| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Department of Radiology, Hôpital Erasme (A.M., P.A.G.), and Statistical Unit, Institut de Recherche Interdiscisplinaire en Biologie Humaine et Moléculaire (V.D.M.), Université Libre de Bruxelles, Brussels, Belgium; and Department of Histology, Université de Mons-Hainaut, Mons, Belgium (J.Z.). Received February 19, 2006; revision requested April 19; revision received May 21; accepted June 6; final version accepted July 25. Supported by the Erasme Funds for Medical Research. Address correspondence to A.M., Department of Radiology, CHU de Charleroi, Boulevard Paul Janson 92, 6000 Charleroi, Belgium (e-mail: Afarine{at}ladner-madani.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: The institutional review board approved this protocol. Written informed consent was obtained from all patients. Seventy patients (49 men, 21 women; age range, 38–79 years) referred for surgical resection of a lung tumor underwent multidetector CT with 4 x 1-mm collimation, 120 kVp, and 20 and 120 effective mAs. At each radiation dose, 1.25-, 5.0-, and 10.0-mm-thick sections were reconstructed at 10-mm intervals. From scans of the lobe or whole lung to be resected, relative areas (RAs) of lung with attenuation coefficients lower than nine thresholds and eight percentiles of the distribution of attenuation coefficients were compared with the histopathologic extent of emphysema, which was measured microscopically—by using the corrected mean interwall distance (MIWD) and the corrected mean perimeter (MP)—and macroscopically. Correlations between the data obtained by using attenuation thresholds and percentiles and the parameters macroscopic extent of emphysema, MIWD, and MP were investigated by using Spearman coefficients.
Results: The 1st percentile (r range, –0.394 to –0.675; P < .001) and attenuation coefficients of –980, –970, and –960 HU (r range, 0.478–0.664; P < .001) yielded the strongest correlations with macroscopic extent, MIWD, and MP, regardless of radiation dose or section thickness. The effects of radiation dose and section thickness on RAs of lung with attenuation coefficients lower than –960 HU (P = .007 and P < .001, respectively) and lower than –970 HU (P = .001 and P < .001, respectively) were significant. The effect of section thickness on the 1st percentile was significant (P < .001), whereas the effect of dose was not (P = .910).
Conclusion: At CT quantification of pulmonary emphysema, the tube current–time product can be reduced to 20 mAs, but both tube current–time product and section thickness should be kept constant in follow-up examinations.
© RSNA, 2007
Pulmonary emphysema is a chronic obstructive pulmonary disease and worldwide the sixth and 12th most common cause of mortality and morbidity, respectively (1). Computed tomography (CT) yields densitometric measurements that are highly reproducible and that have been correlated with morphometric measurements of alveolar tissue. As a result, this technique has been recommended for follow-up examinations, particularly those to evaluate therapeutic interventions (2–5). Although multi–detector row CT is of interest for the quantification of heterogeneously distributed lung disorders such as pulmonary emphysema, the use of this technique increases the radiation dose by an additional 300% per examination compared with the dose required for incremental single–detector row CT, which yields thin sections in 1-cm intervals (6). This increased radiation dose is of special concern in patients with emphysema, because these individuals may be young and may have a favorable prognosis. The relatively high level of radiation that patients with emphysema are exposed to at multidetector CT is compounded by the repeated follow-up examinations that they undergo throughout their life. Although reduced radiation doses have been successfully used in patients with various thoracic disorders (7–12), the effect of dose on densitometric measurements is not known.
Multi–detector row CT yields a large number of images—generally, 350 images with contiguous 1-mm-thick sections throughout the thorax. Quantifying the pulmonary emphysema on such an amount of images is very time consuming. By increasing the section thickness, one could reduce this number and still appreciate the lung parenchyma in its entirety on contiguous thicker sections. Nevertheless, to our knowledge, the effect of section thickness on densitometric measurements is also not known. Thus, the purpose of our study was to prospectively investigate the effects of radiation dose and section thickness on quantitative multidetector CT indexes of pulmonary emphysema.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Twenty-eight patients were not included in our research protocol because they had pneumonia (n = 4), pulmonary atelectasis (n = 4), or interstitial lung disease (n = 3) at admission or because they were referred for tumorectomy (resection of tumor only rather than of entire lobe) (n = 17). Two additional patients were excluded after undergoing multi–detector row CT owing to breathing artifacts (n = 1) or evidence of interstitial lung disease in the nontumoral parenchyma (n = 1). Five patients were excluded after undergoing thoracic surgery because they underwent segmentectomy, a procedure that would prevent us from appropriately fixing the lung specimen in preparation for morphometric measurements.
A total of 70 patients (49 men, 21 women) aged 38–79 years (mean, 63 years ± 1 [standard error of the mean]) were included in our current study. The mean time from CT to surgery was 3 days (range, 1–7 days). Sixty-four patients underwent lobe resection, and six underwent whole-lung resection. Thirteen patients were nonsmokers, 14 were ex-smokers, and 43 were current smokers. The current smokers and ex-smokers had smoked a mean of 38 pack-years ± 2 (range, 10–100 pack-years). The mean body mass index, calculated from data in the medical chart (14), was 25.1 kg/m2 ± 0.5 (range, 15.6–37.6 kg/m2).
CT Examinations
CT scans were obtained by using a commercially available four-channel multi–detector row scanner (Somatom Volume Zoom; Siemens Medical Solutions, Forchheim, Germany). The scale of attenuation coefficients in this CT scanner ranges from –1024 to 3072 HU. The scanner was calibrated periodically and after major maintenance work. Patients were scanned craniocaudally during a full-inspiration breath hold. No spirometric controlling of lung volume was performed because this procedure has not been shown to substantially improve the repeatability of quantitative CT scanning in the assessment of the extent of emphysema (3). No patient received intravenous contrast material. CT was performed twice with 4 x 1-mm collimation, 120 kVp (constant), and 20 and 120 effective mAs. No tube current modulation was applied during the acquisitions. Effective milliampere-second, as defined by Mahesh et al (15), refers to the milliampere-second value divided by the pitch, where pitch is defined, according to Silverman et al (16), as the ratio between the table feed per rotation and the x-ray beam width. The table feed was 6 mm per 0.75-second rotation (8 mm/sec). These parameters resulted in a pitch of 1.5:1. The total absorbed doses—expressed as weighted CT dose indexes—were 2.66 and 15.96 mGy, respectively, at 20 and 120 mAs. The weighted CT dose index was calculated as the sum of two-thirds the peripheral dose and one-third the central dose measured in a plastic phantom (17). From the raw data derived with each radiation dose, 1.25-, 5.0-, and 10.0-mm-thick sections were reconstructed with a soft-tissue algorithm (20S; Siemens Medical Solutions) at 10-mm intervals. We used this interval because it has been widely used in CT studies of diffusely distributed lung disease.
By using the Pulmo CT program (Siemens Medical Solutions), which automatically recognizes the lungs and traces the lung contours, the attenuation coefficients of each lung CT section were measured by a radiologist (A.M.) with 11 years experience in chest CT. The interlobar limits were traced manually in the cases of patients who underwent lobe resection. From these coefficients, we calculated for each lobe or whole lung to be resected relative areas of lung (expressed as percentages) with attenuation coefficients lower than thresholds ranging from –900 to –980 HU (–900, –910, –920, –930, –940, –950, –960, –970, and –980 HU) and eight percentiles (1st, 3rd, 5th, 7th, 10th, 12th, 15th, and 18th percentiles) of the distribution of attenuation coefficients, at each radiation dose and each section thickness.
Macroscopic Quantification of Emphysema
The resected lung samples were prepared for macroscopic quantification as extensively described elsewhere (13): The resected lobes and whole lungs were inflated and were immersed in a formalin solution for 48 hours immediately after surgery. By using a modified Gough-Wentworth technique, two 1-mm-thick vertical whole-lung sections were obtained from the fixed specimens (18). The sections were cut by a pathologist who was blinded to the quantitative CT results about the remaining nontumoral lung parenchyma and were mounted (A.M.) between two transparency films for plain paper copiers (3M, St Paul, Minn), digitized (Hewlett Packard Scan JET 6100C/T; Hewlett Packard, Palo Alto, Calif), and stored on a personal computer by a physicist (J.Z.) with more than 10 years of experience regarding lung pathology measurements and who was blinded to the quantitative CT results. The area macroscopically shown to be occupied by emphysema was measured on these sections by using a previously validated computer-assisted method (19). The relative area (expressed as a percentage) with a low densitometric reading corresponded to the pathologic extent of emphysema seen macroscopically.
Microscopic Quantification of Emphysema
The resected lung samples were prepared for microscopic quantification as extensively described elsewhere (13): For each patient, the material to be used for microscopic quantification consisted of 18 samples from the nontumoral lung parenchyma, which the pathologist obtained from each of nine regions in the lung specimen that were selected by dividing the specimen along the cephalocaudal direction in the upper, middle, and lower zones and along the horizontal direction in the centroperipheral, middle, and peripheral zones. Lung sections were observed at x25 magnification by using a microscope connected to a high-resolution video camera. From each lung section, two microscopic fields were digitized (J.Z.), and after the image threshold was established, the total perimeter of the alveoli and alveolar ducts in the field was measured with an image analyzer system (model KS400; Carl Zeiss Vision, Hallbergmaas, Germany).
In a second step, the distances between intersections of the alveolar and duct walls and seven horizontal lines arbitrarily drawn by the computer were automatically measured with the image analyzer. These interwall distance values reflected the diameters of the alveoli and alveolar ducts combined. Perimeters and interwall distances were measured in 36 fields, 35 of which having been recommended by Gould et al (20). From the measurements obtained from one lung specimen, the mean perimeter per field and the mean interwall distance were calculated. These values, originally expressed in pixels, were converted to micrometers: With the optical system used, 1 pixel corresponded to 3.205 µm.
The mean perimeter per field and the mean interwall distance calculated from the histologic slides were corrected for shrinkage and distortion (mainly due to dehydration) caused by processing and cutting the tissue (21).
Statistical Analyses
Quantitative variables were expressed as means ± standard errors of the mean. Relative areas of lung with attenuation coefficients lower than each considered threshold and all eight percentiles were compared with the macroscopic and microscopic quantitative indexes.
For each radiation dose and each section thickness, we calculated Spearman correlation coefficients for the relationships between each set of CT data obtained with the nine tested attenuation thresholds and the eight percentiles and the corrected mean interwall distance (MIWD) and between each set of CT data and the corrected mean perimeter (MP). The higher the coefficient (for a given set of data), the more appropriate was the threshold for quantifying emphysema.
Three analyses of variance for repeated measurements were performed to investigate the most appropriate threshold and percentile as functions of (a) the potential effect of radiation dose—that is, the tube current–time product (factor at two levels), (b) the potential effect of section thickness (factor at three levels), and (c) the potential effect of the interaction of these two factors. For all tests, P < .05 indicated statistical significance. The statistical software used was SPSS for Windows, release 13.0 (SPSS, Chicago, Ill).
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Attenuation Coefficient Thresholds
The highest correlation coefficients for the relationship between macroscopic extent of emphysema and relative area of lung with an attenuation coefficient lower than a given tested threshold (n = 9), derived at each radiation dose and each section thickness, were obtained for an attenuation coefficient of –980 HU, regardless of radiation dose or section thickness (Table 1).
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Effect of Radiation Dose
Radiation dose does not substantially influence the strength of correlations between histopathologic indexes and either relative areas or percentiles. This finding suggests that reducing the dose to 20 mAs should be recommended for the CT quantification of pulmonary emphysema, especially in patients who will undergo repeated follow-up examinations. However, because radiation dose does affect RA960 and RA970, it should be kept constant in comparative and follow-up examinations based on relative area. Our study results complement those of Stolk et al (22), who found that measurements of both the relative area of lung with an attenuation coefficient lower than –910 HU and the 15th percentile on 2.5-mm-thick CT sections obtained at 20 mAs are as repeatable as pulmonary function tests in individuals who have 
1-antitrypsin deficiency–associated emphysema. However, our study revealed, on the basis of comparisons with macroscopic and microscopic indexes, that RA960, RA970, and the 1st percentile are more appropriate for emphysema quantification than are relative area of lung with an attenuation coefficient lower than –910 HU and the 15th percentile.
Other studies of the objective CT quantification of pulmonary emphysema have been based on tube current–time products ranging from 88 to 255 mAs (5,17,18,23–27). Results of numerous studies in which the effect of dose reduction on chest CT examinations was investigated by means of subjective assessment of image quality or diagnostic performance have suggested that the tube current–time product could be dramatically reduced for certain applications (28,29). These applications include pulmonary angiography (12) and the detection of lung nodules (7), bronchiectasis (8), ground-glass opacity (9), chronic infiltrative lung disease (10), and pleural and pulmonary asbestos-related abnormalities (11). Our study results show that radiation dose reduction could be extended to the objective CT quantification of pulmonary emphysema.
Effect of Section Thickness
Section thickness does not substantially influence the strength of correlations between histopathologic indexes and either relative areas or percentiles. This finding suggests that increased section thickness has no advantage over thin sections obtained in 1-cm intervals. Nevertheless, section thickness does affect RA960, RA970, and the 1st percentile. There is an average difference of 2% among RA960 and RA970 values measured on CT sections of different thicknesses ranging from 1.25 to 10.0 mm. For the 1st percentile, this represents a difference of 25 HU, indicating that the attenuation coefficient corresponding to the 1st percentile increases slightly when the section thickness increases—for example, from –968 HU at 1.25 mm to –943 HU at 10.0 mm in our current study group. These differences prevent comparisons of values measured at different section thicknesses and suggest that the section thickness should be kept constant in follow-up examinations.
Thin-section incremental CT has been proposed for the quantification of pulmonary emphysema (24,25,27). Spiral CT has the major advantage of enabling coverage of the entire thorax during one breath hold. Our analyses were based on two-dimensional image findings because macroscopic and microscopic quantifications are also based on two-dimensional measurements. A very strong correlation between two-dimensional and three-dimensional image findings has been shown and suggests that the results of our study could be applied to three-dimensional models (30).
Our study had limitations. First, only a few patients had severe emphysema; the majority of patients had moderate emphysema. This factor was inherent in the recruitment process, in which only those patients referred for thoracic surgery were accepted. It is not known whether the correlations between the CT measurements and the morphometric data would have been strengthened if a higher proportion of patients with severe emphysema had been included. Another limitation was related to the macroscopic sampling. We analyzed two macroscopic sections that were obtained mainly on the basis of the available nontumoral lung parenchyma and therefore might not have been perfectly representative of the entire lung parenchyma. However, to consider the cephalocaudal gradient in the distribution of pulmonary emphysema, these sections were obtained vertically (22). A third limitation may have stemmed from the fact that analysis of variance procedures require the variables to be normally distributed. Nevertheless, according to the central limit theorem (31), the number of patients in our study (n = 70) was sufficiently high such that we could be reasonably confident in the results obtained at these analyses.
In conclusion, we believe that reducing the radiation dose to 20 mAs should be recommended for patients undergoing CT quantification of pulmonary emphysema and that comparisons of findings between examinations, such as in follow-up studies, require that both the dose and the section thickness be kept constant.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: MIWD = corrected mean interwall distance • MP = corrected mean perimeter • RA960 = relative area of lung with attenuation coefficient lower than –960 HU • RA970 = relative area of lung with attenuation coefficient lower than –970 HU
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, A.M., P.A.G.; 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; literature research, A.M., P.A.G.; clinical studies, A.M., J.Z.; statistical analysis, V.D.M.; and manuscript editing, A.M., V.D.M., P.A.G.
|
References |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. Sverzellati, E. Calabro, G. Randi, C. La Vecchia, A. Marchiano, J-M. Kuhnigk, M. Zompatori, P. Spagnolo, and U. Pastorino Sex differences in emphysema phenotype in smokers without airflow obstruction Eur. Respir. J., June 1, 2009; 33(6): 1320 - 1328. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Rampinelli, E. De Fiori, S. Raimondi, G. Veronesi, and M. Bellomi In Vivo Repeatability of Automated Volume Calculations of Small Pulmonary Nodules With CT Am. J. Roentgenol., June 1, 2009; 192(6): 1657 - 1661. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E Ogawa, Y Nakano, T Ohara, S Muro, T Hirai, S Sato, H Sakai, M Tsukino, D Kinose, M Nishioka, et al. Body mass index in male patients with COPD: correlation with low attenuation areas on CT Thorax, January 1, 2009; 64(1): 20 - 25. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. O. Coxson Quantitative chest tomography in COPD research: chairman's summary. Proceedings of the ATS, December 15, 2008; 5(9): 874 - 877. [Full Text] [PDF]
|
|
|
|
|
|
J. D. Newell Jr. Quantitative Computed Tomography of Lung Parenchyma in Chronic Obstructive Pulmonary Disease: An Overview Proceedings of the ATS, December 15, 2008; 5(9): 915 - 918. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. O. Coxson Quantitative Computed Tomography Assessment of Airway Wall Dimensions: Current Status and Potential Applications for Phenotyping Chronic Obstructive Pulmonary Disease Proceedings of the ATS, December 15, 2008; 5(9): 940 - 945. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Matsuoka, Y. Kurihara, K. Yagihashi, M. Hoshino, and Y. Nakajima Airway Dimensions at Inspiratory and Expiratory Multisection CT in Chronic Obstructive Pulmonary Disease: Correlation with Airflow Limitation Radiology, September 1, 2008; 248(3): 1042 - 1049. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. C. Stoel, D. G. Parr, E. M. Bakker, H. Putter, J. Stolk, H. A. Gietema, A. M. Schilham, B. van Ginneken, R. J. van Klaveren, J. W. J. Lammers, et al. Can the Extent of Low-Attenuation Areas on CT Scans Really Demonstrate Changes in the Severity of Emphysema? Radiology, April 1, 2008; 247(1): 293 - 294. [Full Text] [PDF]
|
|
|
|
|
|
H. O. Coxson Computed tomography and monitoring of emphysema Eur. Respir. J., June 1, 2007; 29(6): 1075 - 1077. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) and 120 effective mAs (
) on 1.25-, 5.0-, and 10.0-mm-thick CT sections. Dose (P = .007 and P = .001, respectively), section thickness (P < .001 for both), and the interaction between dose and thickness (P = .036 and P = .005, respectively) were found to have significant effects on RA960 and RA970.
