Chest ACCP Education Calendar
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Guest Access | Sign In via User Name/Password
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tateishi, U.
Right arrow Articles by Arai, Y.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tateishi, U.
Right arrow Articles by Arai, Y.
(Chest. 2005;128:3276-3283.)
© 2005 American College of Chest Physicians

Prediction of Lung Adenocarcinoma Without Vessel Invasion*

A CT Scan Volumetric Analysis

Ukihide Tateishi, MD, PhD; Hajime Uno, PhD; Kan Yonemori, MD; Mistuo Satake, MD; Masahiro Takeuchi, ScD, MPH and Yasuaki Arai, MD, PhD

* From the Division of Diagnostic Radiology (Drs. Tateishi, Yonemori, Satake, and Arai), National Cancer Center Hospital, Tokyo, Japan; the Department of Biostatistics (Dr. Uno), Harvard School of Public Health, Boston, MA; and the Division of Biostatistics (Takeuchi), Kitasato University Graduate School, Tokyo, Japan.

Correspondence to: Ukihide Tateishi, MD, PhD, Division of Diagnostic Radiology, National Cancer Center Hospital, Tsukiji, Chuo-Ku, 104-0045, Tokyo, Japan; e-mail: utateish{at}ncc.go.jp


   Abstract
TOP
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Study objectives: Patients with lung adenocarcinoma without vessel invasion have a favorable prognosis after resection and are among the candidates for limited surgery. The purpose of the present study was to predict lung adenocarcinoma without vessel invasion based on a volumetric analysis of the lesion with a CT scan prior to the operation.

Methods: CT scan images were obtained in 288 consecutive patients with adenocarcinoma of the lung before surgical resection. Total tumor volume, the volume of the nonsolid component, and the proportion occupied by the nonsolid component were calculated by the perimeter method. The performance of the derived logistic regression model and the volumetric results were evaluated by receiver operating characteristic analysis. The model derived for the prediction of tumors without vessel invasion was assessed by means of the leave-one-out cross-validation technique.

Results: The pathologic diagnosis was adenocarcinoma with vessel invasion in 160 cases, and without vessel invasion in 128 cases. The median total tumor volume, the median volume of the nonsolid component, and median proportion occupied by the nonsolid component were 1,123.7 mm3, 253.4 mm3, and 58.0%, respectively. With the derivation of the predictive rule, stepwise regression yielded the following five features: the proportion occupied by the nonsolid component; spiculation; pleural indentation; gender; and tumor size. The Az value, a measure of diagnostic power represented as the area under the curve, was 0.957 for prediction of lung adenocarcinoma without vessel invasion. The cross-validation accuracy achieved by applying the rule was 90.3%.

Conclusions: The proportion occupied by the nonsolid component based on a CT scan volumetric analysis was a reliable predictor of tumors without vessel invasion in patients with adenocarcinoma of the lung.

Key Words: CT scan • lung • lung cancer


   Introduction
TOP
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Patients with lung adenocarcinoma without vessel invasion clearly have the best outcome after resection, and they are among the candidates for limited surgery.12 Some studies34 have highlighted the potential diagnostic role of thin-section CT scanning in identifying nonsolid components of adenocarcinoma of the lung. The nonsolid component is larger in patients with noninvasive tumors and has been shown to discriminate between patients with noninvasive tumors and patients with advanced tumors with a high degree of power.5 In addition, the proportion of the tumor occupied by the nonsolid component correlates well with the absence of vascular or lymphatic invasion and with better outcome.678910 Thus, identification of the size of the nonsolid component on CT scan images of lung adenocarcinoma is a potential surrogate measurement for tumor aggressiveness.

The reliability of almost all data on the proportion occupied by the nonsolid component is limited, because the investigators assumed that the lesions were spherical and used the maximum cross-sectional diameter on CT scan images to calculate it,56 and the only large series to date used the maximum cross-sectional area of the tumor to predict invasion.11 To our knowledge, few studies have used CT scanning to calculate the volume of the nonsolid component within tumors and to correlate the proportion occupied by the nonsolid component with pathologic characteristics of tumor invasion. Moreover, most studies to date have proposed diagnostic criteria for lesions without testing their diagnostic validity.67891011

We therefore conducted both a derivation and validation cohort study of patients with adenocarcinoma of the lung in order to predict tumors without vessel invasion. Our results will assist physicians in estimating more accurately the probability of a tumor being unassociated with vessel invasion and to decide whether further investigation is necessary to rule the presence of a noninvasive tumor in or out.


   Materials and Methods
TOP
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
A retrospective review of the pathologic records for the period between October 2001 and January 2004 identified 288 who had been patients treated for adenocarcinoma of the lung the maximum dimension of which was < 2 cm. A consecutive subset of 288 patients contributed to the derivation analysis and the cross-validation analysis based on the leave-one-out method. The study population consisted of 113 men and 175 women, and their mean age was 64.7 years (age range, 41 to 86 years). All patients had undergone surgical resection consisting of wedge resection or lobectomy. Complete dissection (n = 96; 33%) and sampling of mediastinal or hilar lymph nodes (n = 192; 67%) were performed. The nodes included high and low ipsilateral, paratracheal, subcarinal, and inferior pulmonary ligament lymph nodes, and any other suspicious lymph nodes identified at surgery. Surgical specimens were fixed in formalin and embedded in paraffin. Four-micrometer sections were stained with hematoxylin and eosin and elastica-van Gieson stain. Tumors without vessel invasion were diagnosed when no lymphatic or blood vessel invasion was identified within the lesion microscopically. Based on the current international TNM classification for staging lung cancer,12 274 tumors (95%) were classified as stage Ia, and the other 14 tumors (5%) were classified as stage IIa. The clinical records of all patients were available for review. No patients were lost to follow-up, which began on the date of the initial operation. The median duration of follow-up was 22 months, and ranged from 1 to 40 months. This study was approved by our institutional review board after confirmation of informed consent by the patients to a review of their records and images.

CT scanning was performed with a multidetector scanner (Aquilion V-detector; Toshiba Medical Systems; Tokyo, Japan) by using axial 2.0-mm x 4 or 1.0-mm x 16 modes (ie, 4 or 16 images per gantry rotation), 120 kVp, 200 to 250 mA, and a 0.5-s scanning time. Thin-section CT scan images were obtained using 2.0-mm sections that were reconstructed at 2.0-mm intervals by means of a high-spatial frequency algorithm and were retrospectively retargeted to each lung with a 20-cm field of view. All images were displayed at window settings for lung (center, –600 Hounsfield units; width, 2,000 Hounsfield units). Image viewing and manipulation were controlled with a workstation (ZIOSOFT M900, Quadra, version 3.10f; ZIOSoft Inc; Tokyo, Japan) that allows the reader to draw lines through regions of interest and the perimeters around them.

The CT scan images were assessed in random order by two independent observers without reference to the clinical findings. The observers examined the images for the following: maximum tumor diameter; nodular edge (irregular or not); presence of a nonsolid component; presence of an air-bronchogram; presence of cavitation, lobulation, pleural indentation, bubble-like lucencies, spiculation, vascular convergence, bronchiectasis, or bronchiolectasis; and the lobe in which the lesion was located (eg, upper, middle, or lower). Nonsolid component was defined as an area of ground-glass attenuation or hazy increased parenchymal attenuation without obscuring of the underlying vascular markings. We distinguished between nonsolid tumors and part-solid tumors, with the former being defined as tumors containing only a nonsolid component, and the latter as tumors that contained a partially solid component. Bubble-like lucencies were diagnosed when there were multiple cystic air spaces measuring ≤ 5 mm in diameter within the lesion surrounded by a wall of variable thickness.13 After an initial independent evaluation, the two observers reviewed all cases in which their interpretations disagreed and reached a final decision by consensus.

The volume of each tumor was calculated by the perimeter method.141516 The cross-sectional areas of the entire tumor and of the solid component were calculated by a workstation that manipulates a voxel matrix of 512 x 512 pixels. Two board-certified radiologists who were experienced with image-viewing and image-manipulation software drew a line around the perimeter of each tumor twice. The total tumor volume and the volume of the solid component were calculated by summing the cross-sectional areas and multiplying by the section increment. The volume of the nonsolid component was calculated by subtracting the volume of the solid component from the total volume. The averages of the two volume values calculated by each of the two observers were used in the analyses.

Statistical Analysis
Interobserver variation in relation to CT scan findings was quantified as the weighted {kappa} coefficient of agreement. The predictive performance of volumetric data was evaluated by receiver operating characteristic (ROC) analysis, and the areas under curve were represented by the Az values, which are a measure of diagnostic power represented by the area under the curve.1718 A stepwise procedure was used in the logistic regression analysis to select the independent variables that should have been included in the model to predict tumors without vessel invasion. A variable was entered into the model if the probability of its score statistic was < 0.05. The odds ratio (OR) and 95% confidence interval (CI) for the multivariate predictors were estimated. The Hosmer-Lemeshow test was also performed to evaluate goodness-of-fit.19 Calibration curves comparing the observed proportion of tumors without vessel invasion with the probability of tumors without vessel invasion ordered by the increasing probability of tumors without vessel invasion were constructed. The performance of the learned model was verified by using the leave-one-out cross-validation method, in which all cases but one were used to train the prediction rule, which was then applied to the single excluded case.202122 This procedure was repeated for each case, until each case had been left out only once. Cross-validation accuracy was calculated by comparing the predicted response and the observed response.23 The following variables were considered for their prognostic value: age; gender; presence or absence of vessel invasion; tumor size; total tumor volume; the volume of the nonsolid component; the proportion occupied by the nonsolid component; and CT scan findings. Univariate analysis was performed by comparing Kaplan-Meier disease-free or recurrence-free survival curves and carrying out log-rank tests. All analyses were conducted using a statistical software package (SAS, version 8.2; SAS Institute, Cary, NC; and R software, version 1.9.0; R project, Center for Computational Intelligence; Vienna, Austria).


   Results
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
The clinical characteristics and outcomes of all patients with adenocarcinoma of the lung are summarized in Table 1 . The pathologic diagnosis was lung adenocarcinoma with vessel invasion in 160 of the 288 patients (55.6%), and without vessel invasion in the other 128 patients (44.4%). Female patients had a predilection for tumors without vessel invasion according to the results of the univariate analysis. The median age at presentation was 66 years (age range, 41 to 86 years). Age was not statistically associated with tumor type. Most tumors (n = 169; 58.7%) were in the upper lobe. The median tumor size was 15.0 mm (range, 5.0 to 20.0 mm). Tumors with vessel invasion were significantly larger than tumors without vessel invasion. Among the tumors without vessel invasion, nonsolid tumors (n = 81; 28.1%) were more common than part-solid or solid tumors (n = 47; 16.3%); whereas, among the tumors with vessel invasion, solid or part-solid tumors (n = 159; 55.2%) were more common than nonsolid tumors (n = 1; 0.3%). Tumors with a nonsolid component were more common among the tumors without vessel invasion than among the tumors with vessel invasion.


View this table:
[in this window]
[in a new window]

  		Table 1.. Patient Demographics*

 
There was good interobserver agreement in the analysis of the CT scan findings (weighted {kappa} coefficient, 0.63 to 0.79). The following CT scan findings were more frequently identified in tumors with vessel invasion than in tumors without vessel invasion according to the univariate analysis: cavitation; air bronchogram; irregular margin; speculation; lobulation; and bronchiectasis or bronchiolectasis. Irregular margins were common in tumors without vessel invasion but were more frequent in tumors with vessel invasion. Vascular convergence was a less common finding and was noted only in patients with tumors that were associated with vessel invasion. Bubble-like lucencies were observed in one tumor with vessel invasion and in one tumor without vessel invasion.

Volumetric Analysis
The results of the volumetric analysis are summarized in Table 2 . The median total tumor volume, the median volume of the nonsolid component, and the median proportion occupied by the nonsolid component were 1,123.7 mm3 (range, 43.6 to 6,912.2 mm3), 253.4 mm3 (range, 0 to 6,593.4 mm3), and 58.0% (range, 0 to 100%), respectively. Significant differences were found between the mean values for total volume and the proportion values in two groups. The ROC analyses to predict tumors without vessel invasion revealed that Az values of total tumor volume, the volume of the nonsolid component, and the proportion occupied by the nonsolid component were 0.699 (95% CI, 0.638 to 0.760), 0.714 (95% CI, 0.653 to 0.775), and 0.928 (95% CI, 0.895 to 0.961), respectively (Fig 1 ). The potential predictors of tumors without vessel invasion at the threshold of each median value were as follows: total tumor volume; the volume of the nonsolid component; and the proportion occupied by the nonsolid component (Table 3 ). Univariate analysis showed that our predictor of interest, the proportion occupied by the nonsolid component, was highly predictive of tumors without vessel invasion at an OR of 42.8 (95% CI, 21.6 to 84.8) with 80.0% as the threshold value.


View this table:
[in this window]
[in a new window]

  		Table 2.. Summary Statistics of Volumetric Measurements*

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1.. Top, A: ROC curve for total tumor volume determined by the perimeter method. The mean (± SE) Az value was 0.699 ± 0.030. Middle, B: the ROC curve for the volume of the nonsolid component determined by the perimeter method. The mean Az (± SE) value was 0.714 ± 0.030. Bottom, C: ROC curve for the proportion occupied by the nonsolid component determined by the perimeter method. The mean Az (± SE) value was 0.928 ± 0.015.

 

View this table:
[in this window]
[in a new window]

  		Table 3.. Potential Predictors of Tumors Without Vessel Invasion*

 
Derivation of the Prediction Rule
From the 13 variables that reached the level of statistical significance in the univariate analyses comparing tumors with and without vessel invasion, we excluded location of the tumor because the absolute difference fell within the precision range of the test. The logistic regression analysis identified the following five significant predictors of tumors without vessel invasion: proportion occupied by the nonsolid component; spiculation; pleural indentation; gender; and tumor size (Table 4 ). The Az value of the ROC analysis for prediction of a tumor without vessel invasion in the derivation phase was 0.957 (Fig 2 ). We adopted the threshold that yielded an appropriate tradeoff between sensitivity and specificity (ie, probability of a tumor without vessel invasion, 0.5). At that point in the ROC curve, the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of the rule were 88.3%, 91.9%, 90.3%, 89.7%, and 90.7%, respectively. The results of the goodness-of-fit test ({chi}2, 3.6563; degrees of freedom, 8; p = 0.89) indicated that the observed proportion of patients with tumors without vessel invasion was similar to the predicted proportion in the derivation group. The calibration curves for the derivation data demonstrated good calibration of the prediction rule.


View this table:
[in this window]
[in a new window]

  		Table 4.. Significant Predictors of Tumors Without Vessel Invasion

 


View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2.. ROC curve for the derived model to predict tumor without vessel invasion. The probability of the occurrence of a tumor without vessel invasion based on the derived model can be considered diagnostic, and the tradeoff between sensitivity and specificity at various thresholds of the probability of a tumor without vessel invasion is given by the ROC curve.

 
Cross-Validation Accuracy
When the rule derived from the 288 patients was applied to the leave-one-out cross-validation cohort, the validation accuracy based on the leave-one-out method was 90.3%, which was quite similar to the model accuracy, suggesting that the rule that was derived to predict tumors without vessel invasion is stable.

Prognostic Analysis
At the last follow-up, 1 of the 288 patients (0.3%) had died, 12 patients (4.2%) were alive with recurrent disease, and the 5-year overall survival rate was 98.7%. Univariate analysis revealed that none of the variables had a significant impact on overall survival. The 5-year recurrence-free survival rate was 83.5%. The patients with tumors without vessel involvement had a 5-year recurrence-free rate of 88.0%, which was significantly better than the rate of 74.6% among the patients with vessel invasion (p < 0.05). Cavitation and vascular convergence were significantly associated with recurrence-free survival according to the univariate analysis (p < 0.01).


   Discussion
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
The proportion occupied by the nonsolid component is characteristically higher in patients with lung adenocarcinoma without vessel invasion, and several studies45 have demonstrated that it distinguishes tumors without vessel invasion from tumors with vessel invasion. Nevertheless, controversy remains regarding its reproducibility in optimally distinguishing tumors without vessel invasion from tumors with vessel invasion. In this study, we investigated the diagnostic capacity of the proportion occupied by the nonsolid component to predict tumors without vessel invasion in a well-characterized study population.

We performed ROC analysis to assess the ability of the proportion occupied by the nonsolid component to discriminate between tumors with and without vessel invasion. We have documented that the diagnostic probability of the proportion occupied by the nonsolid component for tumors without vessel invasion is accurate. The Az value observed for the proportion occupied by the nonsolid component was 0.957. In our study population, its discriminatory capacity yielded positive and negative predictive values for tumors without vessel invasion of 89.7% and 90.7%, respectively. These observations suggest that the identification of the proportion occupied by the nonsolid component on CT images may be useful for predicting tumors without vessel invasion in patients with adenocarcinoma of the lung.

At both baseline screenings and repeat CT screenings for lung cancer, tumors containing a nonsolid component have been found to be a significant sign of malignancy.24 Henschke and colleagues25 found that 19% of the 233 cases with positive results at baseline screening had a tumor with the nonsolid component and that the tumors were predominantly bronchioloalveolar carcinoma and adenocarcinoma with bronchioloalveolar carcinoma features. These predominant histologic types of malignancy corresponded to noninvasive and invasive tumors of lung adenocarcinoma. The quantification of the extent or growth rate of the solid and nonsolid components of tumors is necessary during CT scan screening for lung cancer.

Swensen and colleagues18 created a multivariate logistic regression model to predict a malignant solitary pulmonary nodule (SPN) in a derivation and validation analysis. Of 629 radiologically intermediate nodules in their study, 23% were malignant SPNs. The investigators identified the following three independent findings that predicted malignant SPNs: upper lobe distribution; tumor size; and spiculation. Among these findings, spiculation was associated with the prediction model in our results. Pleural indentation and male predilection were also significant predictors of tumors with vessel invasion in our study.

Some studies262728 using the segmentation algorithm of software have yielded calculations of tumor volume in three dimensions. The excellent interobserver variability suggests that tumor volume estimations by different observers can be reliably compared when three-dimensional volumetric software is used. However, this technique does not enable the segmentation of tumors that contain a nonsolid component.2728 Since most tumors in our study contained a nonsolid component and were not appropriate for three-dimensional volumetric analysis, tumor volume was calculated by the perimeter method, which had potential sources of error that affected the results of volumetric analyses.29

The proportion occupied by the nonsolid component was dichotomized using the median value for a threshold level in the univariate and multivariate analyses performed. In both the univariate and multivariate analyses, the proportion occupied by the nonsolid component yielded the highest point estimates and CIs for the ORs of tumors without vessel invasion. However, it should be noted that there was enough of a difference between the values for patients in whom the nonsolid component occupied a high proportion of the tumor and those for patients in whom the nonsolid component occupied a small proportion of the tumor that similar results could have been obtained with different threshold values.

There are other potential limitations of this study. The size of the sample may also have led to false-positive results because of the number of covariates included in the initial analysis. However, the strength of the association with our primary outcome of interest, as well as the historical precedence of other significant predictors in our multivariate analysis, lends credence to our conclusions. In summary, the results of our study confirm that the proportion occupied by the nonsolid component of a tumor on CT scans is a reliable predictor of tumors without vessel invasion with much greater confidence than was possible in the past.


   Footnotes
 
Abbreviations: CI = confidence interval; OR = odds ratio; ROC = receiver operating characteristic; SPN = solitary pulmonary nodule

Dr. Uno received support for this research by Banyu Life Science Foundation International.

Received for publication January 12, 2005. Accepted for publication May 4, 2005.


   References
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Travis, WD, Colby, TV, Corrin, B, et al (1999) Histological typing of lung and pleural tumors. Springer. Berlin, Germany: [Medline]
  2. Noguchi, M, Morikawa, A, Kawasaki, M, et al Small adenocarcinoma of the lung: histologic characteristics and prognosis. Cancer 1995;75,2844-2852[CrossRef][ISI][Medline]
  3. Jang, HJ, Lee, KS, Kwon, OJ, et al Bronchioloalveolar carcinoma: focal area of ground-glass attenuation at thin-section CT as an early sign. Radiology 1996;199,485-488[Abstract/Free Full Text]
  4. Kuriyama, K, Seto, M, Kasugai, T, et al Ground-glass opacity on thin-section CT: value in differentiating subtypes of adenocarcinoma of the lung. AJR Am J Roentgenol 1999;173,465-469[Abstract/Free Full Text]
  5. Ohde, Y, Nagai, K, Yoshida, J, et al The proportion of consolidation to ground-glass opacity on high resolution CT is a good predictor for distinguishing the population of non-invasive peripheral adenocarcinoma. Lung Cancer 2003;42,303-310[CrossRef][ISI][Medline]
  6. Aoki, T, Tomoda, Y, Watanabe, H, et al Peripheral lung adenocarcinoma: correlation of thin-section CT findings with histologic prognostic factors and survival. Radiology 2001;220,803-809[Abstract/Free Full Text]
  7. Suzuki, K, Yokose, T, Yoshida, J, et al Prognostic significance of the size of central fibrosis in peripheral adenocarcinoma of the lung. Ann Thorac Surg 2000;69,893-897[Abstract/Free Full Text]
  8. Kodama, K, Higashiyama, M, Yokouchi, H, et al Prognostic value of ground-glass opacity found in small lung adenocarcinoma on high-resolution CT scanning. Lung Cancer 2002;33,17-25[Medline]
  9. Matsunaga, H, Yokoi, K, Anraku, M, et al Proportion of ground-glass opacity on high-resolution computed tomography in clinical T1N0M0 adenocarcinoma of the lung: a predictor of lymph node metastasis. J Thorac Cardiovasc Surg 2002;124,278-284[Abstract/Free Full Text]
  10. Kim, EA, Johkoh, T, Lee, KS, et al Quantification of ground-glass opacity on high-resolution CT of small peripheral adenocarcinoma of the lung: pathologic and prognostic implications. AJR Am J Roentgenol 2001;177,1417-1422[Abstract/Free Full Text]
  11. Takashima, S, Maruyama, Y, Hasegawa, M, et al Prognostic significance of high-resolution CT findings in small peripheral adenocarcinoma of the lung: a retrospective study on 64 patients. Lung Cancer 2002;36,289-295[CrossRef][ISI][Medline]
  12. Lababede, O, Meziane, MA, Rice, TW TNM staging of lung cancer: a quick reference chart. Chest 1999;115,233-235[Abstract/Free Full Text]
  13. Adler, B, Padley, S, Miller, RR, et al High-resolution CT of bronchioloalveolar carcinoma. AJR Am J Roentgenol 1992;159,275-277[Abstract/Free Full Text]
  14. Winter-Muram, HT, Jennings, SG, Tarver, RD, et al Volumetric growth rate of stage I lung cancer prior to treatment: serial CT scanning. Radiology 2002;223,798-805[Abstract/Free Full Text]
  15. Winter-Muram, HT, Jennings, SG, Meyer, CA, et al Effect of varying CT section width on volumetric measurement of lung tumors and application of compensatory equations. Radiology 2003;229,184-194[Abstract/Free Full Text]
  16. Jennings, SG, Winter-Muram, HT, Tarver, RD, et al Lung tumor growth: assessment with CT-comparison of diameter and cross-sectional area with volume measurements. Radiology 2004;231,866-871[Abstract/Free Full Text]
  17. Hanley, JA, McNeil, BJ The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143,29-36[Abstract/Free Full Text]
  18. Swensen, SJ, Silverstein, MD, Ilstrup, DM, et al The probability of malignancy in solitary pulmonary nodules: application to small radiologically intermediate nodules. Arch Intern Med 1997;157,849-855[Abstract]
  19. Lemeshow, S, Hosmer, DW A review of goodness-of-fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982;115,92-106[Abstract/Free Full Text]
  20. Wasson, JH, Sox, HC, Neff, RK, et al Clinical prediction rules: applications and methodological standards. N Engl J Med 1985;313,793-799[Abstract]
  21. Scott, JA Pulmonary perfusion patterns and pulmonary arterial pressure. Radiology 2002;224,513-518[Abstract/Free Full Text]
  22. Arana, E, Delicado, P, Marti-Bormati, L Validation procedures in radiologic diagnostic models: neural network and logistic regression. Invest Radiol 1999;34,636-642[CrossRef][ISI][Medline]
  23. Stone, M Cross-validatory choice and assessment of statistical predictions. J R Stat Soc B 1974;36,111-147
  24. Henschke, CI, McCauley, DI, Yankelevitz, DF, et al Early lung cancer action project: overall design and findings from baseline screening. Lancet 1999;354,99-105[CrossRef][ISI][Medline]
  25. Henschke, CI, Yankelevitz, DF, Mircheva, R, et al CT screening for lung cancer: frequency and significance of part-solid and nonsolid nodules. AJR Am J Roentgenol 2002;1178,1053-1057
  26. Yankelevitz, DF, Reeves, AP, Kostis, WJ, et al Small pulmonary nodules: volumetrically determined growth rates based on CT evaluation. Radiology 2000;217,251-256[Abstract/Free Full Text]
  27. Revel, MP, Lefort, C, Bissery, A, et al Pulmonary nodules: preliminary experience with three-dimensional evaluation. Radiology 2004;231,459-466[Abstract/Free Full Text]
  28. Revel, MP, Bissery, A, Bienvenu, M, et al Are two-dimensional CT measurements of small noncalcified pulmonary nodules reliable? Radiology 2004;231,453-458[Abstract/Free Full Text]
  29. Staron, RB, Ford, E Computed tomographic volumetric calcification reproducibility. Invest Radiol 1986;21,272-274[CrossRef][ISI][Medline]



This article has been cited by other articles:


Home page
 ChestHome page
T. Hashizume, K. Yamada, N. Okamoto, H. Saito, F. Oshita, Y. Kato, H. Ito, H. Nakayama, Y. Kameda, and K. Noda
Prognostic Significance of Thin-Section CT Scan Findings in Small-Sized Lung Adenocarcinoma
Chest, February 1, 2008; 133(2): 441 - 447.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tateishi, U.
Right arrow Articles by Arai, Y.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tateishi, U.
Right arrow Articles by Arai, Y.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS