The Study Of CT Spectral Imaging In The Preoperative Diagnosis And Staging Of Pancreatic Ductal Adenocarcinoma

Part 1 The optimal monochromatic imaging of the pancreatic ductal adenocarcinoma by single-source dual-energy spectral CTObjective To explore the optimal monochromatic images for evaluating the pancreatic ductal adenocarcinoma with contrast-enhanced spectral CT.Methods 113 patients were selected,who underwent spectral CT from January 2013 to December 2014,and the diagnosis were pathologically confirmed.All patients underwent triple-phase（plain,late arterial and portal venous phase）CT scans,and the contrast-enhanced scans were performed using a single-source dual-energy spectral CT mode.All monochromatic images were sent to the advantage workstation with GSI viewer（AW 4.4）for analysis.Three circular regions-of-interest（ROIs）were placed in the maximum level of the lesion including the lesion,pancreatic parenchyma and peripancreatic arteries or veins repectively.The dual phases analysis were under at the same location.To obtain more accurate data the work were performed three times,then the mean of contrast-to-noise ratio（CNR）and optimal energy level were calculated including lesion-to-pancreas,artery-to-lesion,vein-to-lesion.The optimal CNRs were compared to the 70 ke V CNRs.Two radiologists assessed the optimal monochromatic images verus 70 ke V image with a 5-points scale for the lesion conspicuity and image quality.Results The optimal energy level present a skewed distribution.Arterial phase: The median of optimal ke V for lesion-to-pancreas is 40±13ke V with the mean of the optimal CNR is 2.31±1.02,and the 70 ke V CNR is 2.08 ± 0.98.The median of optimal ke V for artery-to-lesion is 40±3 ke V with the mean of the optimal CNR is 14.20 ± 3.69,and the 70 ke V CNR is 12.18 ± 3.16.Portal venous phase: The median of optimal ke V for lesion-to-pancreas is 40±12 ke V with the mean of the optimal CNR is 2.38 ± 1.02,and the 70 ke V CNR is 2.12 ± 0.96.The median of optimal ke V for vein-to-lesion is 40±8 ke V with the mean of the optimal CNR is 5.80 ± 1.88,and the 70 ke V CNR is 5.01 ± 1.66.The optimal CNRs and 70 ke V CNRs of the four groups all showed a significant statistical difference（P <0.001）by paired t tests.Two radiologists all regard that the optimal monochromatic images can improve the lesion conspicuity,and have better confidence in vessels involvement.Conclusions The optimal monochromatic image for pancreatic ductal adenocarcinoma were 40 ke V for observing lesions and assessing peripancreatic vascular involvement.The optimal CNR was higher than 70 ke V CNR（equivalent to conventional energy level 120 k Vp）.The optimal monochromatic images can improve the lesion conspicuity.Part 2 Study of charactreristics of pancreatic ductal adenocarcinoma with single-source dual-energy spectral CT imagingObjective To explore the multi-parametric quantitative measurement and characteristics of pancreatic ductal adenocarcinoma（PDAC）by single-source dual-energy spectral CT.Methods From January 2013 to December 2015,190 patients were scanned with Gemstone Spectral CT.All diagnosis were pathologically confirmed by surgery or biopsy.All patients underwent triple-phase（plain,late arterial and portal venous phase）CT scan.The contrast-enhanced CT scans were performed using the rapid k V switch mode.Circular region-of-interest（ROI）was placed on lesion and pancreatic parenchyma at the maximum lesion size level.The ROI files including the monochromatic CT attenuation（CTmono,range from 40 ke V to 140 ke V）,the effective Z(Z_eff),the iodine concentration（IC）,the water concentration（WC）and the corresponding normalized values of the lesion（normalized to pancreatic parenchyma）,and the CTmono attenuation（range from 40 ke V to 140 ke V）,the Z_eff,the IC,the WC of pancreatic parenchyma in late arterial phase（AP）and portal venous phase（PP）respectively.The measurement work was performed three times repeatedly.Paired t-test and Wilcoxon signed-rank test were used for analyzing the differences between different phases and between the PDAC and pancreatic parenchyma.P < 0.05 was considered statistically significant.Results The Z_eff,IC and WC of PDAC in two phases were 8.18 ± 0.31 and 8.48 ± 0.35,9.85 ± 5.09（100ug/cm3）and 15.17 ± 6.24（100ug/cm3）,1026 ± 12（mg/cm3）and 1025 ± 13（mg/cm3）respectively.The CTmono（rang from 40 ke V to 140 ke V）,normalized CTmono（rang from 40 ke V to 89 ke V）,the Z_eff,NZ_eff,IC,and NIC of PDAC have significant differences between AP and PP（P<0.05）,and there were no differences of WC and NWC.The spectral HU curve of PDAC in AP was lower than PP.And the normalized spectral HU curve increased with the increase of energy level.The Z_eff,IC and WC of pancreatic parenchyma in two phases were 8.97 ± 0.39 and 9.25 ± 0.35,24.58 ± 7.70（100ug/cm3）and 30.60 ± 7.77（100ug/cm3）,1032 ± 14（mg/cm3）and 1032±14（mg/cm3）respectively.The Z_eff,IC and WC of PDAC were lower than pancreatic parenchyma in both two phases（P<0.001）.Conclusion CT spectral imaging showed that PDAC was relatively hypo-vascular and progressive enhanced tumor.The differences between PDAC and pancreatic parenchyma were more significant in late arterial phase in CT spectral imaging and the differences of CT spectral HU curves were more distinct at lower energy.Part 3 The value of dual energy CT spectral imaging in TNM staging of Pancreatic ductal adenocarcinoma:initial experienceObjective To investigate the value of dual energy CT spectral imaging in preoperative TNM staging in patients with primary Pancreatic ductal adenocarcinoma（PDAC）.Methods 113 patients with PDAC underwent triple-phase（plain,late arterial and portal venous phase）CT scans,and the contrast-enhanced scans were performed using a single-source dual-energy spectral CT mode.The optimal monochromatic images were interpreted by two observers in consensus for T staging of PDAC and the vascular involvement,and for M staging through the spectral HU curve and Histogram analysis.Then the results were compared to the pathological diagnosis using the Chi-square test.The short-axis diameter greater than 5 mm was the criterion for nodal enlargement.Compared the short axis diameters,long axis diameters,the ratio of short-axis to long-axis diameter,spectral HU curve,effective Z(Z_eff),the iodine concentration（IC）between non-metastatic lymph nodes and metastatic lymph nodes.The Wilcoxon signed-rank test was used to compare the short axis diameters,long axis diameters,the ratio of short-axis to long-axis diameter,Z_eff and IC.The area under the ROC curve（AUC）of all above parameters were tested using the Medcalc software.Stepwise multiple logistic regression analysis was performed to determine which parameters were statistically significant in differentiating non-metastatic lymph nodes from metastatic lymph nodes.Results The overall accuracy for T staging were（100/113）88.5%.The optimal monochromatic imaging had 78.0% sensitivity,98.8% specificity,97.6% accuracy for arteries unresectability,and 85.3% sensitivity,98.0% specificity,96.8% accuracy for veins unresectability.The sensitivity and specificity for diagnosis of hepatic metastasis with spectral imaging were higher than conventional CT.The ratio of short-axis to long-axis diameter had significantly difference（P<0.01）between metastatic lymph nodes and non-metastatic lymph nodes,and the short axis diameters and long axis diameters had no differences（P> 0.05）.The Z_eff and IC of metastatic lymph nodes were significantly lower than non-metastatic lymph nodes in the AP and PP（P<0.01）.There were no differences between AUC of Z_eff,IC and the the ratio of short-axis to long-axis diameter in both AP and PP.The AUC of logistic regression equation had no significant difference with those above parameters.Conclusion Dual energy CT spectral imaging may be helpful to improve the accuracy of T and M staging of PDAC.The optimal monochromatic images had higher sensitivity and specificity than conventional CT for vascular involvement.The quantitative analysis of Z_eff and IC measurements could be useful for differentiating metastatic from non-metastatic lymph nodes,but not better than the the ratio of short-axis to long-axis diameter.