Effect Of Low Dose Combined With Adaptive Statistical Iterative Reconstruction-V And Deep Learning Reconstruction Technique On Abdominal Image Quality And Gastric Cancer Lesion Display

Part One The effects of different tube voltages on abdominal image quality and gastric cancer lesion displayObjective:To compare the effects of 100kVp and 120kVp tube voltages combined with Adaptive Statistical Iterative reconstruction-V（ASiR-V）on abdominal image quality and gastric cancer lesion display.Methods:Sixty newly diagnosed gastric cancer patients who met the inclusion requirements were continuously enrolled,which were divided into low-dose group and conventional dose group,with 30 patients in each group.The tube voltage of the low-dose group was 100k Vp,and that of the conventional dose group was 120k Vp.Other scanning conditions and reconstruction parameters were the same.Objective evaluation and subjective evaluation were performed on the venous phase images with 40%ASiR-V and1.25mm layer thickness.Objective evaluation:CT values,noise（SD）,signal-to-noise ratio（SNR）,contrast-noise ratio（CNR）and fat density resolution of gastric primary tumor,normal stomach wall,abdominal tissues organs and abdominal aorta were compared between the conventional dose group and the low-dose group.Subjective evaluation:The images of two groups were evaluated for noise,the display of gastric wall stratification and fine structure,overall image quality,and the diagnostic confidence of gastric wall serosal infiltration.The CT Dose Index volume(CTDI_vol)and dose-length Product（DLP）automatically generated by computer were recorded,and the Effective Dose（ED）received by patients was calculated.Results:There were no significant differences in gender,age or body mass index between the conventional and low-dose groups（P>0.05）.The effective radiation dose received by the conventional dose group was 10.152.11m Sv,and the effective radiation dose received by the low dose group was7.63 0.85m Sv,and the difference between the two groups was statistically significant（T=6.085,P<0.05）.Under low-dose scanning,the CT values of gastric primary tumor,normal stomach wall and abdominal tissues were higher than those of conventional dose（P<0.05）,while CT value of pericancerous fat in low-dose group was lower than that in conventional dose group（T=-2.54,P<0.05）.The SD,SNR and CNR value measured in all parts of the low-dose group were higher than those in the conventional dose group.The SNR of pericancerous fat and abdominal aorta showed statistically significant difference between the two groups,and the CNR of liver and abdominal aorta showed statistically significant difference between the two groups（P<0.05）.Compared with the conventional dose group,the resolution of fat density calculated in the low-dose group was lower,but the difference between the two groups was not statistically significant（Z=-1.39,P=0.165）.There were significant differences in subjective scores of image noise and overall image quality between the two groups（P<0.05）,there were no statistically significant differences in the subjective score of gastric wall stratification and small anatomical structure display and the confidence of gastric wall serosal surface infiltration（P>0.05）.Conclusion:Low tube voltage combined with Adaptive Statistical Iterative reconstruction-V can reduce the radiation dose received by patients.Although the image quality of the low dose group and the conventional dose group can meet the diagnostic requirements,and has no impact on the diagnostic confidence of whether the gastric wall serosa is infiltrated,the image noise of the low dose group is increased and the sense of granularity is aggravated.Part Two The effect of DLIR and ASiR-V on abdominal image quality and gastric cancer lesion display at low tube voltageObjective: The abdominal venous phase image quality of Adaptive Statistical Iterative reconstruction-V（ASiR-V）and Deep Learning Image reconstruction（DLIR）was compared under 100 k Vp tube voltage to explore the influence of deep learning reconstruction technology on gastric cancer lesions display under low tube voltage.Methods: A total of 30 newly diagnosed gastric cancer patients were enrolled.All patients underwent GE Revolution CT abdominal enhancement scan with a scanning tube voltage of 100 k Vp,using 60% ASiR-V 80%ASiR-V DLIR-low,medium and high-grade（DLIR-L,M,H）.Original data of venous thin layer reconstructed by five reconstruction models.Objective evaluation: CT values and standard deviations of primary gastric cancer in venous stage,normal gastric wall,abdominal tissues and organs,perigastric fat in the greater curvature of the stomach and subcutaneous fat in the anterior abdominal wall were measured,and SNR,CNR and fat density resolution of pericancerous fat in the above areas were calculated.Subjective evaluation:Two radiologists rated the reconstructed images on a scale of 1-4 based on the image noise,the display of gastric wall layering and fine anatomical structures,the overall image quality,and their confidence in the diagnosis of gastric wall serosal infiltration.Finally,the reconstruction mode with better image quality under low dose was selected by comprehensive subjective and objective evaluation,and compared with the image reconstructed under conventional dose of 40% ASiR-V.All data were statistically analyzed by SPSS 22.0.Results:1.There was no significant difference in CT values between primary gastric cancer and various tissues under the five reconstruction modes（P>0.05）.The noise values of all tissues increased in the order of 80% ASiR-V,DLIR-H,DLIR-M,60% ASiR-V,DLIR-L（P<0.05）,SD value of gastric cancer primary lesion was significantly different among all groups（F=3.155,P=0.016）.The SNR value of gastric cancer lesions increased in the order of DLIR-L、60%ASiR-V、DLIR-M、DLIR-H、80%ASiR-V（P<0.05）.The differences between 80% ASiR-V and DLIR-L and between DLIR-L and DLIR-H were statistically significant.With the increase of deep learning reconstruction intensity,the CNR of gastric cancer lesions also increased successively,with the highest DLIR-H,and the CNR values of organs and aorta also gradually increased,60%ASiR-V and DLIR-H,DILR-L and DLIR-H,as well as DLIR-M and DLIR-H showed statistically significant differences（P<0.05）.The difference of the resolution of fat density in the five reconstruction modes was statistically significant（H=16.117,P=0.003）,and DLIR-H was the highest.There was a good agreement between the two observers on the scores of subjective evaluation indicators（Kappa>0.75）.DLIR-M and DLIR-H have higher confidence in the diagnosis of serous surface infiltration than DLIR-L and ASiR-V,which can meet the clinical diagnosis.2.The SD values of gastric cancer primary tumors were reduced by low-dose DLIR-M combined with 40%ASiR-V compared to conventional dose combined with 40%ASiR-V,and the difference was not statistically significant（P> 0.05）.The SNR and CNR values of gastric cancer and the fat density resolution of pericancerous fat also increased,and the difference between the two groups was statistically significant（P< 0.05）.The subjective score of DLIR-M in terms of noise,overall image quality,and diagnostic confidence in the serosal surface of the gastric wall was higher at the low dose than at the conventional dose of 40%ASiR-V,and the difference between the two groups was statistically significant.The subjective score of gastric wall stratification and fine structure with low dose combined with DLIR-M was lower than 40%ASiR-V with conventional dose,but the difference between the two groups was not statistically significant（P=0.557）.Conclusions:Compared with iterative reconstruction,DLIR-M and DLIR-H can improve the quality of abdominal images and improve the diagnostic confidence of serous surface state of primary gastric cancer.Images reconstructed by DLIR-M are better than those obtained by conventional dose.Deep learning reconstruction at low dose can be used for the display of gastric cancer lesions.