Cardiopulmonary Imaging Original Research

Transcription

1 Cardiopulmonary Imaging Original Research Ohno et al. Pulmonary CT Using 3D Adaptive Iterative Dose Reduction Cardiopulmonary Imaging Original Research Yoshiharu Ohno 1,2 Daisuke Takenaka 1,3 Tomonori Kanda 1,3 Takeshi Yoshikawa 1 Sumiaki Matsumoto 1 Naoki Sugihara 4 Kazuro Sugimura 1 Ohno Y, Takenaka D, Kanda T, et al. Keywords: CT, image quality, iterative reconstruction, radiation dosage, reconstruction algorithm DOI: /AJR Received November 18, 2011; accepted after revision February 29, This work was supported by a grant from Toshiba Medical Systems and a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (JSTS KAKEN no ). 1 Department of Radiology, Kobe University Graduate School of Medicine, Kusunoki-cho, Chuo-ku, Kobe, Hyogo , Japan. Address correspondence to Y. Ohno (yosirad@kobe-u.ac.jp, yosirad@med.kobe-u.ac.jp, or yoshiharuohno@aol.com). 2 Advanced Biomedical Imaging Research Center, Kobe University Graduate School of Medicine, Kobe, Hyogo, Japan. 3 Department of Radiology, Hyogo Cancer Center, Akashi, Hyogo, Japan. 4 Toshiba Medical Systems, Ohtawara, Tochigi, Japan. WEB This is a Web exclusive article. AJR 2012; 199:W477 W X/12/1994 W477 American Roentgen Ray Society Adaptive Iterative Dose Reduction Using 3D Processing for Reducedand Low-Dose Pulmonary CT: Comparison With Standard-Dose CT for Image Noise Reduction and Radiological Findings OBJECTIVE. The purpose of this study was to determine the utility of adaptive iterative dose reduction using 3D processing (AIDR 3D) for image noise reduction and assessment of radiologic findings obtained with reduced- and low-dose chest CT in patients with various pulmonary diseases. SUBJECTS AND METHODS. Chest CT examinations at three different tube current settings and using 16- and 64-MDCT scanners were performed for 37 patients. Standarddose (150 mas) data were reconstructed as thin-section CT without AIDR 3D, and low-dose (25 mas) and reduced-dose (50 mas) data were reconstructed as thin-section CT without and with AIDR 3D. To compare image quality, image noises at all CT doses were quantitatively assessed by region of interest measurements. For comparison of radiologic finding assessments, likelihoods of occurrence of emphysema, ground-glass opacity, reticular opacity, bronchiectasis, honeycomb pattern, and nodules were evaluated on a 5-point scale. Then, image noise and agreements of radiologic findings between standard-dose CT and others were statistically evaluated. RESULTS. The image quality scores of reduced- and low-dose CT without AIDR 3D were significantly lower than those of both protocols with AIDR 3D and standard-dose CT (p < 0.05). All intermethod agreements for emphysema, ground-glass opacity, bronchiectasis, honeycomb pattern, and nodules, except for those observed on low-dose CT without AIDR 3D, were almost perfect (κ > 0.81). CONCLUSION. AIDR 3D is useful for image noise reduction and assessment of radiologic findings obtained with reduced- and low-dose CT for patients with various pulmonary diseases. S ince the beginning of clinical application of MDCT in the late 1990s, MDCT has played a very important role in routine clinical practice. In addition, academic and social interest in radiation dose reduction for CT examinations without any decrease in diagnostic capability has been growing. The issue of radiation dose reduction is currently drawing widespread attention. However, the application in routine clinical practice of dose reduction for various CT techniques varies among institutions and scanners. In the last decade, dose reduction strategies have been realized by using various techniques, such as tube current reduction, tube voltage reduction, increased helical pitch, scan length optimization, scan protocol individualization, and utilization of automatic exposure control (AEC) [1]. Such radiation dose reduction should, of course, be attained without deterioration in the image quality of the examination. However, image noise is inversely proportional to the square root of the radiation dose, so that reduceddose CT images have a higher noise level than standard-dose images, and care must be exercised to ensure that the former remain suitable for diagnosis. To overcome the increase in image noise on reduced-dose CT images, various imaging filters, reconstruction algorithms, and kernels have been developed and adapted by institutions, companies, and organizations involved in routine clinical practice. Image filters are software applications designed to improve image quality by removing noise and artifacts. Currently, two types of image filters are in use: one is a spatial domain filter, which manipulates data in the reconstructed images, and the other is a raw data based filter, which modulates the data in the raw data domain before re- W477

2 Ohno et al. construction [1]. These techniques have been used in conjunction with filtered back projection (FBP), which has been adapted for reconstruction of CT images from raw data over the past couple of decades. Recent advancements in central processing units and enhanced computer performance have made it possible to use iterative reconstruction algorithms in routine clinical practice. The standard FBP algorithm is based on several fundamental assumptions about CT geometry and is a compromise between reconstruction speed and image noise. In contrast to the standard FBP algorithm, iterative reconstruction algorithms, such as adaptive statistical iterative reconstruction and iterative reconstruction in image space, are based on different assumptions about CT geometry, combined with multiple iterations of reconstruction, and may result in less image noise from raw data, although longer reconstruction time is usually required [2 7]. To avoid the longer reconstruction time, Toshiba Medical Systems developed a new adaptive iterative dose reduction system using a 3D processing algorithm (AIDR 3D) for the improvement of image quality of reduced-dose CT images and reduction of reconstruction time resulting from the application of the iterative reconstruction algorithm [8 10]. With a low-dose scan, the number of x-ray photons becomes relatively small and electrical noise from the data acquisition system becomes dominant, which then degrades image quality. On the other hand, AIDR 3D processing uses a scanner model and a statistical noise model that takes both the number of photons and electrical noise into consideration to eliminate noise due to photon starvation for the projection data [8, 10]. In the reconstruction domain, AIDR 3D processing selectively extracts and eliminates noise iteratively from the reconstruction data. Finally, a weighted blending of the original reconstruction and the primary reconstruction data helps maintain the noise granularity. Thus, AIDR 3D improves the signal-to-noise ratio while preserving the spatial resolution and produces natural-looking images [9, 10]. However, to our knowledge, no comparative studies have been conducted of the capability of AIDR 3D for image noise reduction and its diagnostic performance using reduced-dose images and standard-dose CT for patients with chest diseases. We hypothesized that the AIDR 3D algorithm makes it possible to improve image noise and reduce radiation dose while maintaining diagnostic accuracy for low-dose chest CT examinations within the limitations of reconstruction times. The purpose of this study was to determine the utility of AIDR 3D for image noise reduction and assessment of radiologic findings obtained with reduced- and low-dose chest CT in patients with various pulmonary diseases. In addition, capabilities for image noise reduction and radiologic findings assessment on both CT protocols without and with AIDR 3D were directly and prospectively compare with those on standard-dose CT. Subjects and Methods Subjects This prospective study was approved by the institutional review board of Kobe University Hospital, and written informed consent was obtained from all patients. The authors had full control over the data for the entire duration of this study. Thirty-seven consecutive patients with pathologically diagnosed lung cancer and various pulmonary diseases underwent chest MDCT examination at three different tube currents (25, 50, and 150 mas). This study group of 37 patients (mean age, 66 years; age range, years) comprised 17 men (mean age, 65 years; age range, years) and 20 women (mean age, 66 years; age range, years) with 10 lung cancers, seven cases of asbestos-related pleural disease, five cases of organizing granulomas, four metastatic lung tumors, four thymomas, three granulomas, two cases of pulmonary tuberculosis, and one case each of cryptococcosis and giant bulla. Ten patients had pulmonary emphysema, and eight had interstitial lung disease due to connective tissue diseases, the latter consisting of four cases of progressive systemic sclerosis or scleroderma, three cases of rheumatoid arthritis, and one case of mixed connective tissue disorder. CT Protocol All plain chest CT examinations were performed with 16- or 64-MDCT scanners (Aquilion 16 or 64, Toshiba Medical Systems). A CT image of the whole chest was obtained with a single breath-hold. With three different tube current settings (150, 50, and 25 mas), three consecutive helical acquisitions were performed for each patient with the same length of helical run and FOV to obtain two volume datasets of the whole chest. Other scan parameters were the same for each CT protocol: peak tube voltage, 120 kv; gantry speed, 0.5 s/ rotation; slice collimation, 16 or mm; and beam pitch, 0.83 or We therefore had three raw data files of the same size for all 37 patients. Then the following datasets were reconstructed for each patient: standard-dose CT (150 mas) without AIDR 3D, reduced-dose CT (50 mas) without and with AIDR 3D, and low-dose CT (25 mas) without and with AIDR 3D at the lung and mediastinal window settings. All datasets were reconstructed for the lung window setting by using a high-frequency reconstruction algorithm (FC51, Toshiba Medical Systems) and for the mediastinal window setting by using a standard reconstruction algorithm (FC13, Toshiba Medical Systems). We then reconstructed all CT images without and with AIDR 3D by using a high-frequency reconstruction algorithm (FC51, Toshiba Medical Systems) for the lung window setting and a standard reconstruction algorithm (FC13, Toshiba Medical Systems) for the mediastinal window setting. This means that 10 datasets for each patient (i.e., standard-dose CT [150 mas] without AIDR 3D, reduced-dose CT [50 mas] without and with AIDR 3D, and low-dose CT [25 mas] without and with AIDR 3D at the lung and mediastinal window setting for each tube current) were reconstructed into CT images of 1-mm contiguous section thickness and that a total of 370 datasets was generated for this study. All CT images reconstructed without AIDR 3D were reconstructed by means of FBP. The difference in reconstruction time with and without AIDR 3D was about 1 second per 200 images. The reconstruction time of each CT set without AIDR 3D was seconds, and that with AIDR 3D was seconds. CT dose index was measured with a 32-cm acrylic dosimetry phantom and 100-mm ionization chamber, as described elsewhere [11]. After axial imaging of the phantom with a detector collimation of 16 or mm, reading of the ionizing chamber was recorded. The measurement was repeated 10 times for each of the five slots to diminish error and effect of the fan angle. The weighted CT dose index (CTDI) was calculated as one third of the CTDI at the center plus two thirds of the CTDI at the periphery. Image Analysis All MDCT images were randomized and independently reviewed by a radiologist with more than 10 years of experience in chest CT and a board-certified chest radiologist with 21 years of experience. All MDCT images were randomly interpreted by using a PACS (Shade Quest, Yokogawa Electronic). Both reviewers reviewed all MDCT images at lung (level, 550 HU; width, 1600 HU) and mediastinal (level, 35 HU; width, 250 HU) window settings without access to information about the technical parameters. All CT image evaluations by both radiologists were performed at different times, days, and reading rooms. Quantitative assessment of image noise Images at the level of the lung apices, aortic arch, carina, left atrium, and lung bases were used for W478

3 Pulmonary CT Using 3D Adaptive Iterative Dose Reduction A D quantitative analysis of image quality. First, the scanner table locations corresponding to the five levels were determined on a standard-dose CT series. Next, for assessment of the effect of direction of the x-ray path on the efficiency of the AIDR 3D, circular regions of interest (ROIs) 10 mm in diameter were placed on the lung window and mediastinal window settings at the tracheal lumen, bilateral lung parenchyma, aortic arch or descending aorta, or heart as the mediastinal structure, as well as at the bilateral soft-tissue structures within the chest wall that are adjacent to the peripheral part of the lungs, and the anterior and posterior aspects of the body. All ROIs were then copied to another CT image series, and image noise at each ROI was determined as the SD of the CT value within the ROI. Therefore, a total of 400 ROI (10 series 5 levels 8 ROIs) measurements were performed to determine image noise for each patient and were averaged to determine the final image noise value for each series by averaging the data from 40 ROIs (5 levels 8 ROIs) for each patient. Qualitative assessment of image quality To compare image noise reduction capability of each CT series with that of standard-dose CT, a 5-point visual scoring system for evaluation of image quality (1, nondiagnostic; 2, poor; 3, acceptable; 4, good; and 5, excellent) was adopted. Image quality was thus evaluated at the level of the lung apices, aortic arch, carina, left atrium, and lung bases on all CT images. To assess the adequacy of a series of images as a whole, acceptability of a set of images was defined as a consistent score of 3 (acceptable) or more for all five levels. To compare the capability of each CT protocol with that of standard-dose CT, a 5-point visual scoring system (1, absent; 2, probably absent; 3, equivocal; 4, probably present; and 5, present) was used to determine the presence of emphysema, ground-glass opacity, reticular opacity, interlobular septal thickening, bronchiectasis, honeycomb pattern, and nodules at the lung window setting and the presence of aneurysm, calcification at coronary artery, pericardial or pleural effusion, pleural thickening, pleural calcification, and lymphadenopathy at the mediastinal window setting. Statistical Analysis To determine the utility of AIDR 3D for reduced- and low-dose CT images at the lung and mediastinal window settings, image noises of all series at both window settings were compared by B E using the Fisher protected least significant difference test for quantitative assessment. Weighted kappa analyses were used to determine the interobserver agreement for image quality evaluation and each radiologic finding assessment on standarddose CT, reduced-dose CT without and with AIDR 3D, and low-dose CT without and with AIDR 3D at lung and mediastinal window settings. Interobserver agreement was considered as slight (κ < 0.21), fair (κ = ), moderate (κ = ), substantial (κ = ), and almost perfect (κ = ) [12]. To determine the utility of AIDR 3D for reducedand low-dose CT images at lung and mediastinal window settings, the image qualities of all series at both window settings were compared by using the Fisher protected least significant difference test for qualitative assessment. Weighted kappa statistical analyses were also performed to determine the appropriateness for routine clinical practice of reduced- and low-dose CT with AIDR 3D compared with that of standard-dose CT as well as agreement among radiologic findings obtained with standarddose CT and the other systems. Intermethod agreement was considered as slight (κ < 0.21), fair (κ = ), moderate (κ = ), substantial Fig year-old woman with pulmonary emphysema in right upper lobe; all CT images show low-attenuation area due to pulmonary emphysema. A, Standard-dose CT. B, Reduced-dose CT without adaptive iterative dose reduction using 3D processing (AIDR 3D). C, Reduced-dose CT with AIDR 3D. D, Low-dose CT without AIDR 3D. E, Low-dose CT with AIDR 3D. C W479

4 Ohno et al. (κ = ), and almost perfect (κ = ) [12]. A p value of less than 0.05 was considered significant for statistical analyses. Results Representative CT images for standard-, reduced-, and low-dose CT are shown in Figures 1 3. Results of quantitative image quality analyses are shown in Table 1. On the lung and mediastinal window settings, the image quality of both reduced- and low-dose CT without AIDR 3D was significantly lower than that of standard-dose CT and that of reduced- and lowdose CT with AIDR 3D (p < 0.05). In addition, low-dose CT without AIDR 3D produced significantly poorer image quality than did reduced-dose CT without AIDR 3D (p < 0.05). A D Results of kappa analyses of interobserver agreement for qualitative image quality and radiologic findings assessment at lung and mediastinal window settings are shown in Tables 2 and 3. The findings indicate that all interobserver agreements were substantial or almost perfect (κ 0.72). Results of a comparison of qualitative assessments of image quality attained by standard-dose CT and others are shown in Table 4. At lung and mediastinal window settings, the image quality of reduced- and low-dose CT without AIDR 3D was significantly lower than that of standarddose CT and that of reduced- and low-dose CT with AIDR 3D (p < 0.05). In addition, low-dose CT without AIDR 3D produced significantly poorer image quality than did reduced-dose CT without AIDR 3D (p < 0.05). B E Fig year-old man with ground-glass opacity in left upper lobe, seen in all CT images. A, Standard-dose CT. B, Reduced-dose CT without adaptive iterative dose reduction using 3D processing (AIDR 3D). C, Reduced-dose CT with AIDR 3D. D, Low-dose CT without AIDR 3D. E, Low-dose CT with AIDR 3D. Results of kappa analyses for agreement between standard-dose CT and others on lung and mediastinal window settings are shown in Tables 5 and 6. All agreements for emphysema, honeycomb pattern, and nodules were almost perfect (κ > 0.81). In addition, all agreements for ground-glass opacity and bronchiectasis were substantial or almost perfect (κ 0.81). Moreover, all agreements for reticular opacity and interlobar septal thickening were moderate, substantial, or almost perfect (κ 0.72), respectively. On the other hand, when assessing agreements between standard-dose CT and others on the mediastinal window setting, all agreements for mediastinal findings were almost perfect (κ > 0.81). Discussion Our results suggest that AIDR 3D is useful for reduced- and low-dose chest CT examinations at 50 and 25 mas without significant degradation of image quality and radiologic findings assessment compared with those obtained with standard-dose CT protocols. The tube currents used in this study were equal to or lower than the tube currents for chest CT examinations used in previously reported studies, which tried to reduce radiation dose C W480

5 Pulmonary CT Using 3D Adaptive Iterative Dose Reduction A D by using other techniques, such as 3D adaptive raw data filter [11], adaptive statistical iterative reconstruction [2 5], and iterative reconstruction in image space [6, 7]. In addition, our results suggest that AIDR 3D makes it possible to reduce tube current equal to 25 mas as a low-dose protocol for chest CT examination without causing significant disagreement in radiologic finding assessments, except for the detection of interstitial septal thickening. To the best of our knowledge, no reports have been published on this utility of AIDR 3D for reduced- and low-dose CT examinations of the chest. As for quantitative and qualitative assessment of image quality and noise, the quality of reduced- and low-dose CT with AIDR 3D at lung and mediastinal window settings showed no significant differences with those of standard-dose CT. The mainstay of radiation dose reduction methods has been reduction of tube current, which is the simplest such method. On the other hand, image noise is inversely proportional to the square root of the radiation dose. In addition, with lower tube current, the image noise becomes significantly worse while streak artifacts also increase, so that the electrical noise becomes significant. Moreover, image noise on each CT protocol at the lung window setting was higher than that at the mediastinal window setting. This difference was related to the difference in reconstruction kernels for lung and mediastinal window images on standard-, reduced-, and low-dose CT. Because AIDR 3D successfully removes the effect of electrical noise from the projection data, the image noise with lower tube current becomes relatively stable. Therefore, our results suggest that AIDR 3D is useful for radiation dose reduction of chest CT examinations in conjunction with significant improvements B E Fig year-old woman with lung collagen in left upper lobe; all CT images show interstitial septal thickening and reticular opacity. A, Standard-dose CT. B, Reduced-dose CT without adaptive iterative dose reduction using 3D processing (AIDR 3D). C, Reduced-dose CT with AIDR 3D. D, Low-dose CT without AIDR 3D. E, Low-dose CT with AIDR 3D. in image quality and noise in routine clinical practice. As for kappa statistics for qualitative image quality and radiologic finding assessment, all interobserver agreements were substantial or almost perfect. This suggests that our findings were reproducible between the two observers and that assessments of image quality and radiologic findings were reliable for all CT protocols [12]. In terms of agreement for assessment of radiologic findings at lung and mediastinal window settings obtained with standard-dose CT and the other protocols, assessments of the detection of emphysema, ground-glass opacity, reticular opacity, bronchiectasis, honeycomb pattern, and nodules at the lung window setting and all findings at the mediastinal window setting obtained with reduced- and low-dose CT with AIDR 3D almost perfectly matched those obtained with standard-dose CT. In addition, agreements for the detection of interstitial septal thickening on reduced- and lowdose CT with AIDR 3D were almost perfect or substantial. These results suggest that the use of AIDR 3D for reduced- and low-dose CT can improve radiologic finding agreements obtained with standard-dose CT. Previous studies in which MDCT was used for plain chest CT examinations for humans have suggested that tube current can be C W481

6 Ohno et al. TABLE 1: Quantitative Image Quality Assessments of All CT Series at Lung and Mediastinal Window Settings Window Setting, CT Protocol Image Noise (HU) Lung Standard-dose CT 31.4 ± 12.6 Reduced-dose CT without AIDR 3D 48.4 ± 17.8 a,b,c Reduced-dose CT with AIDR 3D 30.1 ± 10.1 Low-dose CT without AIDR 3D 66.7 ± 26.9 a,b,c,d Low-dose CT with AIDR 3D 32.1 ± 7.5 Mediastinal Standard-dose CT 9.8 ± 12.6 Reduced-dose CT without AIDR 3D 12.7 ± 10.8 a,b,c Reduced-dose CT with AIDR 3D 9.0 ± 10.9 Low-dose CT without AIDR 3D 15.7 ± 7.4 a,b,c,d Low-dose CT with AIDR 3D 9.6 ± 6.2 Note Data are mean ± SD. AIDR 3D = adaptive iterative dose reduction using 3D processing. a p < 0.05, versus standard-dose CT. b p < 0.05, versus reduced-dose CT with AIDR 3D. c p < 0.05, versus low-dose CT with AIDR 3D. d p < 0.05, versus reduced-dose CT without AIDR 3D. reduced to 30 mas or higher for pulmonary nodule, emphysema, and asbestosis assessments and follow-up CT examinations [13 22]. Increased image noise and streak artifacts were found to diminish image quality and the depiction of radiologic findings on reduced- and low-dose CT. Moreover, agreement among standard-, reduced-, and lowdose CT for diffuse lung diseases was only assessed by using cadaveric lung inflated and fixed with the method of Heitzman [23], whereas no study, to our knowledge, has directly compared the results among standard-, reduced-, and low-dose CT for humans. Although doses lower than those used in the present study have been suggested as acceptable without and with adaptive statistical iterative reconstruction, those studies could not assess artifacts from chest wall, fat, vertebrae, ribs, and mediastinal structures such as trachea, great vessels, and cardiac motions [5]. The lower tube current used in the cadaveric lung study could, therefore, not be directly adopted for routine clinical study, so that, to our knowledge, our study constitutes the first and direct comparison of reduced- or low-dose CT with standard-dose CT for patients with various pulmonary diseases, including interstitial lung diseases. According to our findings, we can reduce tube current from 150 to 25 mas without significantly diminishing the image quality and capabilities of radiologic findings, except for detection of reticular opacity and interstitial septal thickening, in routine clinical practice. In addition, if we hope to assess all radiologic findings without any reduction in image quality or radiologic findings capability, we can reduce tube current to 50 mas with the adaptation of AIDR for chest CT examination in routine clinical practice. In addition, AIDR 3D involves almost no reconstruction time penalty compared with protocols using the standard field back projection algorithm, so that the use of AIDR 3D can be consid- TABLE 2: Interobserver Agreement for Image Quality and Radiologic Findings for All CT Series at Lung Window Setting Method Image Quality Score Emphysema Ground-Glass Opacity Reticular Opacity Interlobular Septal Thickening Bronchiectasis Honeycomb Pattern Nodules Standard-dose CT Reduced-dose CT without AIDR 3D Reduced-dose CT with AIDR 3D Low-dose CT without AIDR 3D Low-dose CT with AIDR 3D Note Data are kappa values. AIDR 3D = adaptive iterative dose reduction using 3D processing. TABLE 3: Interobserver Agreement for Image Quality and Radiologic Findings for All CT Series at Mediastinal Window Setting Method Image Quality Score Aortic Aneurysm Calcification at Coronary Artery Pericardial or Pleural Effusion Pleural Thickening Pleural Calcification Tumor Lymphadenopathy Standard-dose CT Reduced-dose CT without AIDR 3D Reduced-dose CT with AIDR 3D Low-dose CT without AIDR 3D Low-dose CT with AIDR 3D Note Data are kappa values. AIDR 3D = adaptive iterative dose reduction using 3D processing. W482

7 Pulmonary CT Using 3D Adaptive Iterative Dose Reduction TABLE 4: Qualitative Image Quality Assessments of All CT Series at Lung and Mediastinal Window Settings Window Setting, CT Protocol Image Quality Score Lung Standard-dose CT 4.9 ± 0.3 Reduced-dose CT without AIDR 3D 4.3 ± 0.5 a,b,c Reduced-dose CT with AIDR 3D 4.8 ± 0.4 Low-dose CT without AIDR 3D 3.4 ± 0.7 a,b,c,d Low-dose CT with AIDR 3D 4.7 ± 0.5 Mediastinal Standard-dose CT 4.9 ± 0.3 Reduced-dose CT without AIDR 3D 4.1 ± 0.3 a,b,c Reduced-dose CT with AIDR 3D 4.8 ± 0.3 Low-dose CT without AIDR 3D 3.1 ± 0.7 a,b,c,d Low-dose CT with AIDR 3D 4.7 ± 0.6 Note Data are mean ± SD. AIDR 3D = adaptive iterative dose reduction using 3D processing. a p < 0.05, versus standard-dose CT. b p < 0.05, versus with reduced-dose CT with AIDR 3D. c p < 0.05, versus low-dose CT with AIDR 3D. d p < 0.05, versus reduced-dose CT without AIDR 3D. ered acceptable for routine clinical practice. We, therefore, think that AIDR is a useful algorithm for chest CT from the point of view of radiation dose and clinical practice issues. There are several limitations to this study. First, we directly and prospectively tried to evaluate the utility of AIDR 3D for radiation dose reduction for chest CT examinations in this study and compared image qualities and agreements of radiologic findings among low- and reduced-dose CT without and with AIDR 3D and standard-dose CT without AIDR 3D, which is used in routine clinical practice. However, AIDR 3D can reduce image noise and may also improve image quality of standard-dose CT protocol. Therefore, it is necessary for us to show the clinical relevance of AIDR 3D for improving the image quality of currently used standard-dose CT protocols. In addition, we also have to assess the potential of AIDR 3D for standard-dose chest CT as well as low-dose CT examinations by direct comparison of diagnostic performance between standard-dose CT without and with AIDR 3D. Second, it has been suggested in previous publications that AEC is useful for radiation dose reduction [24 31]. Although AEC has been used in routine clinical practice, we did not use this technique for radiation dose reduction. The use of AEC in CT scanners essentially provides programmed dynamic adjustment of the tube current, which is adjusted to achieve consistent image quality among patients and for a single patient. When reduced or low tube currents are used, AEC can reduce the radiation dose without further image degradation on chest CT. Therefore, further radiation dose reduction can be achieved by using AEC for reducedor low-dose CT examinations, so that a study using a combination of with AEC and AIDR is clearly warranted. Third, we assessed image quality and agreement regarding radiologic findings among standard-, reduced-, and low-dose CT. However, we did not assess diagnostic performance in this study. The results of previous studies [32, 33] of image quality and diagnostic quality showed that degradation of image quality can occur at tube current settings higher than those compromising the capability of depicting abnormalities. In addition, all thin-section CT images on reduced- and low-dose CT without and with AIDR 3D and standard-dose CT without AIDR 3D protocols were assessed in this TABLE 5: Agreement for Assessments of Radiologic Findings Between Standard-Dose CT and Low- and Reduced-Dose CT at Lung Window Setting CT Protocol Emphysema Ground-Glass Opacity Reticular Opacity Interlobular Septal Thickening Bronchiectasis Honeycomb Pattern Nodules Reduced-dose CT without AIDR 3D Reduced-dose CT with AIDR 3D Low-dose CT without AIDR 3D Low-dose CT with AIDR 3D Note Data are kappa values. AIDR 3D = adaptive iterative dose reduction using 3D processing. TABLE 6: Agreement for Assessments of Radiologic Finding Between Standard-Dose CT and Low- and Reduced-Dose CT at Mediastinal Window Setting CT Protocol Aortic Aneurysm Calcification at Coronary Artery Pericardial or Pleural Effusion Pleural Thickening Pleural Calcification Tumor Lymphadenopathy Reduced-dose CT without AIDR 3D Reduced-dose CT with AIDR 3D Low-dose CT without AIDR 3D Low-dose CT with AIDR 3D Note Data are kappa values. AIDR 3D = adaptive iterative dose reduction using 3D processing. W483

8 Ohno et al. study. However, CT images reconstructed as more than or equal to 5-mm section thickness is used in routine clinical practice. Reconstruction of CT images as thicker sections can reduce image noise and may reduce the influence of AIDR 3D on reduced- and lowdose CT images. Therefore, it is necessary to show that there is no degradation of diagnostic performance of reduced- and low-dose CT compared with that of standard-dose CT as thin-section and routinely adapted section thickness before this technique can be adopted for routine clinical practice. Fourth, a radiologist with more than 10 years of experience in chest CT and a boardcertified chest radiologist with 21 years of experience quantitatively and qualitatively assessed image noise reduction and radiologic findings assessment to determine the capability of AIDR 3D for radiation dose reduction on chest CT examinations. However, other radiologists who had other experiences as chest radiologists or other specialties were not included in this study. Therefore, our qualitatively assessed image qualities and agreements of radiologic findings on all CT protocols and statistical results might have some biases. In addition, both readers were educated and trained at the same institution and country. Therefore, a study involving multiple readers from multiple centers would be warranted to show the real significance of the utility of AIDR 3D for reduced- and lowdose CT protocols in routine clinical practice. Finally, our study population was not large enough to perform a comparison of image quality and radiologic findings for patients classified by body mass index. Body mass index has been shown to influence artifacts on chest CT [34 36]. Therefore, a large prospective cohort study as well as a diagnostic performance study appears advisable for further evaluation of low- and reduceddose CT with AIDR. In conclusion, AIDR 3D was found to be useful for assessment of image noise reduction and radiologic findings on reduced- and low-dose CT for patients with various pulmonary diseases. Acknowledgments We thank Naho Kishitani, Hiroyasu Inokawa, and Yasuko Fujisawa (Toshiba Medical Systems) for their technical support for this study. We also thank Yoshiyuki Ohno (Nagoya University, Department of Preventive Medicine, Graduate School of Medicine) for his advice for the statistical component of this study. References 1. Kubo T, Lin PJ, Stiller W, et al. Radiation dose reduction in chest CT: a review. AJR 2008; 190: Silva AC, Lawder HJ, Hara A, Kujak J, Pavlicek W. Innovations in CT dose reduction strategy: application of the adaptive statistical iterative reconstruction algorithm. AJR 2010; 194: Prakash P, Kalra MK, Digumarthy SR, et al. Radiation dose reduction with chest computed tomography using adaptive statistical iterative reconstruction technique: initial experience. J Comput Assist Tomogr 2010; 34: Prakash P, Kalra MK, Ackman JB, et al. Diffuse lung disease: CT of the chest with adaptive statistical iterative reconstruction technique. Radiology 2010; 256: Yanagawa M, Honda O, Yoshida S, et al. Adaptive statistical iterative reconstruction technique for pulmonary CT: image quality of the cadaveric lung on standard- and reduced-dose CT. Acad Radiol 2010; 17: Pontana F, Pagniez J, Flohr T, et al. Chest computed tomography using iterative reconstruction vs filtered back projection. Part 1. Evaluation of image noise reduction in 32 patients. Eur Radiol 2011; 21: Pontana F, Duhamel A, Pagniez J, et al. Chest computed tomography using iterative reconstruction vs filtered back projection. Part 2. Image quality of low-dose CT examinations in 80 patients. Eur Radiol 2011; 21: Yang Z, Zamyatin AA, Akino N. Effective datadomain noise and streak reduction for x-ray CT. In: Kachelriess M, Rafecas M, eds. 11th International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, Fully 3D Website. org/2011/fully3d2011proceedings.pdf. Accessed June 4, Yang Z, Silver MD, Noshi Y. Adaptive weighted anisotropic diffusion for computed tomography denoising. In: Kachelriess M, Rafecas M, eds. 11th International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, Fully 3D Website. pdf. Accessed June 4, Gervaise A, Osemont B, Lecocq S, et al. CT image quality improvement using adaptive iterative dose reduction with wide-volume acquisition on 320-detector CT. Eur Radiol 2012; 22: Kubo T, Ohno Y, Gautam S, Lin PJ, Kauczor HU, Hatabu H; ilead Study Group. Use of 3D adaptive raw-data filter in CT of the lung: effect on radiation dose reduction. AJR 2008; 191:1071; [web]w167 W Svanholm H, Starklint H, Gundersen HJ, Fabricius J, Barlebo H, Olsen S. Reproducibility of histomorphologic diagnoses with special reference to the kappa statistic. APMIS 1989; 97: Ravenel JG, Scalzetti EM, Huda W, Garrisi W. Radiation exposure and image quality in chest CT examinations. AJR 2001; 177: Prasad SR, Wittram C, Shepard JA, McLoud T, Rhea J. Standard-dose and 50%-reduced-dose chest CT: comparing the effect on image quality. AJR 2002; 179: Zhu X, Yu J, Huang Z. Low-dose chest CT: optimizing radiation protection for patients. AJR 2004; 183: Karabulut N, Toru M, Gelebek V, Gulsun M, Ariyurek OM. Comparison of low-dose and standard dose helical CT in the evaluation of pulmonary nodules. Eur Radiol 2002; 12: Weng MJ, Wu MT, Pan HB, Kan YY, Yang CF. The feasibility of low-dose CT for pulmonary metastasis in patients with primary gynecologic malignancy. Clin Imaging 2004; 28: Dinkel HP, Sonnenschein M, Hoppe H, Vock P. Low-dose multislice CT of the thorax in follow-up of malignant lymphoma and extrapulmonary primary tumors. Eur Radiol 2003; 13: Yamada T, Ono S, Tsuboi M, et al. Low-dose CT of the thorax in cancer follow-up. Eur J Radiol 2004; 51: Chiu CH, Chern MS, Wu MH, et al. Usefulness of low-dose spiral CT of the chest in regular follow up of postoperative non-small cell lung cancer patients: preliminary report. J Thorac Cardiovasc Surg 2003; 125: Remy-Jardin M, Sobaszek A, Duhamel A, Mastora I, Zanetti C, Remy J. Asbestos-related pleuropulmonary diseases: evaluation with low-dose four-detector row spiral CT. Radiology 2004; 233: Zaporozhan J, Ley S, Weinheimer O, et al. Multidetector CT of the chest: influence of dose onto quantitative evaluation of severe emphysema a simulation study. J Comput Assist Tomogr 2006; 30: Markarian B, Dailey ET. Preparation of inflated lung specimens. In: Groskin SA, ed. Heitzman s the lung: radiologic-pathologic correlation, 3rd ed. St. Louis, MO: Mosby, 1993: Jung KJ, Lee KS, Kim SY, Kim TS, Pyeun YS, Lee JY. Low-dose, volumetric helical CT: image quality, radiation dose, and usefulness for evaluation of bronchiectasis. Invest Radiol 2000; 35: Yi CA, Lee KS, Kim TS, Han D, Sung YM, Kim S. Multidetector CT of bronchiectasis: effect of radiation dose on image quality. AJR 2003; 181: Ley-Zaporozhan J, Ley S, Krummenauer F, Ohno Y, Hatabu H, Kauczor HU. Low dose multi-detector CT of the chest (ilead Study): visual ranking of different simulated mas levels. Eur J Radiol 2010; 73: W484

9 Pulmonary CT Using 3D Adaptive Iterative Dose Reduction 27. Papadakis AE, Perisinakis K, Damilakis J. Automatic matic exposure control for 64-detector row CT: Low-dose spiral computed tomography of the tho- exposure control in pediatric and adult multidetector radiation dose reduction in chest phantom study. rax: comparison with the standard-dose tech- CT examinations: a phantom study on dose reduction Eur J Radiol 2011; 77: nique. Invest Radiol 1998; 33:68 73 and image quality. Med Phys 2008; 35: Koyama H, Ohno Y, Yamazaki Y, et al. Reduced- 34. d Agostino AG, Remy-Jardin M, Khalil C, et al. 28. Söderberg M, Gunnarsson M. Automatic expo- dose chest CT with 3D automatic exposure control Low-dose ECG-gated 64-slices helical CT angi- sure control in computed tomography: an evalua- vs. standard chest CT: quantitative assessment of ography of the chest: evaluation of image quality tion of systems from different manufacturers. Acta Radiol 2010; 51: Israel GM, Cicchiello L, Brink J, Huda W. Patient size and radiation exposure in thoracic, pelvic, and abdominal CT examinations performed with automatic exposure control. AJR 2010; 195: Matsumoto K, Ohno Y, Koyama H, et al. 3D auto- emphysematous changes in smokers lung parenchyma. Eur J Radiol 2012; 81: 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 1995; 164: Takahashi M, Maguire WM, Ashtari M, et al. in 105 patients. Eur Radiol 2006; 16: Sun Z. Multislice CT angiography in aortic stent grafting: relationship between image noise and body mass index. Eur J Radiol 2007; 61: Taylor CM, Blum A, Abbara S. Patient preparation and scanning techniques. Radiol Clin North Am 2010; 48: W485