Electron Density Measurement With Dual-energy CT:Phantom Study

In X(γ)-ray radiotherapy,CT images provide information not only about target volumes but about critical(normal)organs as well.CT images also provide tissue information for heterogeneity-based dose calculations.The conventional method is the stoichiometric calibration method by which one can establish the CT values-electron density(ρ_e)conversion using an electron density phantom.For the diagnostic X-ray,tissue attenuation coefficient depends on the electron density and effective atomic number(Z eff).There is no one-to-one correspondence between CT value and ρ_e.Studies have shown that dose errors in some regions like head and neck can be more than 40% using the stoichiometric calibration method.Fortunately,dual-energy computed tomography(DECT)acquiring CT data sets at two different X-ray spectra can fundamentally measure the electron density and effective atomic number directly and without the calibration process.In recent years,with the development of DECT technology,radiotherapy simulation studies have made a breakthrough with ρ_e errors less than 10% based on clinical DECT scanners.In the future DECT simulation,the role of a phantom will change completely.A phantom is used only to determine the parameters in a DECT-ρ_e or Zeff parametric equation.However,for the current researches,experiments are generally conducted using the tissue-equivalent materials,such as CIRS 062 and Gammex 467.Therefore,the purpose of this study is to propose a phantom prototype for DECT simulation,and to develop the prototype according to the law of phantom research.Then accurate DECT-ρ_e conversion can be implemented.In the first part of the dissertation,the performance difference of electron density measurement between the diagnostic CT,big core CT,and a Hi-ART tomotherapy MVCT scanner was compared.According to the interaction between 3.5 MV X-ray and matter,a linear model of attenuation coefficient with electron density is obtained.The CIRS 062 phantom was scanned on a MVCT scanner,a GE Light Speed 16 scanner and a Siemens SOMATOM Sensation Open scanner.Errors of ρ_e are calculated according to the fitting equation for MVCT.The CT value-ρ_e conversion curves are established according to the stoichiometric calibration method for the diagnostic CT and big core CT.The determination coefficient of the linear fitting for MVCT values and nominal ρ_e,r2=0.9988.The relative errors of ρ_e calculated values were within ±2% except for lung equivalent insert of the CIRS 062 phantom(-7.16%).The CT value of the k VCT image increases with ρ_e,showing a linear segmental relationship.When ρ_e <1.1,the corresponding CT value is not very different;and when ρ_e > 1.1,the CT value of big core CT image is gradually larger than that of diagnostic CT image.Especially for the CT value of the dense material,such as bone,CT value difference is about 100 HU or more.There is no difference in the CT-ρ_e conversion curves for the two kVCT.The MVCT value-ρ_e calibration line with a single linear relationship would facilitate the construction of a well-calibrated ρ_e image from acquired MVCT image.Tomotherapy MVCT may be used for dose calculation.But from MVCT image one cannot get the Zeff.In the second part,a prototype of DECT simulated phantom is proposed.According to the characteristics of DECT simulation,the phantom prototype chose iodine as the photoelectric absorption basis,and water and ethanol as Compton scattering basis.The phantom prototype’s parameters and element composition had no corresponding relationship with the tissues.The phantom prototype and a CIRS 062 phantom were scanned on a Siemens SOMATOM Definition Flash scanner,respectively.Then the differences of ρ_e between the two measurements were compared.The results show that there is a linear relationship between ΔHU and ρ_e(r2 = 0.9962,w =-0.428).The coefficients(a = 0.972,b = 0.995)are consistent with the theoretical values.The ρ_e mean error and relative standard deviation of the phantom prototype are lower than those of the CIRS 062.The Bland-Altman plot shows that the ρ_e measurements are correct.Like a CIRS 062 phantom,the phantom prototype can also be used to determine the parameters of ΔHU-ρ_e model proposed by Saito.The feasibility of the DECT simulated phantom is verified.In the third part,the phantom is constructed according to the principle of DECT basis material decomposition.Its electronic density and effective atomic number are consistent with the range of human tissue.A DECT simulated phantom was prepared by using a variety of materials with atomic numbers above 10 as the photoelectric absorption basis,and materials with atomic numbers below 10 as the Compton scattering basis.The DECT simulated phantom was scanned on a Siemens SOMATOM Definition Flash scanner.The accuracy of the measured ρ_e was analyzed.The results show that there is a linear relationship between ΔHU and ρ_e(r2 = 0.9690,α =-0.489).The coefficients a and b are 0.942 and 0.986,respectively.The ρ_e and Zeff of the in-house DECT simulated phantom are between 0.8 to 1.3 and 6.4 to 13.1,respectively.The phantom achieves radiation equivalent to human tissue parameters in a certain degree.The results showed that accuracy of ρ_e measurement was within 10%,which was not 1% as reported in the literatures.The accuracy of ρ_e measurement using DECT needs to be further improved.In the fourth part,the influence of the size of the phantom profile on the CT value and ρ_e measurement is discussed from the perspective of beam hardening effect.According to the BMI(body mass index),three groups of patients examined by routine upper abdominal DECT protocol and with image data were collected in Southwest Hospital.CT differences in patients with different sizes were analyzed.On this basis,the design of a cone-shaped water phantom,whose cross-section diameter is 22 ~ 30 cm.Inserts are deposited the inner and outer rings(4.5,9.0 cm from the center,respectively)on a disc,which is fix ed in the water phantom.Ethanol,water,ethylene glycol,glycerol,calcium chloride solution(3 g / 5 m L)were selected as the phantom inserts.SECT scanning and DECT scanning are performed on the Siemens SOMATOM Definition Flash scanner.SECT scanning vo ltages selected 100,120 k Vp;DECT scanning voltage pairs were 80/140 Sn,80/140,100/140 Sn k Vp.The results show that the CT value of the material with a smaller density than water increases with the increase of the diameter of the cone-shaped phantom.The CT value of the substance with larger density than that of the water decreases with the increase of the diameter.In the inner and outer rings,the CT value of the change is different,the inner ring’s changes more obvious,about 2 times the outer ring.The change in CT values for different substances is proportional to Zeff.The Saito model coefficients α = 0.482,0.485,0.476,0.494 and 0.492,respectively,at different diameters,α shows a tendency to increase with the increase of the diameter.When the phantom diameter corresponds to α,ρ_e measurement error is within ±0.5%.If the value of α is not appropriate,it will cause ρ_e measurement error increased.The larger the size of the phantom,or the greater the atomic number of the substance,the greater the impact on the CT value.The change in phantom size affects the determination of the ρ_e model of the DECT measurement.Population of different sizes should have a corresponding phantom to match.