Table 4. Qualitative Measurements of Image Quality.

	Reader 1		Reader 2
	Score	P	Score	P
Overall image quality	3.26 ± 0.48	< 0.001	3.46 ± 0.52	< 0.001
Image noise	3.54 ± 0.58	< 0.001	3.63 ± 0.57	< 0.001
Image sharpness	2.99 ± 0.29	> 0.999	3.04 ± 0.19	0.134
Image texture	2.82 ± 0.39	< 0.001	2.76 ± 0.43	< 0.001
Lesion conspicuity	3.09 ± 0.36	0.219	3.29 ± 0.51	0.001

Data are mean ± standard deviation. A score of 3 was assigned if the LDCT with DLIR-M images were equivalent to SDCT with h-IR images. Higher scores reflected better overall image quality, less image noise, better image sharpness, more preferred image texture, and better lesion conspicuity. A one-sample Wilcoxon signed-rank test was used to reject the null hypothesis: the median of qualitative scores for LDCT with DLIR-M images equals a hypothetical value of 3. DLIR-M = deep learning image reconstruction medium-strength, h-IR = hybrid iterative reconstruction, LDCT = lower-dose CT, SDCT = standard-dose CT