Iterative reconstruction for quantitative computed tomography analysis of emphysema: consistent results using different tube currents

Figures & data

Table 1 Comparisons of quantitative pulmonary measurements using different tube current and reconstruction settings (n=22)

CT index	Reconstruction method	Mean ± SD			P-value*
		240 mA	120 mA	60 mA
LV (L)	AIDR3D	4.28±1.05	4.32±1.08	4.38±1.13	NS
	FBP	4.27±1.05	4.30±1.08	4.36±1.13	NS
MLD (HU)	AIDR3D	−821.4±52.9	−821.4±52.0	−820.9±50.6	NS
	FBP	−822.8±52.3	−824.7±51.7	−826.2±50.1	NS
LAA% (%)	AIDR3D	9.9±15.1	10.0±14.9	10.1±14.8	NS
	FBP	10.9±15.0	12.1±14.6	14.2±14.2	<0.0001
15th percentile (HU)	AIDR3D	−919.9±45.9	−922.1±45.2	−922.3±43.6	NS
	FBP	−924.4±44.6	−929.1±43.2	−936.2±39.1	<0.01
Image noise (HU)	AIDR3D	10.7±2.0	13.1±2.3	15.6±2.3	<0.0001
	FBP	18.6±5.0	26.1±7.2	38.1±11.5	<0.0001

Note:

* P-values were obtained from all three comparisons among different tube current settings.

Abbreviations: LV, lung volume; MLD, mean lung density; LAA%, percent low attenuation area; HU, Hounsfield unit; AIDR3D, Adaptive Iterative Dose Reduction using Three Dimensional Processing; FBP, filtered-back projection; SD, standard deviation; NS, not significant; CT, computed tomography.

Figure 1 Example of emphysema measurements made by the workstation used in this study.

Notes: Low attenuation areas (LAA, < −950 HU) are shown as red-colored areas. For this case (81 year old male, 62 kg), when using AIDR3D, measured LAA% results were 6.8% at 240 mA, 7.7% at 120 mA, and 8.0% at 60 mA. When not using AIDR3D (ie, when using FBP), LAA% results were 9.2% at 240 mA, 12.0% at 120 mA, and 15.3% at 60 mA. Note that LAA% measurements were obtained automatically for the entire lung and not from these single axial images.

Abbreviations: LAA%, percent low attenuation area; AIDR3D, Adaptive Iterative Dose Reduction using Three Dimensional Processing; FBP, filtered-back projection; HU, Hounsfield unit.

Figure 2 Image noise analysis.

Notes: At each tube current, greater image noise was observed on scans with filtered-back projection (FBP; without AIDR3D) than on those with AIDR3D. An FBP scan at 60 mA showed the greatest image noise. Note that each AIDR3D and FBP pair at three tube current settings was reconstructed from a single set of raw data.
Abbreviations: AIDR3D, Adaptive Iterative Dose Reduction using Three Dimensional Processing; SD, standard deviation; HU, Hounsfield unit.

Table 2 Correlations between patients’ body weights and measurement differences among three tube current settings

CT index	Reconstruction method	Correlation coefficient (ρ) and P-value
		240–120 mA	120–60 mA	240–60 mA
LAA%	AIDR3D	0.30 (NS)	0.24 (NS)	0.52 (P<0.05)
	FBP	0.52 (P<0.05)	0.68 (P<0.001)	0.68 (P<0.001)
15th percentile	AIDR3D	−0.21 (NS)	−0.33 (NS)	−0.47 (P<0.05)
	FBP	−0.42 (P<0.05)	−0.50 (P<0.05)	−0.64 (P<0.01)
Image noise	AIDR3D	0.01 (NS)	0.02 (NS)	0.06 (NS)
	FBP	0.44 (P<0.05)	0.61 (P<0.01)	0.53 (P<0.05)

Abbreviations: LAA%, percent low attenuation area; AIDR3D, Adaptive Iterative Dose Reduction using Three Dimensional Processing; FBP, filtered-back projection; NS, not significant; CT, computed tomography.