Study of Reconstruction Parameters of Non-Linear Image Blending in Dual Energy CT Imaging for Pancreatic Mass

王慧慧,刘婧,刘建新,曹新鸣,王子岩,王霄英
2017-01-01
Abstract:Objective To evaluate the image quality of dual-energy abdominal computed tomography (DECT) with a non-linear image blending reconstruction algorithm as compared to the conventional linear blending reconstruction images.Methods Thirty-two patients suspected of pancrentic lesions (15 solid lesion and 17 cystic lesion) underwent dualsource,dual-energy abdominal CT (80 kVp/Sn140 kVp) examination.Conventional linear blending algorithm (M =0.3) was used to reconstruct CT data of pancreatic parenchymal phase,which was taken as the reference image.Non-linear blending images of the pancreatic parenchymal phase were also obtained by using different reconstruction protocols.The blending center was set as the median value between the lesion and the normal pancreatic parenchyma with different blending width (BW =0,BW =50,BW =100,BW =200 and BW =500 HU,respectively).Contrast of lesion,lesion CNR and image noise were calculated and compared with the reference image and various nonlinear blending images.Two radiologists assessed subjective image quality,lesion contrast and conspicuity and selected the preferred images.Results For solid lesions,the non-linear blending image sets with level width of BW =0,BW =50 and BW =100 HU were estimated as the most preferred image sets.For cystic lesions,the non-linear blending image set with level width of BW =0,BW =50,BW =100 and BW =200 HU were estimated as the most preferred ones.There was no significant difference in noise of all images (P =0.147).CNR in non-linear blending image was significantly higher (P < 0.001 for both solid and cystic lesions) than that of reference image.Among non-linear blending images,CNR of solid lesion was the highest in BW =50 HU without statistical difference compared with BW =0 and BW =100 HU (P =0.819,0.117),and CNR of cystic lesion was the highest in BW =0 HU without statistical difference compared with BW =50 HU (P =0.053).Conclusion The non-linear blending images provide improved CNR over the conventional linear blending method (M =0.3).The best blending width is 50 HU for solid lesion and 0 HU for cystic lesion when blending center is set as the median value between pancreatic parenchyma and lesion.
What problem does this paper attempt to address?