Application Of CT Reconstruction Noise Reduction Technology Assisted Perfusion CT Instead Of Conventional Enhanced CT In Pancreatic Solid Tumor

Part 1Detection of Insulinoma:One-stop Pancreatic Perfusion CT with Calculated Mean Temporal Images can Replace the Combination of Bi-phasic plus Perfusion ScanAbstractObjective:To evaluate the feasibility of one-stop pancreatic perfusion CT with mean temporal(MT)imaging replacing the combination of a bi-phasic scan plus a perfusion scan to detect insulinoma.Material and methods:Forty-five patients with suspected insulinoma,who underwent both biphasic and perfusion CT,were enrolled in this retrospective study.MT arterial,portal venous and tumour phase images were generated by averaging 3 dynamic datasets from perfusion CT.The image quality and diagnostic performance for biphasic and MT datasets were assessed and compared.Radiation dose were also recorded.Paired t-tests,Wilcoxon signed-rank tests and McNemar’s tests were applied for comparison.Results:Compared to bi-phasic CT images,image noise,SNR and CNR of the MT datasets were all non-inferior(noise and CNR of the portal vein,p=0.565 and p=0.227,respectively)or superior(p ≤0.001).The subjective image quality(including pancreatic parenchyma,pancreatic duct,small arteries and portal vein depiction as well as overall image quality)was better in the MT images(p<0.001 to p=0.039).The sensitivity and NPV of MT images were also better(95%vs 75%and 75%vs 37.5%for reader 1;97.5%vs 72.5%and 85.7%vs 35.3%for reader 2).Omitting the bi-phasic scan resulted in a dose reduction of 25%±4%.Conclusion:MT imaging can allow pancreatic perfusion CT to be used alone without the need for an additional biphasic CT in the detection of insulinoma.Part 2Feasibility of Ultra-low Dose Pancreatic Perfusion CT using AiCE Algorithm in Replacing the Combination of a Bi-phasic Scan plus a Perfusion Scan to Detect InsulinomaAbstractObjective:To evaluate the feasibility of ultra-low dose(CTDI:1.1mGy)pancreatic perfusion CT using AiCE algorithm in replacing the combination of a bi-phasic scan plus a perfusion scan to detect insulinoma.Material and methods:Forty-seven patients with suspected insulinoma,who underwent both biphasic and perfusion CT,were enrolled in this prospective study.The raw data of perfusion CT was reconstructed with conventional iterative reconstruction algorithm(AIDR)and two deep learning methods(AiCE-Body and AiCE-Bodysharp).Conventional CT was reconstructed with AIDR.Two readers assessed the image quality and diagnostic performance separately for the 4 group images(perfusion AIDR,AiCE-Body,AiCEBodysharp and conventional AIDR).ANOVA,Friedman,Delong and Cochran’s Q test were used to do analysis.Results:Qualitative scores of overall image quality and pancreatic parenchyma depiction were best in AIDR,followed by AiCE-Bodysharp,AiCE-Body and were worst in perfusion AIDR.Quantitative image noise of AiCE-Body sharp,AiCE-Body and conventional AIDR were basically comparable and all were better than perfusion AIDR.SNR of aorta,portal vein and pancreas were highest in AiCE-Bodysharp,followed by AiCE-Body,conventional AIDR and were lowest in perfusion AIDR.CNR of aorta,portal vein and pancreas were comparable in AiCE-Bodysharp and AiCE-Body,followed by conventional AIDR and were lowest in perfusion AIDR.The sensitivity for insulinoma detection was lowest in conventional AIDR(reader 1:74.2%,reader 2:77.4%),followed by perfusion AIDR(reader 1:90.3%,reader 2:87.1%),and was highest in AiCE-Body and AiCEBodysharp(reader 1:96.8%for both images,reader 2:93.6%for both images).The specificity was lowest in AiCE-Body and AiCE-Bodysharp(reader 1:85.7%for both images,reader 2:85.3%for both images),followed by perfusion AIDR(100%for reader 1 and 96.4%for reader 2),and was highest in conventional AIDR(100%for both readers).Conclusion:AiCE algorithm can reduce image noise of ultra-low dose perfusion CT and improve the image quality to a certain extent.However,the image quality ultra-low dose pancreatic perfusion CT using AiCE algorithm cannot reach the level of conventional CT.In addition,ultra-low dose pancreatic perfusion CT using AiCE algorithm has certain false positive results for the diagnosis of insulinoma.Thus,ultra-low dose pancreatic perfusion CT using AiCE algorithm cannot replace the combination of a bi-phasic scan plus a perfusion scan to detect insulinoma.Part 3Feasibility of Ultra-low Dose Pancreatic Perfusion CT using Calculated MT Images in Replacing the Combination of a Biphasic Scan plus a Perfusion Scan to Detect InsulinomaAbstractObjective:To evaluate the feasibility of ultra-low dose(CTDI:1.1mGy)pancreatic perfusion CT using calculated MT images in replacing the combination of a bi-phasic scan plus a perfusion scan to detect insulinoma.Material and methods:Forty-seven patients with suspected insulinoma,who underwent both biphasic and perfusion CT,were retrospectively included in this study.Three adjacent phases were calculated into one image set to obtain MT arterial,portal and tumour phase images,respectively.Two readers assessed the image quality and diagnostic performance separately for conventional CT and MT images.Paired t-tests,Wilcoxon signed-rank tests and McNemar’s tests were applied for comparison.Results:All quantitative parameters of MT images were superior to conventional enhanced CT(p≤0.01).Insulinoma detection of MT images was also better than conventional CT,with AUC values of 0.936 vs 0.871 for reader 1 and 0.936 vs 0.887 for reader 2,respectively.There were no statistically significant differences of image noise,pancreatic parenchyma depiction and arterial overall image quality between MT images and conventional CT(p-values range 0.052-0.670).The arterial image contrast of MT images was better than that of conventional CT(p<0.001).However,qualitative scores of image artefacts,pancreatic duct depiction,contrast and overall image quality in the portal phase were lower in MT images than conventional CT and all have statistically significant differences(p-value range<0.001-0.026),except for pancreatic duct depiction and image contrast.Conclusion:Ultra-low dose pancreatic perfusion CT using calculated MT images cannot replace the combination of a bi-phasic scan plus a perfusion scan to detect insulinoma.Part 4Comparison of AiCE Algorithm and Calculated MT Images in Improving the Image Quality and Diagnostic Performance of Ultra-low Dose Pancreatic Perfusion CTAbstractObjective:To compare the ability of AiCE algorithm and calculated MT imaging in improving the image quality and diagnostic performance of ultra-low dose pancreatic perfusion CT.Material and methods:Forty-seven patients with suspected insulinoma were retrospectively included in this study and AAPM phantom images were prospectively obtained.Patients and phantom underwent the same scanning protocol and reconstruction methods.The raw data of perfusion CT was reconstructed with conventional iterative reconstruction algorithm(AIDR)and AiCE algorithm(AiCE-Bodysharp).AIDR images were used to calculate MT images.Two group of images(AiCE-Bodysharp and MT images)were compared.Patient images were used for the comparison of qualitative and quantitative image quality as well as diagnostic performance.Phantom images were used to obtain task-based transfer functions(TTF),noise power spectrum(NPS)and detectability index(d’).Results:There were no statistically significant differences in qualitative image noise,artefacts,contrast and overall image quality between AiCE-Bodysharp and MT images(all p-values>0.05).The depiction of pancreatic contours,pancreatic duct,common bile duct within the pancreatic head,diaphragmatic crura and duodenum in both arterial phase and portal venous phase,the depiction of pancreatic parenchyma in arterial phase and superior mesenteric vein depiction in portal venous phase was better in MT images than AiCEBodysharp images(p-values ranged from<0.001 to 0.023).Except for SNR and CNR of aorta,all quantitative parameters in MT images were slightly higher than AiCE-Bodysharp images.Insulinoma detection of MT images was also better than conventional CT,with AUC values of 0.936 vs 0.845 for reader 1 and 0.936 vs 0.829 for reader 2,respectively.MT images detected more lesions than AiCE-Bodysharp images(reader 1:96 vs 85;reader 2:116 vs 103).Except for the TTF50%for high contrast objective(the two images were equal),TTF50%of MT images was higher than AiCE-Bodysharp images,and the difference was largest for low contrast objective(38 vs 26).All d’ values of MT images were higher than AiCE-Bodysharp images.Conclusion:MT imaging outperforms AiCE algorithm in improving the image quality of ultra-low dose pancreatic perfusion CT.Part 5Feasibility of Low Dose Pancreatic Perfusion CT using Calculated MT Images in Replacing Conventional Bi-phasic CT in the Diagnosis of Pancreatic carcinoma:a Preliminary StudyAbstractObjective:To initially explore the feasibility of low dose pancreatic perfusion CT using calculated MT images in replacing conventional bi-phasic CT in the diagnosis of pancreatic carcinoma by evaluating the feasibility of MT images 1)to clearly display pancreatic carcinoma;2)to classify of the resectability of pancreatic carcinoma;3)to detect non-metastatic lesions in abdomen.Material and methods:Forty-seven patients with pancreatic carcinoma who underwent both low-dose pancreatic perfusion CT and conventional CT at our institution were prospectively included.Tumour contrast,tumour CNR and qualitative tumour conspicuity were compared between MT images and conventional CT.The tumour resectability classification,diagnostic confidence and the number of non-metastatic lesions detected in abdomen were also compared between the two groups of images.The radiation dose and iodine dose of perfusion CT and conventional CT were recorded.Paired t-tests and Wilcoxon signed-rank tests were used to do analysis.Results:Qualitative score of tumour conspicuity were higher in MT images than conventional CT(Reader 1:5(4,5)vs 4(4,5),p<0.001;Reader 2:5(4,5)vs 4(3,5),p<0.001).Tumour contrast(95.9±35.4HU vs 47.3±27.6HU,p<0.001)and CNR(7.63±3.03 vs 4.51± 1.89,p<0.001)were also higher in MT images than conventional CT.There was good to excellent agreement between the two observers in tumour resectability classification using MT images and conventional CT,with κ values of 0.703 and 0.728 respectively.Both observers had high diagnostic confidence in tumour resectability classification,with qualitative scores of 5(4,5)for both readers and both image sets(p-value of 0.317 for both observers).MT images detected more non-metastatic lesions in abdomen than conventional enhanced CT(reader 1:217 vs 203;reader 2:223 vs 206),and they were comparable to detect low contrast lesions(reader 1:160 vs 163;reader 2:167 vs 167)and small lesions under 1 cm(reader 1:164 vs 168;reader 2:172 vs 172).The effective radiation dose for perfusion CT and conventional bi-phase CT was 11.86(9.56,15.57)mSv and 10.17±3.31 mSv,respectively.The iodine dose for perfusion CT was 23.3±3.0g.The iodine dose for conventional CT was 25.6 g or 29.6 g.Conclusion:Low-dose pancreatic perfusion CT using calculated MT images is superior to conventional enhanced CT in pancreatic carcinoma depiction.Two images were comparable in tumour resectability classification and non-metastatic lesions detection.The above shows that suggesting that low-dose pancreatic perfusion CT using calculated MT images may have the potential to replace conventional enhanced CT for the diagnosis of pancreatic carcinoma.