This copy is for personal use only. To order printed copies, contact [email protected] Purpose: To evaluate the effect of iterative reconstruction on the depiction of systemic sclerosis–related interstitial lung disease (ILD) when the radiation dose is reduced by 60%. Materials and Methods: This study was based on retrospective interpretation of prospectively acquired data over a 12-month period and approved by the institutional review board. The requirement to obtain informed consent was waived. Fifty-five chest computed tomographic (CT) examinations were performed in 38 women and 17 men (mean age, 55.8 years; range, 23–82 years) by using a dual-source CT unit with (a) both tubes set at similar energy (120 kVp) and (b) the total reference milliampere seconds (ie, 110 mAs) split up in a way that 40% was applied to tube A and 60% to tube B. Two series of images were generated simultaneously from the same dataset: (a) standard-dose images (generated from both tubes) reconstructed with filtered back projection (group 1, the reference standard) and (b) reduced-dose images (generated from tube A; 60% dose reduction) reconstructed with sinogram-affirmed iterative reconstruction (SAFIRE) (group 2). In both groups, the analyzed parameters comprised the image noise and the visualization and conspicuity of CT features of ILD. Two readers independently analyzed images from both groups. Results were compared by using the Wilcoxon test for paired samples; the 95% confidence interval was calculated when appropriate. Results: The mean level of objective noise in group 2 was significantly lower than that in group 1 (22.02 HU vs 26.23 HU, respectively; P , .0001). The CT features of ILD in group 1 were always depicted in group 2, with subjective conspicuity scores (a) improved in group 2 for ground-glass opacity, reticulation, and bronchiectasis and/or bronchiolectasis and (b) identical in both groups for honeycombing. The interobserver agreement for their depiction was excellent in both groups (k, 0.84–0.98). Conclusion: Despite a 60% dose reduction, images reconstructed with SAFIRE allowed similar detection of systematic sclerosis– related ILD compared with the reference standard. 1 From the Department of Thoracic Imaging (EA 2694) (F.P., A.S.B., J.B.F., J.R., M.R.J.), Department of Medical Statistics (EA 2694) (A.D.), Department of Internal Medicine (E.H.), and Department of Pulmonary Function (R.M.), Hôpital Calmette, Université Lille II, Boulevard Jules Leclercq, 59037 Lille, France; and Research and Development Department, Siemens Healthcare, Forchheim, Germany (B.S.). Received April 10, 2015; revision requested June 2; revision received August 10; accepted September 2; final version accepted September 21. Address correspondence to M.R.J. (e-mail: [email protected] ). q q RSNA, 2015 RSNA, 2015 Radiology: Volume 279: Number 1—April 2016 n radiology.rsna.org 297 n Thoracic Imaging François Pontana, MD Anne-Sophie Billard, MD Alain Duhamel, PhD Bernhard Schmidt, PhD Jean-Baptiste Faivre, MD Eric Hachulla, MD, PhD Régis Matran, MD Jacques Remy, MD Martine Remy-Jardin, MD, PhD Original Research Effect of Iterative Reconstruction on the Detection of Systemic Sclerosis–related Interstitial Lung Disease: Clinical Experience in 55 Patients1 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease P ulmonary involvement is common in patients with systemic sclerosis, and lung complications represent the most frequent cause of death in this population (1,2). In addition to clinical findings, staging of disease severity usually relies on pulmonary function tests and high-spatial-resolution computed tomography (CT), with the latter enabling characterization and quantification of lung changes. This morphologic information is important to identify early changes in asymptomatic patients and represents a method with which to monitor lung disease progression in patients with or without therapy (2–4). The CT features of systematic sclerosis–related interstitial lung disease (ILD) are characterized by subtle parenchymal changes (eg, ground-glass opacities, fine reticulation, and microcystic honeycombing), the detection of which could be altered on grainy images of reduced-dose acquisitions (5–8). To minimize noise in CT images, iterative reconstruction has been introduced as an alternative to the conventional reconstruction mode based on filtered back projection (FBP). The first generations of iterative reconstruction provided reduced-dose images with a similar quality to that of images acquired with a standard dose and reconstructed with FBP (9–11). However, these reconstruction algorithms (ie, image domain–based iterative reconstruction) modified the visual appearance of images, which could appear smooth (9) or pixelated (11), thus raising concerns about the possibility of maintaining the diagnostic capability of CT. A second generation of iterative reconstruction, active in the raw data space, was introduced with the aim of reducing noise and maintaining image sharpness. Their effect on lung images has mainly been evaluated with anthropomorphic phantoms, showing a similar detection of ground-glass opacities and pulmonary nodules on reduced-dose and standard-dose images (12–15). In clinical situations, a few studies have evaluated the visualization of lesions of lung infiltration in various disorders; however, none, to our knowledge, has specifically investigated ILD (9,11,16). Because systemic sclerosis–related ILD is characterized by mild forms of lung infiltration on CT scans sequentially indicated for follow-up purposes, this population can be considered as an interesting target to investigate the effect of a second-generation iterative reconstruction on reduced-dose images in comparison with images acquired with a standard dose and reconstructed with FBP. The purpose of this study was to evaluate the effect of iterative reconstruction on the depiction of systematic sclerosis–related ILD when the radiation dose is reduced by 60%. Materials and Methods One author (B.S., an employee of Siemens) provided technical support for implementation of the research prototype enabling reconstruction of reduced-dose and full-dose images from each data set. One author received research grants (M.R.J.) and one received consultant fees (J.R.) from Siemens. The other authors (F.P., A.S.B., A.D., J.B.F., E.H., R.M.), who are not associated with Siemens, maintained full control of the data at all times. The study was approved by the institutional ethics committee, and the requirement to obtain informed consent was waived. Advance in Knowledge Implication for Patient Care nn Mild interstitial lung abnormalities related to systemic sclerosis are confidently depicted at radiation dose levels of 1.8 mSv, with very good or excellent image quality in 96% of reduced-dose examinations nn Reduced-dose chest CT examinations reconstructed with sinogram-affirmed iterative reconstruction allow accurate depiction of CT features of systemic sclerosis–related interstitial lung disease. 298 Pontana et al Population During a 12-month period (May 2011– April 2012), 55 consecutive patients with systematic sclerosis (mean age, 55.8 years; range, 23–82 years) underwent a chest CT examination indicated for the diagnosis or follow-up of lung disease; detailed information about disease severity is given in the first paragraph of the Results section. The population comprised 38 women (mean age, 55.2 years; range, 23–77 years) and 17 men (mean age, 57.1 years; range, 32–82 years); there was statistically significant difference in the mean age of female and male patients (P = .64, Student t test). Recruitment for CT and assessment of clinical parameters was done prospectively, and image analysis was performed retrospectively. CT Scan Acquisition CT was performed with a dual-source 128-section multidetector CT system (Somatom Definition Flash; Siemens Healthcare, Germany) by using a dualsource research protocol provided by the manufacturer. Both x-ray tubes of the CT system were operated simultaneously at the same tube potential (120 kV) with the following parameters: collimation, Published online before print 10.1148/radiol.2015150849 Content code: Radiology 2016; 279:297–305 Abbreviations: BMI = body mass index FBP = filtered back projection ILD = interstitial lung disease SAFIRE = sinogram-affirmed iterative reconstruction Author contributions: Guarantors of integrity of entire study, F.P., A.S.B., R.M., J.R., M.R.J.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, F.P., A.S.B., J.R., M.R.J.; clinical studies, F.P., A.S.B., J.B.F., R.M., J.R., M.R.J.; statistical analysis, F.P., A.S.B., A.D.; and manuscript editing, F.P., A.S.B., B.S., E.H., J.R., M.R.J. Conflicts of interest are listed at the end of this article. See also the editorial by Fletcher et al and articles by Solomon et al and Mileto et al in this issue. radiology.rsna.org n Radiology: Volume 279: Number 1—April 2016 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease 64 3 2 3 0.6 mm with z-flying focal spot; rotation time, 0.28 second; and pitch, 2. The total reference milliampere second was split between both x-ray tubes in a way that 40% of the reference milliampere second was applied to tube A (44 mAs) and 60% of the reference milliampere second was applied to tube B (66 mAs) with a four-dimensional dose modulation system (Care Dose 4D, Siemens Healthcare). There were 43 unenhanced and 12 contrast material–enhanced CT examinations. Image Reconstruction Two sets of lung images were reconstructed from each acquisition. The first set of images were standard-dose images (group 1) consisting of 1-mmthick contiguous images reconstructed with FBP by using a high-spatial-resolution kernel (B50). The images were generated by combining data from both tubes so that the resulting images used the total applied reference milliampere second, that is, 110 mAs. The second set of images were reduced-dose images (group 2) consisting of 1-mm-thick contiguous images reconstructed with sinogram-affirmed iterative reconstruction (SAFIRE). The images were generated only with data from tube A, which used 40% of the total applied reference milliampere second and thus represent images acquired with 60% dose reduction. The manufacturer designed the iterative reconstruction kernel (I50) to closely match the spatial resolution of the corresponding FBP kernel (B50). The modulation transfer functions of both kernels are designed to be very similar in terms of their characteristic spatial frequencies—namely the 50% value, 10% value, and 2% value. Images were viewed by using standard lung parenchymal window settings (window width, 1600 HU; window center, 2600 HU). Analyzed Parameters Population characteristics and radiation dose estimation.—We recorded the patients’ age, weight, body mass index (BMI), CT dose index, doselength product, and size-specific dose estimate. Radiology: Volume 279: Number 1—April 2016 n Image noise.—The objective assessment of noise was performed by measuring the standard deviation of pixel values in homogeneous regions of interest within the tracheal lumen above the aortic arch. Circular regions of interest within the tracheal lumen (mean 6 standard deviation of the region of interest surface: 1.48 cm2 6 0.53; range: 0.69–3.10 cm2) were placed by one reader (A.S.B.), a junior radiologist with 3 years of experience in chest CT. The visual perception of noise, defined by the grainy appearance of images, was independently evaluated by the junior reader (A.S.B.) and a senior reader (F.P., with 8 years of experience in chest CT) with use of a threepoint scale, as follows: 1 = minimal, 2 = moderate, and 3 = severe. A noise level score of 1 or 2 indicated that the noise did not alter the identification of normal and/or abnormal structures. A noise level score of 3 indicated that the identification of normal and/or abnormal structures was altered. Overall image quality.—The overall image quality of group 1 and group 2 images was rated by the same individuals by using a five-point scale (11), as follows: 1 = excellent, 2 = very good, 3 = satisfactory, 4 = suboptimal, and 5 = nondiagnostic. Scores of 1–3 indicated diagnostic image quality; scores of 4 and 5 corresponded to nondiagnostic image quality. Visualization of normal and abnormal lung structures.—The following anatomic structures were analyzed: (a) fissures; (b) proximal bronchi and adjacent pulmonary vessels (ie, structures down to the subsegmental level, further referred to as proximal bronchi and vessels); (c) peripheral bronchi and adjacent pulmonary vessels (ie, structures beyond the subsegmental level); and (d) vascular structures located within 10 mm of the pleura, further referred to as subpleural vessels (17). The presence of five abnormal structures was determined at a lobar level, considering six lobes per patient (namely the right upper, middle, and lower lobes; left upper culmen; and lingula and lower lobes) and a total of 330 lobes in our study group. The following five abnormal structures were analyzed: radiology.rsna.org Pontana et al (a) ground-glass opacities; (b) reticulations; (c) CT features of lung fibrosis, including bronchiectasis and/or bronchiolectasis; (d) signs of architectural distortion (on bronchi, vessels, fissures, and/or interlobular septa); and (e) honeycombing. The visibility of normal and abnormal lung structures was analyzed on all images by using a five-point scale (18), as follows: 1 = excellent image quality with sharp demarcation of structures; 2 = slight blurring of the structures, with unrestricted image evaluation; 3 = moderate blurring of the interface structures, with slight restricted assessment; 4 = severe blurring and poorly defined structures, with uncertainty about the evaluation; and 5 = severely reduced image quality, making reliable interpretation impossible. The rating of abnormal structures also included (a) an additional score (score, 0) in the absence of abnormal findings (18) and (b) comparison of the visibility of the five abnormal structures within the 330 lobes examined in our study group. Lung infiltration quantification scores.—Because image quality can influence the analysis of the extent and severity of ILD, the study included the evaluation of the coarseness score, as described by Desai et al (6), and the extent score, as described by Goh et al (19), on the group 1 and group 2 images from each patient. Lung images were scored at five levels: (a) origin of great vessels, (b) main carina, (c) pulmonary venous confluence, (d) halfway between the third and fifth sections, and (e) immediately above the right hemidiaphragm. As proposed by Desai et al (6), the coarseness of fibrosis was quantified semiquantitatively as follows: 0 = ground-glass opacification alone, 1 = fine intralobular fibrosis, 2 = microcystic reticular pattern (ie, airspaces 4 mm), and 3 = macrocystic reticular pattern (ie, airspaces .4 mm).The total coarseness score was the summed score for all five levels (range, 0–15). The total extent of ILD was estimated at the nearest 5% in each of the five sections, with global extent of disease on lung images (ie, Goh score) as the mean of the scores. 299 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease Conditions of Image Analysis Images from groups 1 and 2 were anonymized and transferred to a clinical workstation (Syngo CT Workplace, Siemens Healthcare). Two readers (A.S.B., F.P.) analyzed group 1 and group 2 images from each patient in random order while blinded to the reconstruction technique. This was possible because of the lack of pixilated and/ smooth appearance of images reconstructed with this second-generation algorithm, a finding derived from our daily use of SAFIRE on reduced-dose chest CT examinations. After anonymization, the images from both groups were randomized and mixed together. The two types of images from one patient were not read at the same session. Subjective noise, visibility of normal and abnormal structures, and overall image quality of group 1 and group 2 images were independently evaluated by the two readers. Determination of the coarseness and extent was obtained by consensus between the two readers, considering that the analysis of the effect of iterative reconstruction on the extent and coarseness of ILD was a secondary objective of this study. Measurement of objective image noise was performed by one reader (A.S.B.). This analysis had been undertaken before the anonymization of images by positioning a circular region of interest at the level of the trachea on group 1 and group 2 images from each patient, both being simultaneously available on the workstation. This allowed the reader to generate strictly similar regions of interest on both images for each patient. Statistical Analysis We did not calculate an a priori sample size and decided to include all consecutive patients with systemic sclerosis referred to our department during the study period, that is, 55 patients. On the basis of this study population, we performed a post hoc power analysis by calculating the smallest difference (expressed as effect size by using the standardized mean difference) that we could detect with 80% power. The minimum effect size was 0.39, which was considered moderate (20). 300 Pontana et al Table 1 Interobserver Agreement for Subjective Criteria Analysis Parameter Patient-based analysis (n = 55)* Subjective noise Overall image quality Visibility of normal lung structures Fissures Proximal bronchi Peripheral bronchi Subpleural vessels Lobar-based analysis (n = 330)† Ground-glass opacities Reticulation Bronchiectasis and bronchiolectasis Architectural distortion Honeycombing Group 1 Group 2 52 (94.5) 49 (89) 52 (94.5) 49 (89) 49 (89) 38 (69) 43 (78) 38 (69) 48 (87) 41 (74.5) 43 (78) 48 (87) 0.96 0.96 0.90 0.84 0.98 0.94 0.92 0.88 0.84 0.98 Note.—Group 1 = standard-dose images with FBP, group 2 = reduced-dose images with SAFIRE. * Data are numbers of patients (n = 55), with percentages in parentheses, and are the concordance rate. † Data represent the visibility of the abnormal structures and are k values. Statistical analyses were performed by using software (SAS, version 9.3; SAS Institute, Cary, NC). Results were compared by using the Wilcoxon test for paired samples. The 95% confidence interval was described when appropriate. The interobserver agreement was expressed with a concordance rate for the patient-based analysis and with a weighted k value for the lobarbased analysis. The k coefficients were interpreted as follows: less than 0, disagreement; 0.0–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, good agreement; and 0.81–1.00, almost perfect agreement (21). P , .05 was indicative of a statistically significant difference. Results Characteristics of the Study Population The mean age (6standard deviation) of the study population was 55.8 years 6 13.8 (range: 23–82 years), with a female (38 women, 17 men) and nonsmoker (31 nonsmokers, 24 active or ex-smokers) predominance. On the basis of clinical findings, systemic sclerosis was limited in 38 of the 55 patients (69%) and diffuse in 17 (31%). The mean duration of the disease at the time of CT examination was 7.6 years (range: 0.1–31 years; median, 5 years). The mean BMI was 27.2 kg/m2 6 5.8 (range, 16.9–37.8 kg/m2). Two patients were underweight (BMI ,18.5 kg/m2), 20 were within the normal range (BMI, 18.5–24.9 kg/m2), 13 were overweight (BMI, 25–29.9 kg/m2), and 19 were obese (BMI .30 kg/m2). Results of pulmonary function tests, expressed as mean percentages (6standard deviations) of predicted values, were within normal limits except for diffusing capacity of lung for carbon monoxide. Results are as follows : (a) total lung capacity, 87.7% 6 20.4; (b) forced vital capacity, 91.5% 6 23; (c) forced expiratory volume in 1 second, 85.5% 6 19.5; (d) ratio of forced expiratory volume in 1 second to vital capacity, 78.7% 6 8.8; and (e) diffusing capacity of lung for carbon monoxide, 53.5% 6 22.4. The population comprised 35 patients with diffusing capacity of lung for carbon monoxide less than 70% predicted, 10 with forced expiratory volume in 1 second less than 70% predicted, and 11 with total lung capacity less than 70% predicted. CT features of lung infiltration were present in 38 of the 55 patients (69%) and included ground-glass opacities in 38 radiology.rsna.org n Radiology: Volume 279: Number 1—April 2016 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease patients (69%), reticulation in 38 (69%), bronchiectasis and/or bronchiolectasis in 27 (49%), architectural distortion in 27 (49%), and honeycombing in 17 (31%). The mean number of lobes with abnormal CT features was 5.2 6 1.4 (range: 2–6), and the mean number of abnormal CT features per lobe was 3.3 6 1.5 (range: 1–5). The mean dose-length product of CT examinations was 216.09 mGy · cm 6 58.83 (range: 118–343 mGy · cm). The mean size-specific dose estimate was 9.94 mGy 6 3.62 (range: 4.98–25.52 mGy). The mean effective dose, calculated by multiplying the doselength product by a conversion factor of 0.017 (22), was 3.67 mSv 6 1.00 (range: 2.01–5.83 mSv). In group 2, the theoretical mean dose-length product was 86.44 mGy · cm (range: 48.8–208.8 mGy · cm), which corresponds to a mean effective dose of 1.8 mSv. Interobserver Agreement for Subjective Analysis Concordance rates were similar between groups 1 and 2 for subjective noise analysis and overall image quality and higher in group 2 for the visibility of proximal bronchi and subpleural vessels, whereas they were similar for peripheral bronchi in both groups and inferior for fissure analysis in group 2 (Table 1). The interobserver agreement was excellent for the analysis of abnormal lung structures (k: 0.84–0.98). Noise and Overall Image Quality The mean level of objective noise in group 2 images was significantly lower than that in group 1 images (P , .0001, Table 2). There was no significant difference in the distribution of subjective noise scores between the two groups, which was mainly rated as moderate (94%) in both groups and never downgraded in group 2 compared with group 1 images. There was no significant difference in the distribution of image quality scores between the two groups, with images classified as very good (score, 2) for 52 of 55 patients in group 1 (94.5%) and 49 of 55 patients in group 2 (89%); the proportion of images with excellent image quality (score, 1) was higher in group 2 (four Radiology: Volume 279: Number 1—April 2016 n radiology.rsna.org Pontana et al Table 2 Comparison of Noise and Image Quality between Groups 1 and 2 Parameter Objective noise (HU) Mean Range Median Subjective noise Minimal (score, 1) Moderate (score, 2) Severe (score, 3) Overall image quality Excellent (score, 1) Very good (score, 2) Satisfactory (score, 3) Suboptimal (score, 4) Nondiagnostic (score, 5) Group 1 Group 2 26.23 6 9.40 15.6–52.4 22.6 22.02 6 6.95 13.2–44 19.6 2 (3.6) 52 (94.5) 1 (1.8) 3 (5.5) 52 (94.5) 0 1 (1.8) 52 (94.5) 2 (3.6) 0 0 4 (7.3) 49 (89) 2 (3.6) 0 0 P Value ,.0001 NA NA Note.—Except where indicated, data are numbers of patients (n = 55), with percentages in parentheses. Group 1 = standarddose images with FBP, group 2 = reduced-dose images with SAFIRE. NA = not applicable. Table 3 Comparison of Normal Lung Structure Visibility between Groups 1 and 2 Parameter Fissures Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Proximal bronchi and vessels Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Peripheral bronchi and vessels Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Subpleural vessels Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Group 1 Group 2 5 (9.1) 49 (89.1) 1 (1.8) 0 0 7 (12.7) 47 (85.5) 1 (1.8) 0 0 7 (12.7) 47 (85.5) 1 (1.8) 0 0 10 (18.2) 45 (81.8) 0 0 0 2 (3.7) 52 (94.5) 1 (1.8) 0 0 7 (12.7) 47 (85.5) 1 (1.8) 0 0 1 (1.8) 50 (90.9) 4 (7.3) 0 0 2 (3.7) 51 (92.6) 2 (3.7) 0 0 Note.—Data are numbers of patients (n = 55), with percentages in parentheses. Group 1 = standard-dose images with FBP, group 2 = reduced-dose images with SAFIRE. 301 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease of 55 patients, 7.3%) than in group 1 (one of 55 patients, 1.8%). A single set of group 2 images was rated with lower image quality (score, 3) compared with the corresponding group 1 set of images (score, 2) (one of 55 patients, 1.82% [95% confidence interval = 0.05%, 9.72%]). No series of images was classified as having suboptimal (score, 4) or nondiagnostic (score, 5) image quality. Visualization of Normal and Abnormal Lung Structures There was no difference between groups in the distribution of conspicuity scores of normal lung structures (Table 3). Most structures were given a score of 2 in both groups; there was a higher proportion of scores of 1 in group 2 than in group 1 and no scores of 4 or 5 in either group. There was no significant difference in the distribution of scores with regard to the visibility of abnormal lung structures (Table 4). Most CT features were given a score of 2 in both groups. Except for the honeycombing, which received a similar rating for both groups, there was a higher proportion of scores of 1 in group 2 than in group 1 (Figs 1–4). Extent of ILD Because the coarseness score was based on the analysis of five sections per patient, a total of 275 sections were analyzed in the study group; 164 of the 275 sections (60%) had signs of infiltration in groups 1 and 2. There was no difference in the mean coarseness score between the two groups (group 1: mean = 6.34 6 3.17, range = 0–12, median = 5; group 2: mean = 6.29 6 3.17, range = 0–12, median = 5). The extent score was similar in both groups: 39 of the 55 patients (70.9%) had lung infiltration of less than 20% and 16 (29.1%) had lung infiltration of more than 20%. Discussion The results of our study showed no significant difference in the distribution of overall image quality scores between our two groups. All series of images were rated with a diagnostic image quality—even those in group 2, which 302 Pontana et al Table 4 Comparison of Abnormal Lung Structure Visibility between Groups 1 and 2 Parameter Ground-glass opacity Absent (score, 0) Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Reticulation Absent (score, 0) Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Bronchiectasis and bronchiolectasis Absent (score, 0) Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Honeycombing Absent (score, 0) Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Architectural distortion Absent (score, 0) Sharp delineation (score, 1) Slight blurring (score, 2) Moderate blurring (score, 3) Severe blurring (score, 4) Unreliable analysis (score, 5) Group 1 Group 2 127 (38.5) 4 (1.2) 199 (60.3) 0 0 0 127 (38.5) 14 (4.2) 189 (57.3) 0 0 0 143 (43.3) 4 (1.2) 183 (55.5) 0 0 0 143 (43.3) 14 (4.2) 173 (52.4) 0 0 0 188 (57.0) 4 (1.2) 134 (40.6) 4 (1.2) 0 0 188 (57.0) 7 (2.1) 130 (39.4) 5 (1.5) 0 0 244 (73.9) 0 82 (24.8) 3 (0.9) 0 0 244 (73.9) 0 82 (24.8) 3 (0.9) 0 0 186 (56.4) 3 (0.9) 138 (41.8) 3 (0.9) 0 0 186 (56.4) 8 (2.4) 131 (39.7) 5 (1.5) 0 0 Note.—Data are numbers of lobes (n = 330), with percentages in parentheses. Group 1 = standard-dose images with FBP, group 2 = reduced-dose images with SAFIRE. were generated with a 60% dose reduction corresponding to a mean effective dose of 1.8 mSv. A higher proportion of images were rated with excellent image quality in group 2 than in group 1 that is likely to be linked to the significantly lower noise level, a major factor in the analysis of subjective image quality (23). Our findings are consistent with those of Baumueller et al (24), who reported a significant reduction in image noise on submillisievert images reconstructed with SAFIRE compared with standard-dose images with FBP. In contrast to studies performed with the first generations of iterative reconstruction (9,11), we did not observe a pixelated and/or smoothed appearance for images reconstructed with SAFIRE. With regard to the visibility of CT features of systematic sclerosis–related ILD, there was no significant difference in the distribution of ratings between the two groups. The extent of disease and the severity of lung infiltration, as assessed with the coarseness and extent scores, radiology.rsna.org n Radiology: Volume 279: Number 1—April 2016 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease Figure 1 Figure 1: Transverse CT scans obtained in lower lung zones (magnified view of right lower lobe) in 27-year-old patient with normal BMI (23.8 kg/m2). Subpleural ground-glass opacity with fine reticulation was given a similar rating (score, 2) on (a) standard-dose image reconstructed with FBP and (b) reduceddose image reconstructed with SAFIRE. Objective noise was 17.8 HU on group 1 images and 17.2 HU on group 2 images. Figure 2 Figure 2: Transverse CT scans obtained in lower lung zones (magnified view of right lower lobe) in 65-year-old patient with normal BMI (23.4 kg/m2). Ground-glass opacity of low attenuation, which is diffusely distributed in lower lobe and disseminated in right middle lobe, was given a similar rating (score, 2) on (a) standard-dose image reconstructed with FBP and (b) reduced-dose image reconstructed with SAFIRE. Objective image noise was lower in group 2 images (16.8 HU) than in group 1 images (17.4 HU). did not significantly differ between the two groups. These results demonstrate the effectiveness of SAFIRE in restoring a diagnostic image quality of reduceddose acquisitions. In a recent study performed to evaluate the ability of low-dose CT to depict and help characterize the most common patterns of pulmonary disease, Christe et al (8) reported a 20% Radiology: Volume 279: Number 1—April 2016 n decrease in sensitivity but also a drift in recognition of faint abnormalities on images reconstructed with FBP instead of iterative reconstruction. Because honeycombing is known to be detectable even on grainy images (8), it was not surprising to observe similar ratings in our two groups. More interesting to note was the higher scores of excellent radiology.rsna.org Pontana et al visibility observed in group 2 for the other subtle interstitial patterns. These results are consistent with the improved depiction of ground-glass opacities, pulmonary nodules, and emphysema previously reported with the first generation of iterative reconstruction compared with FBP (9). They also confirm in vivo the results recently reported in an experimental study concerning the detection of ground-glass nodules (25). They are comparable to the excellent intermethod agreement demonstrated by Ohno et al (16) for the detection of emphysema, ground-glass opacities, bronchiectasis, honeycombing, and nodules when reduced-dose images are reconstructed with iterative reconstruction. The absence of negative effect of iterative reconstruction on the detection of groundglass opacities is particularly important for systematic sclerosis when keeping in mind that it may reflect not only the presence of interstitial inflammation but also fibrotic changes (26). The excellent interobserver agreement observed in our study is in keeping with that reported by Baumueller et al (24) in a wide variety of chest diseases and should be linked to the low level of image noise (27). Our study design provided a means of comparing image quality on full-dose and reduced-dose images for each patient. This situation is more appropriate for image quality analysis than is relying on the comparison of distinct groups of patients. Although often paired according to BMI, populations may differ in chest dimension, degree of inspiration, and/or underlying disorders that may affect comparative analysis of image quality. Our approach also allows more accurate intertechnique comparison because the whole lung volume is analyzed on the two sets of images. This is more robust than the methodology proposed by Sverzellati et al (28), who added three CT sections acquired at reduced dose to a standard-dose study covering the whole chest volume. Iterative reconstructions are expected to influence the dose reduction strategy that can be proposed to patients with systematic sclerosis–related ILD. The similarities in radiation dose levels reported in our study (1.8 mSv in group 2) and with sequential 303 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease Pontana et al Figure 3 Figure 3: Transverse CT scans obtained in lower lung zones (magnified view of right lower lobe) in 66-year-old overweight patient (BMI: 26.2 kg/m2). CT features of lung fibrosis in right middle lobe and right lower lobe are ground-glass opacities of high attenuation, traction bronchiolectasis, and fissural distortion. Visibility scores of individual CT features were worse on (a) standard-dose image reconstructed with FBP (ground-glass opacity: score, 2; traction bronchiolectasis: score, 2; distortion: score, 2) than on (b) reduceddose images reconstructed with SAFIRE (ground-glass opacity: score, 1; bronchiolectasis: score, 1; distortion: score, 1). Objective noise was lower in group 2 images (17.4 HU) than in group 1 images (19.6 HU). Figure 4 Figure 4: Transverse CT scans obtained at level of right inferior pulmonary vein (magnified view centered on right lung) in 47-year-old obese patient (BMI: 31.3 kg/m2). CT features of lung fibrosis were traction bronchiectasis in right middle lobe and microcystic changes, reticulation, and signs of distortion in right lower lobe, with concurrent presence of mild ground-glass opacities in both lobes. Visibility scores of individual CT features were identical for both (a) standard-dose image reconstructed with FBP and (b) reduced-dose image reconstructed with SAFIRE (traction bronchiectasis: score, 1; microcystic changes suggestive of honeycombing: score, 2; reticulation: score, 2; distortion: score, 2). Objective noise was lower in group 2 images (24.1 HU) than in group 1 images (28.8 HU). scanning (1.6 mSv) strongly favor reduced-dose volumetric acquisitions with images reconstructed with iterative 304 reconstruction instead of sequential scanning at 10-mm intervals with images reconstructed with FBP (29). This approach can be extended to the entire spectrum of ILDs because it allows a more comprehensive assessment than that provided by a reduced number of sections (30). This approach should also facilitate quantitative assessment of ILD, which is useful for screening but also for longitudinal monitoring (28,31–33). Our study has some limitations. With regard to the scanning parameters, we used the same kilovoltage and same reference milliampere second for all patients regardless of their BMI. Although this had no adverse consequence for thin patients, it could have been responsible for high noise levels in obese patients–especially with reduced-dose images. This limitation was partly compensated for by the systematic use of automatic tube current modulation but could not be avoided because of our study design. Second, the study population was composed of only 55 patients, a situation linked to the low prevalence of this disease and the limited availability of the research prototype. Third, we should underline that 12 chest CT examinations were performed after injection of contrast material, which may influence the assessment of ground-glass opacity on lung images (34). Fourth, a limitation related to the power of the study must be underlined. We cannot exclude that some differences between the two groups could have been overlooked because of the effect size. However, results of post hoc analysis showed that a moderate range of differences between the two techniques was accessible, thus limiting the risks of overestimation of the effect of iterative reconstructions. Fifth, one should underline a limitation of the subjective scoring system used. The use of a five-point scale may have obscured small image quality differences between the two series of images. Last, the CT features studied did not include all categories of lung abnormalities that can be encountered in respiratory diseases. However, the absence of a negative effect of iterative reconstruction for the depiction of subtle lesions suggests that similar conclusions could be drawn for more severe forms of infiltration and/or pulmonary destruction. Despite a 60% dose reduction, SAFIRE allows evaluation of systematic radiology.rsna.org n Radiology: Volume 279: Number 1—April 2016 THORACIC IMAGING: Detection of Systemic Sclerosis–related Interstitial Lung Disease Pontana et al sclerosis–related ILD on reduced-dose images as accurate as that provided by standard-dose images reconstructed with FBP. 10. Pontana F, Pagniez J, Flohr T, et al. Chest computed tomography using iterative reconstruction vs filtered back projection. I. Evaluation of image noise reduction in 32 patients. Eur Radiol 2011;21(3):627–635. 23. Mayo JR, Hartman TE, Lee KS, Primack SL, Vedal S, Müller NL. CT of the chest: minimal tube current required for good image quality with the least radiation dose. AJR Am J Roentgenol 1995;164(3):603–607. Disclosures of Conflicts of Interest: F.P. disclosed no relevant relationships. A.S.B. disclosed no relevant relationships. A.D. disclosed no relevant relationships. B.S. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is an employee of Siemens Healthcare. Other relationships: disclosed no relevant relationships. J.B.F. disclosed no relevant relationships. E.H. disclosed no relevant relationships. R.M. disclosed no relevant relationships. J.R. disclosed no relevant relationships. M.R. disclosed no relevant relationships. 11. Prakash P, Kalra MK, Ackman JB, et al. Diffuse lung disease: CT of the chest with adaptive statistical iterative reconstruction technique. Radiology 2010;256(1):261–269. 24. Baumueller S, Winklehner A, Karlo C, et al. Low-dose CT of the lung: potential value of iterative reconstructions. Eur Radiol 2012; 22(12):2597–2606. 12. Higuchi K, Nagao M, Matsuo Y, et al. Detection of ground-glass opacities by use of hybrid iterative reconstruction (iDose) and low-dose 256-section computed tomography: a phantom study. Radiol Phys Technol 2013;6(2):299–304. 25. Christe A, Lin MC, Yen AC, et al. CT patterns of fungal pulmonary infections of the lung: comparison of standard-dose and simulated low-dose CT. Eur J Radiol 2012;81(10):2860– 2866. References 1. Ioannidis JP, Vlachoyiannopoulos PG, Haidich AB, et al. Mortality in systemic sclerosis: an international meta-analysis of individual patient data. Am J Med 2005;118(1):2–10. 2. Steen VD, Medsger TA. Changes in causes of death in systemic sclerosis, 1972–2002. Ann Rheum Dis 2007;66(7):940–944. 3. Wells AU, Rubens MB, du Bois RM, Hansell DM. Functional impairment in fibrosing alveolitis: relationship to reversible disease on thin section computed tomography. Eur Respir J 1997;10(2):280–285. 4. Wells AU, Rubens MB, du Bois RM, Hansell DM. Serial CT in fibrosing alveolitis: prognostic significance of the initial pattern. AJR Am J Roentgenol 1993;161(6):1159–1165. 5. Remy-Jardin M, Remy J, Wallaert B, Bataille D, Hatron PY. Pulmonary involvement in progressive systemic sclerosis: sequential evaluation with CT, pulmonary function tests, and bronchoalveolar lavage. Radiology 1993;188(2):499–506. 6. Desai SR, Veeraraghavan S, Hansell DM, et al. CT features of lung disease in patients with systemic sclerosis: comparison with idiopathic pulmonary fibrosis and nonspecific interstitial pneumonia. Radiology 2004;232(2):560–567. 7. Zwirewich CV, Mayo JR, Müller NL. Lowdose high-resolution CT of lung parenchyma. Radiology 1991;180(2):413–417. 8. Christe A, Charimo-Torrente J, Roychoudhury K, Vock P, Roos JE. Accuracy of lowdose computed tomography (CT) for detecting and characterizing the most common CT patterns of pulmonary disease. Eur J Radiol 2013;82(3):e142–e150. 9. Pontana F, Duhamel A, Pagniez J, et al. Chest computed tomography using iterative reconstruction vs filtered back projection. II. Image quality of low-dose CT examinations in 80 patients. Eur Radiol 2011;21(3):636–643. Radiology: Volume 279: Number 1—April 2016 n 13. Mathieu KB, Ai H, Fox PS, et al. Radiation dose reduction for CT lung cancer screening using ASIR and MBIR: a phantom study. J Appl Clin Med Phys 2014;15(2):4515. 14. Hashemi S, Mehrez H, Cobbold RS, Paul NS. Optimal image reconstruction for detection and characterization of small pulmonary nodules during low-dose CT. Eur Radiol 2014;24(6):1239–1250. 15. Rampinelli C, Origgi D, Vecchi V, et al. Ultra-low-dose CT with model-based iterative reconstruction (MBIR): detection of ground-glass nodules in an anthropomorphic phantom study. Radiol Med (Torino) 2015;120(7):611–617. 16. Ohno Y, Takenaka D, Kanda T, et al. Adaptive iterative dose reduction using 3D processing for reduced- and low-dose pulmonary CT: comparison with standarddose CT for image noise reduction and radiological findings. AJR Am J Roentgenol 2012;199(4):W477–W485. 17. Bongartz G, Golding SJ, Jurik AG, et al. European guidelines for multislice computed tomography. Contract number FIGM-CT 2000-20078-CT-TIP. http://www.msct.eu/ CT_Quality_Criteria.htm. Published online March 2004. Accessed April 2012. 18. Studler U, Gluecker T, Bongartz G, Roth J, Steinbrich W. Image quality from high-resolution CT of the lung: comparison of axial scans and of sections reconstructed from volumetric data acquired using MDCT. AJR Am J Roentgenol 2005;185(3):602–607. 19. Goh NSL, Desai SR, Veeraraghavan S, et al. Interstitial lung disease in systemic sclerosis: a simple staging system. Am J Respir Crit Care Med 2008;177(11):1248–1254. 20. Cohen J. A power primer. Psychol Bull 1992; 112(1):155–159. 21. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159–174. 22. McCollough CH, Primak AN, Braun N, Kofler J, Yu L, Christner J. Strategies for reducing radiation dose in CT. Radiol Clin North Am 2009;47(1):27–40. radiology.rsna.org 26. Remy-Jardin M, Giraud F, Remy J, Copin MC, Gosselin B, Duhamel A. Importance of ground-glass attenuation in chronic diffuse infiltrative lung disease: pathologic-CT correlation. Radiology 1993;189(3):693– 698. 27. Mayo JR, Kim KI, MacDonald SL, et al. Reduced radiation dose helical chest CT: effect on reader evaluation of structures and lung findings. Radiology 2004;232(3):749–756. 28. Sverzellati N, Zompatori M, De Luca G, et al. Evaluation of quantitative CT indexes in idiopathic interstitial pneumonitis using a low-dose technique. Eur J Radiol 2005; 56(3):370–375. 29. Frauenfelder T, Winklehner A, Nguyen TD, et al. Screening for interstitial lung disease in systemic sclerosis: performance of highresolution CT with limited number of slices—a prospective study. Ann Rheum Dis 2014;73(12):2069–2073. 30. Bendaoud S, Remy-Jardin M, Wallaert B, et al. Sequential versus volumetric computed tomography in the follow-up of chronic bronchopulmonary diseases: comparison of diagnostic information and radiation dose in 63 adults. J Thorac Imaging 2011;26(3):190–195. 31. Park SC, Tan J, Wang X, et al. Computeraided detection of early interstitial lung diseases using low-dose CT images. Phys Med Biol 2011;56(4):1139–1153. 32. Orlandi I, Camiciottoli G, Diciotti S, et al. Thin-section and low-dose volumetric computed tomographic densitometry of the lung in systemic sclerosis. J Comput Assist Tomogr 2006;30(5):823–827. 33. Yabuuchi H, Matsuo Y, Tsukamoto H, et al. Evaluation of the extent of ground-glass opacity on high-resolution CT in patients with interstitial pneumonia associated with systemic sclerosis: comparison between quantitative and qualitative analysis. Clin Radiol 2014;69(7):758–764. 34. Lloyd CR, Walsh SL, Hansell DM. Highresolution CT of complications of idiopathic fibrotic lung disease. Br J Radiol 2011;84 (1003):581–592. 305