Study On Testing Techniques Of Spectral CT Imaging Key Parameters

Computed tomography (CT) has become one of the most important methods of medical imaging diagnosis. Since several years’continuous development, CT has achieved great progress in the scanning speed, precision and scope of confirmed diagnosis. However, there are many problems existing in the CT imaging, such as single imaging parameters, lacking of function imaging and beam hardening artifact, which confused the radiologists for a long time. The clinical application of spectral CT is the key means to solve these problems. The theory of spectral CT is that different attenuation absorption appears when the material exposed in different level energy X-ray. By this way, it provides more clinical image information. Spectral CT scans material by two or more different X-ray energy. The attenuation absorption of organization presents the relevant variation in different energy level. This variation is quantifiable and identification, based on this, we can get more information about the tested material.Spectral CT testing technique is still in the stage of continually study exploration, compared with the spectral CT imaging technology development and popularization, is relatively lagging. This paper introduced the imaging techniques and clinical application of spectral CT and finished the following work that focused at the key imaging parameters of spectral CT:(1) Focusing on the characteristic of the spectral CT key parameters, this paper carried on the preliminary exploration study on the technique of quantitative analysis ability of spectral CT and designed the following experiments. Designed and made an iodine solution testing phantom. This testing phantom uses bottled pure water and iopamidol 300 contrast medium that iodine concentration is 300 mg/mL as raw materials. Keeping the total solution in 600 mL, numbered the water and injected 0,1, 2,3,…8 mL contrast medium to the water with medical syringe. To fixed these bottles with tape and finished the phantom. Put the phantom on the GE Discovery CT750 HD scanner and scanned the phantom with axial and GSI (gemstone spectral imaging) scanning under the head scanning conditions. Obtained scanning data through spectral CT and did the image processing in GE AW4.6 image post-processing workstation. In the 70keV monochromatic energy image, selected the ROI (region of interest) in the center of the bottles and measured the CT value and the standard deviation. Obtained the effective atomic number histogram, water and iodine basis material image scatter plot and spectral HU curve by GSI analysis module software. The experimental results showed that:a. GSI scanning dose index CTDIvol is 38.42mGy which is lower than 55.90mGy of the axial scanning dose index;b. In the 70keV monochromatic energy image, for example, the CT value of numbered 0、4 and 8 is -0.36,49.91, and 109.91, respectively, the value of bottle 8 is double of the bottle 4;c. The histogram peaks of the effective atomic number are 7.7、9.0 and 10.0 and the effective atomic number closed to the true value 7.4;d. The scatter plot shows that water percentage in the solution is 995～1000 mg/cm3, there is no obvious difference. The iodine in the solution is about 1、21 and 44, closed to the nominal value;e. The spectral CT shows that, the 0、2、4、6 and 8 distribute in equal interval, and the lower energy is, the greater difference is.The experimental results indicate that:Through analyzing the three parameters about the effective atomic number、spectral HU curve and water/iodine basis concentration, GSI can distinguish the different iodine concentrations, even quantitative calculate the iodine concentrations of the solution.(2) This paper designed and discussed the method for evaluating CT image noise based on noise power spectrum (NPS) and designed the following experiments. Put the CTP486 module in Catphan 500 in the Somatom Sensation 64 and changed the slice、dose and reconstruction algorithm to obtain the images with different parameters. The detailed parameters are that:slice:3、5、10mm; tube current:80、 133mA; reconstruction algorithm:H30s (moderate smooth)、H60s (moderate sharp)、 U90s (extremely sharp). According the following data processing requirement, the phantom should be scanned 2 times to get the two set of data with the same condition. We finished the programming processing with MATLAB programming software, and measured the SD value at the same time and compared the SD with NPS. The experimental result showed that:a. The NPS curve amplitude is inversely proportional to the slice. The NPS curve amplitude is the highest when the slice is 3mm. The curve amplitude is lowest when the slice is 10mm. The NPS curve peak frequency almost does not change with slice;b. The NPS curve amplitude is proportional to the enhancement degree of reconstruction algorithm. The NPS curve amplitude is the lowest when the reconstruction algorithm is H30s which is moderate smooth. The curve amplitude is highest when the reconstruction algorithm is U90s which is extremely sharp. The NPS curve peak frequency moves to the high frequency when the reconstruction algorithm is enhanced. The NPS curve peak frequency of H30s is lowest and U90s is highest;c. The NPS curve amplitude is inversely proportional to the dose. The NPS curve amplitude of 80mA is higher than 133mA. The NPS curve peak frequency almost does not change with dose.The experimental results indicated that:The amplitude of the NPS curve is in accord with the SD value. They can both reflect the noise variation. The effect caused by slice、dose and reconstruction algorithm can be obviously reflected in the NPS curve. The NPS curve amplitude reflected noise stationarity and the peak frequency reflected the changes in the noise structure. (3) This paper combined the NPS with SD and applied on noise evaluation of the basis material image obtained with three different methods to know more about the noise variation and the characteristic of the three different methods. Three methods are that:The first method is material decomposition based on joint bilateral filter (JBF) low dose image restoration. The second method is material decomposition based on previous normal dose scan-induced non-local means (ndiNLM) low dose image restoration. The last method is image domain iterative material decomposition. The evaluation result shows that:a. The NPS curve amplitude of the basis material images obtained via three methods is higher than the images before decomposition;b. The NPS curve amplitude of the basis material images that are based on two low dose image restoration methods is higher than the images that decomposed via direct matrix inversion and the peak frequency moves to low frequency compared with that one;c. The NPS curve amplitude of the basis material images obtained via material decomposition based on ndiNLM is lower than the material decomposition method based on JBF and the peak frequency moves to low frequency compared with that one;d. The NPS curve amplitude of the basis material images obtained via image domain iterative material decomposition is lowest and the peak frequency is between that via low dose image restoration methods.The experimental results indicated that:The denoising effect of basis material images via three different methods and the image frequency are different. The noise of the basis material images obtained via image domain iterative material decomposition is lowest, and has rich edge information. The experiments also show that the material decomposition itself will cause the change of the image noise.The innovation of this paper lies in:1) According to the imaging parameters and clinical application characteristics of the gemstone spectral imaging (GSI), this paper designed an iodine solution testing phantom and test the GSI quantitative analysis ability, preliminary explore the method of spectral CT detection technique;2) Proposed the method for evaluating CT image noise based on noise power spectrum and obtained the image via change the scanning parameters and reconstruction algorithm. Evaluate the CT images by the NPS and SD and compare the results to verify the validity of the method of NPS.3) Combined the NPS with SD and applied on noise evaluation of the basis material image obtained with three different methods to know more about the noise variation and the characteristic of the three different methods.This paper introduced the imaging techniques and clinical application of spectral CT and analyzed the imaging key parameters via clinical instance. This paper designed an iodine solution testing phantom to verify the quantitative calculation ability of GSI and had a preliminary study on key testing parameters of spectral imaging. Besides that, this paper analyzed the basis material image based on principle that not changed the convention CT scanning method and low dose imaging. Through three different material decomposition methods so as to realize the low dose scan in image domain for the purpose of high quality base material image. Proposed the NPS and combined it with SD and applied on noise evaluation of the basis material image obtained with three different methods to know more about the noise variation and the characteristic of the three different methods. This research is of great significance for promoting the development of spectral CT testing technique、establishing and improving the noise evaluation system、optimizing the basis material quality and improving clinical diagnosis accuracy.