OBJECTIVE: The objective of this study was to evaluate image quality of computed tomography (CT) portal venography using nonlinear blending technique of dual-energy CT (DECT) in portal phase. METHODS: Fifty patients underwent contrast-enhanced abdominal scan with DECT at portal venous phase. The images of 2 energy levels were mixed by a nonlinear blending filter defined by blending center (BC) and blending width (BW) and compared with linear mixing images with a ratio of 0.5. Image quality was evaluated by the images' contrast-to-noise ratios (CNRs) and the grade of visible branches of portal vein (grades 1-6). The BC and BW were optimized for the best CNRs. RESULTS: The portal-to-hepatic contrast was significantly higher when the average of the liver and portal vein CT values was chosen as the BC, and 0 to 20 Hounsfield units were chosen as the BW than when using the default mixing values (80.04 [SD, 28.09] vs 38.55 [SD, 14.49] Hounsfield units, P < 0.001). In the optimized images using the nonlinear mixing technique, the mean CT values for the portal venous system and the CNRs were significantly greater than those of the linear mixing images (P < 0.001). A visible branch grade 5 or greater was assigned to 76% of the patients in the optimized group and 36% of the patients in the default linear mixing group. This difference was significant (P < 0.01). CONCLUSIONS: The nonlinear blending technique of DECT can improve the image quality of portal venography and show more branches of portal vein without changing the conventional scan protocol.
OBJECTIVE: The objective of this study was to evaluate image quality of computed tomography (CT) portal venography using nonlinear blending technique of dual-energy CT (DECT) in portal phase. METHODS: Fifty patients underwent contrast-enhanced abdominal scan with DECT at portal venous phase. The images of 2 energy levels were mixed by a nonlinear blending filter defined by blending center (BC) and blending width (BW) and compared with linear mixing images with a ratio of 0.5. Image quality was evaluated by the images' contrast-to-noise ratios (CNRs) and the grade of visible branches of portal vein (grades 1-6). The BC and BW were optimized for the best CNRs. RESULTS: The portal-to-hepatic contrast was significantly higher when the average of the liver and portal vein CT values was chosen as the BC, and 0 to 20 Hounsfield units were chosen as the BW than when using the default mixing values (80.04 [SD, 28.09] vs 38.55 [SD, 14.49] Hounsfield units, P < 0.001). In the optimized images using the nonlinear mixing technique, the mean CT values for the portal venous system and the CNRs were significantly greater than those of the linear mixing images (P < 0.001). A visible branch grade 5 or greater was assigned to 76% of the patients in the optimized group and 36% of the patients in the default linear mixing group. This difference was significant (P < 0.01). CONCLUSIONS: The nonlinear blending technique of DECT can improve the image quality of portal venography and show more branches of portal vein without changing the conventional scan protocol.