HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Swensen, S. J. Articles by Weaver, A. L. Search for Related Content PubMed PubMed Citation Articles by Swensen, S. J. Articles by Weaver, A. L.

Lung Nodules: Dual-Kilovolt Peak Analysis with CT-Multicenter Study¹

¹ From the Department of Diagnostic Radiology (S.J.S., C.H.M.), the Division of Pulmonary and Critical Care Medicine (D.E.M.), and the Section of Biostatistics (A.L.W.), Mayo Clinic, 200 First St SW, Rochester, MN 55905; the Department of Diagnostic Radiology, Shiga Health Insurance Hospital, Japan (K.Y.); the Division Pulmonary and Critical Care Medicine (R.W.V.) and the Department of Diagnostic Radiology (J.R.M.), Mayo Clinic, Scottsdale, Ariz; and the Department of Diagnostic Radiology, Duke University Medical Center, Durham, NC (E.F.P.). Received January 6, 1999; revision requested February 18; revision received April 7; accepted April 26. Address reprint requests to S.J.S. (e-mail: swensen.stephen@mayo.edu).

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To test the following hypothesis: The greater the increase in the mean computed tomographic (CT) number of a radiologically indeterminate lung nodule from the CT number on a 140-kVp CT image to that on an 80-kVp CT image, the more likely the nodule is benign (ie, contains calcium).

MATERIALS AND METHODS: Two hundred forty indeterminate lung nodules were prospectively studied at four institutions: Mayo Clinic Scottsdale, Ariz (n = 160); Mayo Clinic Rochester, Minn (n = 50); Shiga Health Insurance Hospital, Otsu, Japan (n = 25); and Duke University Medical Center, Durham, NC (n = 5). Of the 240 nodules, 157 met the entrance criteria for this study and had a diagnosis. All nodules included were solid, 5–40-mm diameter, relatively spherical, homogeneous, and without visible evidence of calcification or fat. Each nodule was evaluated by using 3-mm-collimation, nonenhanced CT scans with both 140- and 80-kVp x-ray beams.

RESULTS: There were 86 (55%) benign and 71 (45%) malignant nodules. The median increase in the nodule mean CT number from the CT number on 140-kVp images to that on 80-kVp images was 2 HU for benign nodules and 3 HU for malignant nodules. This difference was not statistically significant. The area under the receiver operating characteristic curve was 0.505.

CONCLUSION: Dual–kilovolt peak analysis with current CT technology does not appear to be helpful in the identification of benign lung nodules.

Index terms: Lung, CT, 60.12111, 60.12115 • Lung, nodule, 60.281 • Lung neoplasms, CT, 60.12111, 60.12115, 60.30 • Lung neoplasms, diagnosis, 60.281, 60.30

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Evaluation of the solitary pulmonary nodule remains a substantial and costly challenge in modern medicine. Approximately 50% of indeterminate lung nodules for which surgery is performed for diagnosis are benign (1,2). Hospitalization for surgical removal of a nodule in 1996 cost on the order of $25,000 (3). A means short of biopsy by which diagnostic radiologists and clinicians can help to substantially reduce the percentage of benign nodules surgically excised for diagnosis is desirable.

Preliminary in vitro and in vivo work has indicated that dual–kilovolt peak computed tomography (CT) may be useful in the identification of benign lung nodules with low levels of calcification (4,5). At the high kilovolt peak values used in CT scanning, most soft-tissue interactions are from Compton scattering. Lowering the kilovolt peak from 140 to 80 kVp will increase dramatically the number of photoelectric interactions (the interaction probability for the photoelectric effect is proportional to Z³/E³, where Z is the atomic number and E is the photon energy). Thus for high-Z materials such as calcium (Z = 20), lowering the kilovolt peak will increase the x-ray attenuation coefficient, as reflected by the CT number (6). Decreasing the Compton interaction and increasing the photoelectric effect by lowering the kilovolt peak should theoretically help to identify nodules with visually undetectable levels of calcification (7–11).

A prospective multicenter study was performed to test the following hypothesis: The greater the increase in the mean CT number of a radiologically indeterminate lung nodule from the CT number on a 140-kVp CT image to that on an 80-kVp CT image, the more likely that the nodule is benign (ie, contains calcium).

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Between August 1995 and June 1997, we studied 240 solid, radiologically indeterminate lung nodules. Four centers participated in this prospective study: Mayo Clinic Scottsdale, Ariz (n = 160); Mayo Clinic Rochester, Minn (n = 50); Shiga Health Insurance Hospital, Otsu, Japan (n = 25); and Duke University Medical Center, Durham, NC (n = 5).

Informed consent and institutional review board approval were not required because kilovolt peak analysis of indeterminate lung nodules was part of routine evaluation based on supportive studies in the literature (4–11). However, institutional review board approval was nonetheless obtained at two centers (the Mayo Clinic and the Shiga Health Insurance Hospital) because of local requirements for all prospective or retrospective studies.

All patients were referred for CT examination of a known pulmonary nodule. The nodules had to be solid and relatively spherical (ie, short- and long-axis diameters were within a factor of 2 of each other). Nodules were excluded if they had benign patterns of calcification or fat on 3-mm-section CT images (12–14). In the judgment of the attending radiologist, there was no substantial CT artifact in the region of the nodule (eg, cardiac motion artifact, beam-hardening artifact from adjacent bone). All nodules were relatively homogeneous in the judgment of the attending radiologist (ie, no signs of necrosis, cavitation, calcification, low signal-to-noise ratio) on nonenhanced images (15). Recent (<1-month-interval) transthoracic needle aspiration biopsy of the nodule was an exclusion criterion. Nodules measured 5–40 mm in diameter. The diameter was calculated as a mean of the short- and long-axis diameters on transverse CT images obtained at lung window settings.

All patients were examined with 3-mm-collimation, nonenhanced CT. A 15-mm, z-axis, spiral cluster of scans was obtained through each nodule with the following technique: 3-mm collimation, 1-mm reconstruction interval, 1:1 pitch, 15-cm field of view, 80 and 140 kVp, and 280 mA. The standard reconstruction algorithm (as designated by the vendor) was used. Studies were performed with the following CT scanners: HiSpeed Advantage (GE Medical Systems, Milwaukee, Wis; n = 55), CT-Twin/RTS (Elscint, Hackensack, NJ; n = 160), and Quantex (GE-Yokogama Medical Systems, Tokyo, Japan; n = 25) scanners.

The nodule attenuation for each examination was quantified in terms of the mean CT number (in Hounsfield units) by one of the investigator chest radiologists (S.J.S., K.Y., E.F.P., and J.R.M.) at each site. All measurements were made at the time of the CT examination without knowledge of the histologic diagnosis. A single region of interest was carefully constructed to approximate the transverse shape of the nodule (Fig 1). The region-of-interest diameters were approximately 70% of the lung nodules' short- and long-axis diameters, as measured at mediastinal window settings on transverse images. All CT number measurements were performed on images obtained with mediastinal windows (window width, 400 HU; window level, 20 HU) to ensure consistency. The circular or oval mean region of interest was centered on the image closest to the nodule equator.

View larger version (115K): [in this window] [in a new window] [Download PPT slide]   Figure 1a. Dual-kilovolt peak CT study of a 1-cm right upper lobe nodule in 51-year-old man. The nodule was diagnosed as benign, given the stability during more than 2 years of radiologic follow-up. (a) Nonenhanced, 3-mm-collimation, transverse helical CT section obtained at 140 kVp. The mean CT number in the circular region of interest (1 ) centered on the nodule was 6.1 HU ± 4.5. (b) Nonenhanced, 3-mm-collimation, transverse helical CT section obtained at 80 kVp. The mean CT number in the circular region of interest (1 ) was 33.8 HU ± 28.9. We have no explanation for the apparent change in the size of the pleural effusion (a vs b).

View larger version (116K): [in this window] [in a new window] [Download PPT slide]   Figure 1b. Dual-kilovolt peak CT study of a 1-cm right upper lobe nodule in 51-year-old man. The nodule was diagnosed as benign, given the stability during more than 2 years of radiologic follow-up. (a) Nonenhanced, 3-mm-collimation, transverse helical CT section obtained at 140 kVp. The mean CT number in the circular region of interest (1 ) centered on the nodule was 6.1 HU ± 4.5. (b) Nonenhanced, 3-mm-collimation, transverse helical CT section obtained at 80 kVp. The mean CT number in the circular region of interest (1 ) was 33.8 HU ± 28.9. We have no explanation for the apparent change in the size of the pleural effusion (a vs b).

The clinical histories of all patients were later reviewed to determine whether a clinical or histologic diagnosis had been made (K.Y., R.W.V., D.E.M., and E.F.P.). Nodules were classified as malignant only if this diagnosis was confirmed histologically or cytologically. All bronchial carcinoids and fibrous tumors of the pleura were classified as malignancies (16). All benign neoplasm diagnoses (hamartomas) were made histologically. Nodules were classified as granulomas if this diagnosis was histologically confirmed or if there was no radiologic evidence of growth during at least 2 years of surveillance (12,13,17,18).

We were unable to obtain a clinical or histologic nodule diagnosis in 83 patients because the surveillance period was less than 2 years, the patient was no longer being managed at the respective institution and could not be contacted, or the diagnosis could not be verified after death. These patients were excluded from statistical analysis. Thus, there were clinical or histologic diagnoses and technically adequate CT images for 157 patients (88 men, 69 women; age range, 21–89 years; mean age, 64.9 years).

To analyze the statistical significance and clinical relevance of the data, we divided the patients into two groups: patients with a malignant neoplasm and patients with granulomas or benign neoplasms. The distributions of nodule attenuation measurement changes and diameters were compared for the two groups by using the Wilcoxon rank sum test, because the distributions of these variables were non-Gaussian.

Receiver operating characteristic analysis was used to summarize the usefulness of the mean change in nodule CT number as a marker for granulomas and benign neoplasms (vs malignant neoplasms). An estimate of the area (and its standard error) under the receiver operating characteristic curve was made by using nonparametric methods, which require no distributional assumptions (19).

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
A total of 157 patients had a nodule with a clinical or histologic diagnosis and had undergone a dual–kilovolt peak nodule study. The prevalence of malignancy was 45.2% (71 of 157 nodules).

Benign Lesions
Eighty-six patients (55%; 43 men, 43 women; age range, 22–88 years; mean age, 62.9 years) had a benign abnormality. There were 77 granulomas, seven other benign nodules, and two benign neoplasms. Fourteen of the 77 granulomas had histologic proof of diagnosis. The remaining 63 were considered granulomas because there was no radiologic evidence of growth during a follow-up of 2 years or more or because the nodule decreased substantially in size or resolved. The two benign neoplasms were diagnosed as hamartomas after surgical removal. The median increase in mean CT nodule attenuation measurement from 140 to 80 kVp images was 2 HU. Dual–kilovolt peak results and nodule sizes are summarized in Table 1.

Comparison between Benign and Malignant Nodules
The difference in the mean CT number changes between benign and malignant nodules was not statistically significant (Fig 2)( Table 1). The area under the receiver operating characteristic curve was 0.505, which indicated that the discriminatory value of this attenuation measurement is no different than what one would expect due to chance alone.

View larger version (15K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Graph shows the distribution of the mean CT number changes between 140- and 80-kVp images for malignant and benign nodules. Note that, except for three outlier benign nodules, there is substantial overlap in the distribution of mean CT number changes between benign and malignant nodules.

Calcium Content of Nodules
Fifteen of the 25 nodules studied in the Japanese arm of this protocol were analyzed for calcium content (Table 2). The relationship between CT number change from 80- to 140-kVp images and the calcium content of the lung nodules was studied. There was a moderate correlation (Pearson correlation, r = 0.57, P = .026).

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

We analyzed the relationship between the amount of histologic calcification in 15 nodules and the change in the mean CT number between 80- and 140-kVp images. These data corroborate the findings of others (4,5) that the difference in CT number has, at best, a moderate relationship with the level of nodule calcification. We found that the mean and median changes in CT number for benign and malignant nodules were not significantly different. The range of change in CT number was substantial: 108 HU for benign nodules and 63 HU for malignant nodules (Table 1).

The three outlier benign nodules (Fig 2) were nodules that were at the threshold for visual detection of calcification (ie, they had high attenuation, but the attenuation was not as high as that of the rib, which is the attenuation necessary to allow confident benign-pattern-calcification diagnosis subjectively). It is possible that this dual–kilovolt peak technique has some clinical utility for that small subset of nodules that have high attenuation but that are subjectively "not definitely calcified." It is interesting that none of the malignant nodules had a change in CT number greater than 40 HU, while three (3%) of the benign ones did. While this small percentage of nodules is too low to make the technique of generalized clinical utility, it may show a potential niche for the dual–kilovolt peak CT technique. All three of these nodules were benign granulomas and visually had an attenuation that was suggestive, but not diagnostic, of benign-pattern calcification.

However, findings of this study did not confirm that these theoretic considerations have practical utility for patient care (4,5). Our study involved the use of spiral scanners, while the clinical studies cited used conventional CT scanners. It is our belief that with spiral CT and overlapping reconstructions of volumetric data, the likelihood that one is actually comparing approximately the same portion of the lung nodule on the 140- and 80-kVp images is higher than it is with the use of conventional CT without volumetric scanning capability. Because nodules are often not homogeneous in their attenuation and distribution of calcifications, it is critical that this same equatorial image be compared for this technique.

It is possible that the failure of this technique in our population was due to a very low rate of calcification in the benign nodules. However, given our large sample size, this is not likely. Even if it were so, this group of nodules is widely representative of a prevalent problem throughout the world. Our sample from Japan and several locales in the United States included histoplasmosis, coccidioidomycosis, and tuberculosis granulomas.

Another possible explanation for the failure of this theoretically plausible CT technique is that approximately 10% of malignancies contain calcification (25,26). This fact, coupled with the likely presence of some uncalcified benign nodules, could have resulted in substantial overlap of attenuation measurements. In fact, in the Japanese arm of the study, the calcium content of 11 malignant and four benign nodules was within the same range (Table 2).

Another consideration is that the 80-kVp images with a relatively low signal-to-noise ratio (Fig 1b) are unreliable. It is possible that in vivo measurements are not accurate enough with today's CT technology for us to consistently detect low levels of calcification.

A final consideration has to do with CT scanner calibration and the use of CT scanners from different manufacturers at different institutions. The comparison of the absolute value of CT numbers between different CT scanners, manufacturers, and institutions can be problematic because of variability in scanner calibration. The calibration procedure, which must be performed at each x-ray beam energy used, determines the CT number assigned to scanned objects. The results of this procedure can be characterized by a calibration line, which relates measured CT numbers to known CT numbers for specific test objects. The calibration line is described both by its slope and y intercept. Ideally, the slope should equal 1 HU, and the y intercept (CT number of water) should equal 0 HU. Typically, the slope of the CT number calibration line is very stable, while the y intercept fluctuates (±5 HU is the regulatory limit in Minnesota). Changes in CT numbers are inherently more precise than absolute CT numbers. This is because subtracting two points from the same calibration line will cancel any effects from fluctuating y intercepts.

The use of two energies (80 and 140 kVp) requires the use of two CT number calibration lines, and fluctuations in y intercepts will not necessary cancel. Thus, the change in CT number assessed in any dual–kilovolt peak study is subject to the effects of calibration errors. The system used in our study will not allow scanning to proceed if a given energy has not been initially calibrated. Again, this initial calibration determines the slope and intercept of the calibration line. More frequent "fast-cals" (which check and reset the y intercept) are not necessary for scanning but are routinely performed (typically daily) at our institution to minimize fluctuations in the y intercept. In our experience, by using well-maintained systems, these fluctuations are less than ±3 HU. Hence, errors in the dual–kilovolt peak mean CT number changes of about 0–6 HU (3 HU on average) may exist because of calibration fluctuations.

Thus, a limitation of any study in which the dual–kilovolt peak technique is attempted is that mean CT number changes between the scans at two energy points will need to overcome errors introduced by fluctuations in the y intercept of the calibration lines. For our systems, we estimate this value to be about 3–6 HU. For systems not as carefully maintained, an even larger offset is required. In addition, systematic errors (eg, consistently depressed or elevated y intercepts) in less frequently calibrated systems could cause nonzero mean CT number changes between energies because of calibration errors (ie, false-positive findings).

Our results did not corroborate the hypothesis. We were not able to show a relationship between nodule diagnosis and an increase from the mean attenuation measurement on a 140-kVp CT image to that on an 80-kVp CT image. We found no evidence to support the clinical utility of dual–kilovolt peak CT for the detection of calcification in radiologically indeterminate lung nodules.

	Footnotes

 
Author contributions: Guarantor of integrity of entire study, S.J.S.; study concepts and design, S.J.S.; definition of intellectual content, S.J.S.; literature research, S.J.S.; clinical studies, all authors; data acquisition, all authors; data analysis, S.J.S., A.L.W.; statistical analysis, S.J.S., A.L.W.; manuscript preparation, editing, and review, all authors.

	References

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 

1. Bernard A. Resection of pulmonary nodules using video-assisted thoracic surgery: the Thorax Group. Ann Thorac Surg 1996; 61:202-205.[Abstract/Free Full Text]
2. Keagy BA, Starek PJ, Murray GF, et al. Major pulmonary resection for suspected but unconfirmed malignancy. Ann Thorac Surg 1984; 38:314-316.[Abstract]
3. 1996 St Anthony's DRG guidebook Reston, Va: St Anthony's Publishing, 1996.
4. Higashi Y, Nakamura H, Matsumoto T, Nakanishi T. Dual-energy computed tomographic diagnosis of pulmonary nodules. J Thorac Imaging 1994; 9:31-34.[Medline]
5. Bhalla M, Shepard JO, Nakamura K, Kazerooni EA. Dual kV CT to detect calcification in solitary pulmonary nodule. J Comput Assist Tomogr 1995; 19:44-47.[Medline]
6. Attix FH. Introduction to radiological physics and radiation dosimetry New York, NY: Wiley, 1986; 138-146.
7. Curry TS, Dowdey JE, Murry RC, Jr. Christensen's physics of diagnostic radiology 4th ed. Philadelphia, Pa: Lea & Febiger, 1990; 61-86.
8. In: Hendee WR, eds. The physical principles of computed tomography. 1st ed. Boston, Mass: Little, Brown, 1983; 101-111.
9. Brooks RA. A quantitative theory of the Hounsfield units and its application to dual energy scanning. J Comput Assist Tomogr 1977; 1:487-493.[Medline]
10. Hickey NM, Niklason LT, Sabbagh E, Fraser RG, Barnes GT. Dual-energy digital radiographic quantification of calcium in simulated pulmonary nodules. AJR Am J Roentgenol 1987; 148:19-24.[Abstract/Free Full Text]
11. Cann CE, Gamsu G, Birnberg FA, Webb WR. Quantification of calcium in solitary pulmonary nodules using single- and dual-energy CT. Radiology 1982; 145:493-496.[Free Full Text]
12. Khouri NF, Meziane MA, Zerhouni EA, Fishman EK, Siegelman SS. The solitary pulmonary nodule: assessment, diagnosis, and management. Chest 1987; 91:128-133.[Abstract/Free Full Text]
13. Swensen SJ, Jett JR, Payne WS, Viggiano RW, Pairolero PC, Trastek VF. An integrated approach to evaluation of the solitary pulmonary nodule. Mayo Clin Proc 1990; 65:173-186.[Medline]
14. Siegelman SS, Khouri NF, Scott WW, Jr, et al. Pulmonary hamartoma: CT findings. Radiology 1986; 160:313-317.[Abstract/Free Full Text]
15. Yamashita K, Matsunobe S, Tsuda T, et al. Intratumoral necrosis of lung carcinoma: a potential diagnostic pitfall in incremental dynamic computed tomography analysis of solitary pulmonary nodules?. J Thorac Imaging 1997; 12:181-187.[Medline]
16. England DM, Hochholzer L, McCarthy MJ. Localized benign and malignant fibrous tumors of the pleura: a clinicopathologic review of 223 cases. Am J Surg Pathol 1989; 13:640-658.[Medline]
17. Good CA. The solitary pulmonary nodule: a problem of management. Radiol Clin North Am 1963; 1:429-438.
18. Good CA, Wilson TW. The solitary circumscribed pulmonary nodule: study of seven-hundred-five cases encountered roentgenologically in a period of three and one-half years. JAMA 1958; 166:210-215.
19. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143:29-63.[Abstract/Free Full Text]
20. Gurney JW. Determining the likelihood of malignancy in solitary pulmonary nodules with Bayesian analysis. Radiology 1993; 186:405-413.[Abstract/Free Full Text]
21. Zerhouni EA, Boukadoum M, Siddiky M, et al. A standard phantom for quantitative CT analysis of pulmonary nodules. Radiology 1983; 149:767-773.[Abstract/Free Full Text]
22. Swensen SJ, Brown LR, Colby TV, Weaver AL, Midthun DE. Lung nodules enhancement at CT: prospective findings. Radiology 1996; 201:447-455.[Abstract/Free Full Text]
23. Patz EF, Lowe VJ, Hoffman JM, et al. Focal pulmonary abnormalities: evaluation with F-18 fluorodeoxyglucose PET scanning. Radiology 1993; 188:487-490.[Abstract/Free Full Text]
24. Gupta NC, Frank AR, Dewan NA, et al. Solitary pulmonary nodules: detection of malignancy with PET with 2-[F-18]fluoro-2-deoxy-D-glucose. Radiology 1992; 184:441-444.[Abstract/Free Full Text]
25. Mahoney MC, Shipley RT, Corcoran HL, Dickson BA. CT demonstration of calcification in carcinoma of the lung. AJR Am J Roentgenol 1990; 154:255-258.[Abstract/Free Full Text]
26. Siegelman SS, Khouri NF, Leo FP, et al. Solitary pulmonary nodules: CT assessment. Radiology 1986; 160:307-312.[Abstract/Free Full Text]

This article has been cited by other articles:

				Y. Nakayama, K. Awai, Y. Funama, M. Hatemura, M. Imuta, T. Nakaura, D. Ryu, S. Morishita, S. Sultana, N. Sato, et al. Abdominal CT with Low Tube Voltage: Preliminary Observations about Radiation Dose, Contrast Enhancement, Image Quality, and Noise Radiology, December 1, 2005; 237(3): 945 - 951. [Abstract] [Full Text] [PDF]

				C. Schaefer-Prokop and M. Prokop New imaging techniques in the treatment guidelines for lung cancer Eur. Respir. J., February 1, 2002; 19(35_suppl): 71S - 83s. [Abstract] [Full Text] [PDF]

				M. Pandit-Bhalla, C. H. McCollough, and S. J. Swensen Dual-Kilovolt CT of Solitary Pulmonary Nodules: Importance of Equipment Calibration and Soft-Tissue Controls Drs McCollough and Swensen respond: Radiology, January 1, 2001; 218(1): 300 - 301. [Full Text]

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Swensen, S. J. Articles by Weaver, A. L. Search for Related Content PubMed PubMed Citation Articles by Swensen, S. J. Articles by Weaver, A. L.