standards, because they can beprecisely controlled and measured. However, liquids requirespecial handling considerations, are sensitive
29、to temperaturevariations, and often tend to precipitate, especially high-concentration aqueous solutions. It is hard to find organicFIG. 1 Density Calibration PhantomE 1935 97 (2003)2liquids with densities above 1.5 g/cm3or inorganic liquidsabove 4.0 g/cm3; but for many purposes, they offer a suitab
30、lechoice.6.2.2 Plastics are popular but in general make the worststandards. Most plastics have at best an approximately knownpolymerization and often contain unknown or proprietaryadditives, making them poor choices for calibration standards.They also tend to vary more than other materials from batc
31、h tobatch. Notable exceptions to these generalizations are brand-name acrylics and brand-name fluorocarbons.6.2.3 Metals are also popular, but they are generally avail-able only in a limited number of discrete densities. They canexhibit important lot-to-lot variations in alloy fractions; butwith car
32、eful selection or characterization, they can make gooddensity calibration standards. Pure elements or very wellknown specimens offer an excellent option when they can beobtained in the density range of interest.6.2.4 Each material must be treated on a case-by-case basis.Reactor-grade graphite provid
33、es a good case study. Reactor-grade graphite is available in a variety of shapes, in very pureform, and in a number of densities. At first glance, it appears tooffer an attractive choice in a density range without manyviable alternatives. However, upon closer examination, thematerial is found to be
34、susceptible to surface damage duringmachining and to exhibit important inhomogeneities in densityon linear scales of about 1 mm. Surface damage makes itnearly impossible to determine the core density of the samplegravimetrically, because the total weight is biased by a denserouter shell. Inhomogenei
35、ties make it difficult to extract accu-rate mean CT numbers from an image of a sample that is notlarge in diameter compared to 1 mm.7. Procedure7.1 Unless otherwise agreed upon between the purchaserand supplier, the density calibration phantom shall be scannedas follows:7.1.1 The phantom shall be mo
36、unted on the CT system withthe orientation of its axis of revolution normal to the scanplane.7.1.2 The phantom shall be placed at the same location usedfor examination object scans.7.1.3 The slice plane shall be adjusted to intercept thephantom approximately midway between the flat faces of thedisk.
37、7.1.4 The phantom shall be scanned using the same dataacquisition parameters, and the data shall be processed usingthe same steps (for example, beam-hardening corrections)applied to examination objects.8. Interpretation of Results8.1 Unless otherwise agreed upon between the purchaserand supplier, th
38、e image of the density calibration phantom shallbe analyzed as follows:8.1.1 The phantom scan data shall be reconstructed usingthe same reconstruction parameters and post-processing steps,if any, used for examination object data.8.1.2 The phantom image shall be displayed using the samedisplay parame
39、ters used for viewing examination object im-ages.8.1.3 The mean CT numbers of the density standards in theCT image shall be measured. Special attention needs to be paidto this part of the measurement process. As much of the area ofeach specimen as practical should be used, but care must betaken to i
40、nsure that only valid pixels are included. For example,a square region of interest in a round sample could yield biasedresults if there are significant radial effects, such as from beamhardening or a higher density around the perimeter due tosurface damage caused by machining or compression. Ideally
41、,a circular region of interest should be used that includes ahundred or more pixels but avoids the boundary region aroundeach density standard, especially if edge effects of any type areclearly visible.8.1.4 A table of linear attenuation coefficients versus meanCT numbers shall be prepared.8.1.5 A l
42、east-squares fit to the equation NCT= a + b shallbe performed on the data stored in the table, where is thelinear attenuation coefficient and NCTis the CT number.8.1.6 The resulting linear curve shall be used as the densitycalibration. Using the inferred linear relationship between CTnumber and line
43、ar attenuation coefficient, the measured CTvalue, NCT, of any material can be used to calculate a bestestimate of its associated linear attenuation coefficient, .8.2 Unless otherwise agreed upon between the purchaserand supplier, the density of a region of interest in an exami-nation object shall be
44、 determined as follows:8.2.1 The mean CT number in the region of interest shall bemeasured.8.2.2 From the known calibration parameters, the linearattenuation coefficient of the region of interest shall be ob-tained using the equation NCT= a + b.8.2.3 The density of the region of interest shall be ca
45、lculatedby dividing the obtained linear attenuation by the appropriatetabulated value of /r at the effective energy of the system (see8.3). If /r is not known for the feature of interest, a nominalvalue for /r may be used. Variations in /r are minor, andbasically independent of material in the energ
46、y range of about200 keV to about 2 MeV. Outside this range, the selection of anominal value is more sensitive. Adoption of an appropriatenominal value is a matter of agreement between purchaser andsupplier.8.3 Unless otherwise agreed upon between the purchaserand supplier, the effective energy of th
47、e CT system shall bedetermined as follows:8.3.1 A table of linear attenuation coefficients versus meanCT numbers shall be prepared for several X-ray energiesbracketing the effective energy of the CT system, as shown in8.4.1.8.3.2 For each X-ray energy, a least-squares fit to theequation NCT= a + b s
48、hall be performed and the correlationcoefficient recorded.8.3.3 The energy value in the table that yields the best fit(that is, the largest value of the correlation coefficient) shall beselected as the effective energy of the CT system.8.3.4 If the effective energy has been determined previouslyunde
49、r the same or similar conditions, this step may be skippedwith the consent of the buyer.8.4 Illustrative Examples:E 1935 97 (2003)38.4.1 Effective Energy DeterminationThe process of de-termining the effective X-ray energy of a CT system isillustrated with the following example taken from actualpractice. The data presented were acquired with a 9-MVindustrial CT system. Three materials (see Table 1) were used:methyl methacrylate, tetrafluoroethylene, and aluminum. Theempirical densities of the specimens, not the published values,were determined by accurately mea
