The Value of the Model Based Iterative Reconstruction to Improve the Quality of Abdominal CT Images

Objective To compare the abdominal CT images quality obtained by model based iterative reconstruction( MBIR),adaptive statistical iterative reconste ruction( ASIR) and the filtered back projection( FBP) reconstructions. Methods 36 adult patients underwent Chest CT with automatic exposure control( Auto mA; GE Healthcare,Waukesha, Wis) on Discovery CT750HD. 3 sets of 0. 625mm slice thickness CT images were reconstructed with FBP,40% ASIR and MBIR. Image noises of the background,liver,erector spine muscle,abdominal aorta,and CNR were measured and compared. The abdominal imaging quality was evaluated by two radiologists independently using 5 scales( 5 excellent,4 good, 3 can distinguish,2 unclear,and 1 cannot be used). Results At same image slice thickness,compared with FBP,the average image noise reduction with 40% ASIR and MBIR for the background,liver,erector spine muscle,abdominal aorta were( 21. 82%,16. 62%,23. 10% and24. 74%) and( 57. 00%,70. 17%,67. 77% and74. 49%),respectively. Compared wirh FBP,the CNR for erector spine muscle with 40% ASIR and MBIR were improved by 26. 47% and 235. 29% respectively. The average image noise reduction and CNR improvement had statistically significant difference between groups( P 0. 001). At the same slice thickness of 0. 625mm,MBIR reconstructions revealed statistically lower image noises( P 0. 001). CNR with MBIR was also much higher than that with FBP( P 0. 05). The imaging quality of MBIR was better than that of FBP( 4. 32 ± 0. 55 VS 2. 51 ± 0. 68,P 0. 001). Conclusion MBIR and ASIR can improve abdominal image quality. Especially MBIR reconstruction can reduce image noise and improved abdominal image CNR in low dose CT. It may be helpful for reduction radiation dose in routine abdominal CT with MBIR reconstruction..