Image quality comparison of lower extremity CTA between CT routine reconstruction algorithms and deep learning reconstruction

Table 4 Objective sharpness analysis

Parameter	DLR	MBIR	HIR	FBP	p*
Structures
Aorta	0.47 ± 0.05	0.39 ± 0.04*	0.50 ± 0.04*°	0.37 ± 0.04*˜	< 0.001
Femoral artery	0.39 ± 0.06	0.35 ± 0.06*	0.42 ± 0.06*°	0.35 ± 0.06*˜	< 0.001
Popliteal artery	0.29 ± 0.10	0.28 ± 0.09*	0.32 ± 0.10*°	0.29 ± 0.09°˜	< 0.001
Liver	0.30 ± 0.05	0.26 ± 0.02*	0.33 ± 0.03*°	0.30 ± 0.03°˜	< 0.001
Psoas muscle	0.33 ± 0.03	0.28 ± 0.03*	0.35 ± 0.03*°	0.32 ± 0.03°˜	< 0.001
Whole section image
Aorta level	0.35 ± 0.03	0.29 ± 0.02*	0.36 ± 0.02*°	0.26 ± 0.02*°˜	< 0.001
Femoral artery level	0.31 ± 0.01	0.26 ± 0.01*	0.33 ± 0.01*°	0.26 ± 0.01* ˜	< 0.001
Popliteal artery level	0.31 ± 0.01	0.28 ± 0.01*	0.35 ± 0.01*°	0.33 ± 0.01*°˜	< 0.001

DLR deep learning–based reconstruction, MBIR model-based iterative reconstruction, HIR hybrid-iterative reconstruction, FBP filtered back projection
Pairwise comparisons showed significant differences from DLR (*), MBIR (°), and HIR (˜) (p < 0.05)

ISSN: 1471-2342