75% radiation dose reduction using deep learning reconstruction on low-dose chest CT

BMC Medical Imaging

Table 2 Quantitative Image Quality Assessment Results

	QLD-DLIR, (3-mm thickness, standard kernel)	LD-IR (3-mm thickness, standard kernel)	P-value*	QLD-DLIR (1-mm thickness, sharp kernel)	LD-IR (1-mm thickness, sharp kernel)	P-value*
Noise
Lung	11.5 ± 3.3	16.0 ± 3.8	< 0.001	37.1 ± 6.7	66.5 ± 10.9	< 0.001
Trachea	16.0 ± 3.9	13.4 ± 2.0	< 0.001	44.2 ± 7.7	57.5 ± 8.4	< 0.001
Aorta	8.6 ± 2.3	17.6 ± 2.2	< 0.001	81.8 ± 6.1	104.9 ± 7.2	< 0.001
Muscle	12.2 ± 3.3	22.8 ± 3.3	< 0.001	81.0 ± 9.5	111.8 ± 9.9	< 0.001
Axillary fat	9.2 ± 2.0	17.4 ± 3.3	< 0.001	82.1 ± 11.6	108.3 ± 16.6	< 0.001
SNR
Lung	84.4 ± 24.3	58.9 ± 14.1	< 0.001	23.0 ± 4.6	13.7 ± 2.5	< 0.001
Trachea	64.3 ± 14.6	75.8 ± 11.7	< 0.001	20.1 ± 3.6	16.6 ± 2.7	< 0.001
Aorta	5.6 ± 1.3	2.6 ± 0.4	< 0.001	0.58 ± 0.06	0.43 ± 0.06	< 0.001
Muscle	5.0 ± 1.4	2.5 ± 0.5	< 0.001	0.71 ± 0.12	0.49 ± 0.08	< 0.001
Axillary fat	11.3 ± 2.8	6.0 ± 1.2	< 0.001	1.2 ± 0.2	0.95 ± 0.16	< 0.001
Edge-rise-distance (mm)	1.52 ± 0.08	1.80 ± 0.11	< 0.001	1.15 ± 0.07	1.06 ± 0.08	< 0.001

Note.—Data are presented as means ± standard deviations. Italicized data indicate that the values are lower (for noise and edge-rise-distance) or higher (for SNR) than the compared counterpart. DLIR = deep-learning image reconstruction, IR = iterative reconstruction, LD = low dose, QLD = quarter of the low dose, SNR = signal-to-noise ratio
*P-values were calculated using the paired t-test

ISSN: 1471-2342