Partial Inspection and Dual-energy CT Material Analysis with 950keV/3.95MeV X-Band Linac
公開日:
カテゴリ: 第10回
The University of Tokyo Wenjing WU The University of Tokyo Haito ZHU The University of Tokyo Ming JIN The University of Tokyo Katsuhiro DOBASHI The University of Tokyo Takeshi FUJIWARA The University of Tokyo Mitsuru UESAKA Member
Accuthera Int. Joichi KUSANO Accuthera Int. Naoki NAKAMURA Accuthera Int. Eiji TANABE Riken Hideyuki SUNAGA Riken Yoshie OTAKE
1.Introduction
In Japan many bridges suffering degradation problems due to long years’ service which results to high demand about in situ NDE techniques. Since conventional NDE methods have limitation to inspect inner structure, Computer Tomography becomes a better choice to detect inner steel rod. Portable 950kwV/3.95MeV X-band linac would be adopted as beam source complying with that radiation source under 4 MeV is legally permitted to work outside controlled area.[1] Partial projection problem with limited scanning angle range should be studied because of constraints of bridge shape. With dual-energy CT analysis, the erosion thickness of inner steel rods can be estimated based on different density and attenuation character between iron and iron dioxide. After th, VCAD software, developed by Riken, Japan, is adopted for modeling and further structural analysis. 3D model of inspected module will be built with reconstructed sectional images so that original bridge model can be amended by substituting degradation module. Structural analysis is carried out hereafter with reduced stiffness to evaluate mechanical characters of aging structure. Fig.1 Schematic view of CT inspection system In previous work, a small concrete sample with inner steel rods was scanned and CT reconstruction was carried out with Filtered Back Projection algorithm. However, due to serious noise caused by low energy scattered X-ray, the inner ste rods were not showed clear enough for 3D modeling software VCAD to distinguish the color.[2]
New experiment adopting line sensor with collimators is implemented, as explained in next section, followed by partial CT study. Then in the fourth section, primary simulation work of dual-energy CT for component analysis is presented. 2．CT Experiment with Collimators IIC Todai Hybrid Scintillator is a line sensor compacted with collimators to keep effective incident beam angle as showed in figure 2. The small sample is a concrete cylinder (height: 10mm; diameter: 10mm) that contains steel rods inside it and the figure 3 showed the scanning experiment layout using 3.95MeV linac.[3] Fig.2Line sensor with collimators Fig.3CT scanning experiment Fig.4Transmission images in different directions From transmission images, two steel rods can be confirmed inside the sample, one flat and the other inclined. Comparing the reconstructed images by Filter Back projection algorithm in figure 5, the steel rods have become much clearer than previous results owing to collimators effect. 64 section slices of the whole sample is reconstructed with projection data and 3D model is successfully built by VCAD software. [4] max-sino_52_allmax-sino_10_all(a) (b) (c) (d) Fig.5Reconstructed sectional images (a,b: with collimator; c,d: without collimator; a,c: flat inner steel rod; b,d: inclined inner steel rod) C:\Users\wuwenjing\Desktop\VCAD\3D_linesensor130515\sample.pngFig.63D model of scanned sample (red: steel rods; blue: concrete) 3．Partial CT Reconstruction When reconstructing with incomplete projection data, artifacts would deteriorate image quality. Partial CT reconstruction is studied with experimental data to explore how much influence there is for inner steel rods. 120d120d_10(a) (b) 90d90d_10(c) (d) Fig.7Reconstruction with limited angle range (a,b: 120degree; c,d: 90degree; a,c: flat inner steel rod; b,d: inclined inner steel rod) When reconstructing with data of limited projection angle range, the sample edge shape was damaged seriously and apparent shadow appears around missing angles. The inclined steel rod was damaged a little more serious than flat steel rod with shadow because it located near edge. per8_52per8_10(a) (b) per16_52per16_10(c) (d) Fig.8Reconstruction with few projections (a,b: 8degree increment; c,d: 16degree increment; a,c: flat inner steel rod; b,d: inclined inner steel rod) When increasing projection increment and reconstructing with data of fewer projections but still covering full angle range, apparent line shadow due to incomplete data appears. The steel rod’s diameter and shape was affected especially as increment increased. Inner steel rod is still discernible in reconstructed images under partial projection conditions but the shape is deteriorated to some extent especially under severe lack of projection data. Increasing projection increment properly could be one approach to reduce scanning time but the increment degree should be well controlled. On the other hand, some improvement for projection data recovery algorithms is under discussion such as iterative analytic reconstruction. 4．Dual-energy CT simulation 4.1Method of Material Estimation by Dual-energy CT CT image is representation of linear attenuation coefficientμ, which denotes how strongly the media material absorbs or scatters beam light. This character can be described by the mass attenuation coefficient μ/ρwhereρis the density.[5] Each kind of element media shows distinctive mass attenuation coefficient. For heterogeneous material, the mass attenuation coefficient is defined as below and Wi is fraction by weight of ith atomic constituent: ..miiiiW......At different energy level, the measuredμvalue would vary for same media. Therefore for μ1andμ2at two different energies, we have ....1111222()fZFZZFfZ...............Here ......,iifZwEEZdE..........F(Z) shows as monotonic function withμ1/μ2, so we can get Z=F-1(μ1/μ2) through numerical interpolation. Relatively, density can be obtained by ....112iiiifZfF............For heterogeneous material, atomic number Z and density ρinclude effectiveness of all compound elements: kiikiiiiiZZ.........4.2 Simulation of Dual-Energy CT In the simulation sample there is one steel rod inside concrete with erosion coating with thickness of half its radius. The model is similar with figure 3and EGS5 simulation is calculated about spectrum and detector efficiency. Noise elimination in air region is realized by resetμvalue of air. Fig.9Spectrum energy Fig.10Detector efficiency Fig.11Simulation sample Fig.12Estimation result (left: atomic number; right: density) For steel rod with erosion coating Fe and Fe2O3can be identified. The estimation accuracy is acceptable, for estimation of material atomic number Z within 0.2-0.5and for density ρwithin 0.3 g/cm3. Validation with Experimental data is expected in future work. 5．Conclusion The linac CT system for in situ NDE of bridge is improved for reducing low energy scattered X-ray noise by collimator. Through reconstruction with incomplete projection data, the identification of steel rod is discussed. Lack of data would cause artifacts and shape of inner steel rod would be deteriorated seriously when too much projection data is missing. Data recovery algorithm will be added to the system. Additionally, material estimation with dual-energy CT is programed and tested that Fe and Fe2O3 can be well identified for erosion evaluation. Relative experiment is under plan. References [1] M. Uesaka, T. Natsui, et al.: ““950keV, 3.95MeV and 6MeV X-band linacs for nondestructive evaluation and medicine““, (2011), Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 657:1, 82-87.(2011) [2] Wenjing Wu, Haitao Zhu, et al.: ““Partial CT and Structural Analysis with 950 Kev/3.95 Mev X-Band Linac X-Ray Sources““, (2013), E-Journal of Advanced Maintenance, 5:2, 2013) [3] K. Lee, Dual energy X-ray imaging for material recognition using 950 keV X-band Linac, Doctoral dissertation, The University of Tokyo, 2010. [4] M. Akitake: ““VCAD System: Advanced Software System for Production Engineering and Biological Research““, (2010), JSCES, 15:2, 2309-2315.(2010) [5] B. J. Heismann, Leppert, J., Stierstorfer, K.: ““Density and atomic number measurements with spectral x-ray attenuation method““, (2003), Journal of Applied Physics, 94:3, 2073-2079.(2003)
“ “Partial Inspection and Dual-energy CT Material Analysis with 950keV/3.95MeV X-Band Linac “ “Wenjing WU,Haito ZHU,Ming JIN,Katsuhiro DOBASHI ,Takeshi FUJIWARA,Mitsuru UESAKA,Joichi KUSANO,Naoki NAKAMURA ,Eiji TANABE ,Hideyuki SUNAGA ,Yoshie OTAKE “ “Partial Inspection and Dual-energy CT Material Analysis with 950keV/3.95MeV X-Band Linac “ “Wenjing WU,Haito ZHU,Ming JIN,Katsuhiro DOBASHI ,Takeshi FUJIWARA,Mitsuru UESAKA,Joichi KUSANO,Naoki NAKAMURA ,Eiji TANABE ,Hideyuki SUNAGA ,Yoshie OTAKE
Accuthera Int. Joichi KUSANO Accuthera Int. Naoki NAKAMURA Accuthera Int. Eiji TANABE Riken Hideyuki SUNAGA Riken Yoshie OTAKE
1.Introduction
In Japan many bridges suffering degradation problems due to long years’ service which results to high demand about in situ NDE techniques. Since conventional NDE methods have limitation to inspect inner structure, Computer Tomography becomes a better choice to detect inner steel rod. Portable 950kwV/3.95MeV X-band linac would be adopted as beam source complying with that radiation source under 4 MeV is legally permitted to work outside controlled area.[1] Partial projection problem with limited scanning angle range should be studied because of constraints of bridge shape. With dual-energy CT analysis, the erosion thickness of inner steel rods can be estimated based on different density and attenuation character between iron and iron dioxide. After th, VCAD software, developed by Riken, Japan, is adopted for modeling and further structural analysis. 3D model of inspected module will be built with reconstructed sectional images so that original bridge model can be amended by substituting degradation module. Structural analysis is carried out hereafter with reduced stiffness to evaluate mechanical characters of aging structure. Fig.1 Schematic view of CT inspection system In previous work, a small concrete sample with inner steel rods was scanned and CT reconstruction was carried out with Filtered Back Projection algorithm. However, due to serious noise caused by low energy scattered X-ray, the inner ste rods were not showed clear enough for 3D modeling software VCAD to distinguish the color.[2]
New experiment adopting line sensor with collimators is implemented, as explained in next section, followed by partial CT study. Then in the fourth section, primary simulation work of dual-energy CT for component analysis is presented. 2．CT Experiment with Collimators IIC Todai Hybrid Scintillator is a line sensor compacted with collimators to keep effective incident beam angle as showed in figure 2. The small sample is a concrete cylinder (height: 10mm; diameter: 10mm) that contains steel rods inside it and the figure 3 showed the scanning experiment layout using 3.95MeV linac.[3] Fig.2Line sensor with collimators Fig.3CT scanning experiment Fig.4Transmission images in different directions From transmission images, two steel rods can be confirmed inside the sample, one flat and the other inclined. Comparing the reconstructed images by Filter Back projection algorithm in figure 5, the steel rods have become much clearer than previous results owing to collimators effect. 64 section slices of the whole sample is reconstructed with projection data and 3D model is successfully built by VCAD software. [4] max-sino_52_allmax-sino_10_all(a) (b) (c) (d) Fig.5Reconstructed sectional images (a,b: with collimator; c,d: without collimator; a,c: flat inner steel rod; b,d: inclined inner steel rod) C:\Users\wuwenjing\Desktop\VCAD\3D_linesensor130515\sample.pngFig.63D model of scanned sample (red: steel rods; blue: concrete) 3．Partial CT Reconstruction When reconstructing with incomplete projection data, artifacts would deteriorate image quality. Partial CT reconstruction is studied with experimental data to explore how much influence there is for inner steel rods. 120d120d_10(a) (b) 90d90d_10(c) (d) Fig.7Reconstruction with limited angle range (a,b: 120degree; c,d: 90degree; a,c: flat inner steel rod; b,d: inclined inner steel rod) When reconstructing with data of limited projection angle range, the sample edge shape was damaged seriously and apparent shadow appears around missing angles. The inclined steel rod was damaged a little more serious than flat steel rod with shadow because it located near edge. per8_52per8_10(a) (b) per16_52per16_10(c) (d) Fig.8Reconstruction with few projections (a,b: 8degree increment; c,d: 16degree increment; a,c: flat inner steel rod; b,d: inclined inner steel rod) When increasing projection increment and reconstructing with data of fewer projections but still covering full angle range, apparent line shadow due to incomplete data appears. The steel rod’s diameter and shape was affected especially as increment increased. Inner steel rod is still discernible in reconstructed images under partial projection conditions but the shape is deteriorated to some extent especially under severe lack of projection data. Increasing projection increment properly could be one approach to reduce scanning time but the increment degree should be well controlled. On the other hand, some improvement for projection data recovery algorithms is under discussion such as iterative analytic reconstruction. 4．Dual-energy CT simulation 4.1Method of Material Estimation by Dual-energy CT CT image is representation of linear attenuation coefficientμ, which denotes how strongly the media material absorbs or scatters beam light. This character can be described by the mass attenuation coefficient μ/ρwhereρis the density.[5] Each kind of element media shows distinctive mass attenuation coefficient. For heterogeneous material, the mass attenuation coefficient is defined as below and Wi is fraction by weight of ith atomic constituent: ..miiiiW......At different energy level, the measuredμvalue would vary for same media. Therefore for μ1andμ2at two different energies, we have ....1111222()fZFZZFfZ...............Here ......,iifZwEEZdE..........F(Z) shows as monotonic function withμ1/μ2, so we can get Z=F-1(μ1/μ2) through numerical interpolation. Relatively, density can be obtained by ....112iiiifZfF............For heterogeneous material, atomic number Z and density ρinclude effectiveness of all compound elements: kiikiiiiiZZ.........4.2 Simulation of Dual-Energy CT In the simulation sample there is one steel rod inside concrete with erosion coating with thickness of half its radius. The model is similar with figure 3and EGS5 simulation is calculated about spectrum and detector efficiency. Noise elimination in air region is realized by resetμvalue of air. Fig.9Spectrum energy Fig.10Detector efficiency Fig.11Simulation sample Fig.12Estimation result (left: atomic number; right: density) For steel rod with erosion coating Fe and Fe2O3can be identified. The estimation accuracy is acceptable, for estimation of material atomic number Z within 0.2-0.5and for density ρwithin 0.3 g/cm3. Validation with Experimental data is expected in future work. 5．Conclusion The linac CT system for in situ NDE of bridge is improved for reducing low energy scattered X-ray noise by collimator. Through reconstruction with incomplete projection data, the identification of steel rod is discussed. Lack of data would cause artifacts and shape of inner steel rod would be deteriorated seriously when too much projection data is missing. Data recovery algorithm will be added to the system. Additionally, material estimation with dual-energy CT is programed and tested that Fe and Fe2O3 can be well identified for erosion evaluation. Relative experiment is under plan. References [1] M. Uesaka, T. Natsui, et al.: ““950keV, 3.95MeV and 6MeV X-band linacs for nondestructive evaluation and medicine““, (2011), Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 657:1, 82-87.(2011) [2] Wenjing Wu, Haitao Zhu, et al.: ““Partial CT and Structural Analysis with 950 Kev/3.95 Mev X-Band Linac X-Ray Sources““, (2013), E-Journal of Advanced Maintenance, 5:2, 2013) [3] K. Lee, Dual energy X-ray imaging for material recognition using 950 keV X-band Linac, Doctoral dissertation, The University of Tokyo, 2010. [4] M. Akitake: ““VCAD System: Advanced Software System for Production Engineering and Biological Research““, (2010), JSCES, 15:2, 2309-2315.(2010) [5] B. J. Heismann, Leppert, J., Stierstorfer, K.: ““Density and atomic number measurements with spectral x-ray attenuation method““, (2003), Journal of Applied Physics, 94:3, 2073-2079.(2003)
“ “Partial Inspection and Dual-energy CT Material Analysis with 950keV/3.95MeV X-Band Linac “ “Wenjing WU,Haito ZHU,Ming JIN,Katsuhiro DOBASHI ,Takeshi FUJIWARA,Mitsuru UESAKA,Joichi KUSANO,Naoki NAKAMURA ,Eiji TANABE ,Hideyuki SUNAGA ,Yoshie OTAKE “ “Partial Inspection and Dual-energy CT Material Analysis with 950keV/3.95MeV X-Band Linac “ “Wenjing WU,Haito ZHU,Ming JIN,Katsuhiro DOBASHI ,Takeshi FUJIWARA,Mitsuru UESAKA,Joichi KUSANO,Naoki NAKAMURA ,Eiji TANABE ,Hideyuki SUNAGA ,Yoshie OTAKE