| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Dept of Radiology (Y.O., T.H., H.W., K.S.), Div of Cardiovascular, Thoracic and Pediatric Surgery (M.Y.), and Div of Cardiovascular and Respiratory Medicine, Dept of Internal Medicine (M.S., Y.N.), Kobe Univ Graduate School of Medicine, 7–5-2 Kusunoki-cho, Chuo-ku, Kobe 650-0017, Japan; Dept of Radiology, Beth Israel Deaconess Medical Center, Boston, Mass (H.H.); Dept of Radiology, Kasai Municipal Hosp, Hyogo, Japan (D.T.); and Div of Pathology, Kobe University Hosp, Japan (C.O.). From the 2002 RSNA scientific assembly. Received Jan 15, 2003; revision requested Mar 25; final revision received Sep 2; accepted Sep 25. Address correspondence to Y.O. (e-mail: yosirad@kobe-u.ac.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
MATERIALS AND METHODS: One hundred ten patients (68 men and 42 women) with non–small cell lung cancer who ranged in age from 36 to 82 years (mean age, 64 years) were examined with respiratory-triggered STIR TSE MR imaging. Ratios of signal intensity in a lymph node to that in a 0.9% saline phantom (lymph node–saline ratios [LSRs]) for all lymph nodes were classified into three groups according to nodal short-axis diameter. LSRs of each group were compared by using pathologic diagnosis as the standard of reference. For quantitative analysis, the LSR threshold value for a positive test was determined on a per-node basis and tested for ability to enable a correct diagnosis on a per-patient basis. For qualitative analysis, signal intensities of lymph nodes were assessed by using a five-point visual scoring system. Results of quantitative and qualitative analyses were compared on a per-patient basis with McNemar testing.
RESULTS: In 110 patients, 92 of 802 lymph nodes were pathologically diagnosed as containing metastases, while 710 lymph nodes did not contain metastases. Mean LSR in the lymph node group with metastasis was higher than that in the group without metastasis (P < .05). When an LSR of 0.6 was used as the positive-test threshold at quantitative analysis, sensitivity was 93% (37 of 40 patients) and specificity was 87% (61 of 70 patients) on a per-patient basis. With a score of 4 as the positive-test threshold at qualitative analysis, sensitivity was 88% (35 of 40 patients) and specificity was 86% (60 of 70 patients) on a per-patient basis. There was no significant difference (P > .05) between results of quantitative and those of qualitative analysis.
CONCLUSION: Quantitative and qualitative analyses of STIR TSE MR images enable differentiation of lymph nodes with metastasis from those without. Qualitative analysis can substitute for quantitative analysis of STIR TSE MR imaging data.
© RSNA, 2004
Index terms: Lung neoplasms, 60.3211, 60.3212, 60.3214 • Lymphatic system, CT, 996.12912 • Lymphatic system, MR, 996.129413 • Lymphatic system, neoplasms, 996.33 • Lymphatic system, PET, 996.12963
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
Various diagnostic techniques and procedures, such as computed tomography (CT), magnetic resonance (MR) imaging, bronchoscopy, mediastinoscopy, thoracoscopy, and positron emission tomography (PET), are used for preoperative staging of lung cancer. However, CT is limited in the evaluation of nodal status because it provides only presumptive evidence of metastatic disease on the basis of size criteria (4–10). Sensitivity and specificity of CT in this regard are approximately 60%, which is certainly not optimal for clinical decision making (4–10). Despite the fact that the prevalence of metastases to mediastinal nodes is as high as 26% in this setting (4–10), many patients proceed directly to thoracotomy for primary resection and random hilar lymph node sampling. At other institutions, patients routinely undergo mediastinoscopy before thoracotomy (11–13). A more accurate noninvasive method for determining lymph node status in patients with early-stage NSCLC would be extremely useful for assigning patients to the most appropriate staging procedure.
Promising results have been reported for PET performed with fluorine 18 fluorodeoxyglucose (FDG) or carbon 11 chlorine (14–17). PET imaging has been used to differentiate lymph nodes with metastasis from lymph nodes without metastasis on the basis of the biochemical mechanisms of increased glucose metabolism and duplication of tumor cells (14–17); however, elevated glucose metabolism may be secondary to tumor, infection, or inflammation (18,19). Moreover, the diagnostic capability of FDG PET is limited because standard uptake values at FDG PET are affected by lymph node size (16).
Some investigators have discussed the utility of short inversion time inversion-recovery (STIR) MR imaging for detection of lymph nodes with metastasis in various malignant cancers (20–24). Recently, we (24) reported that the sensitivity, specificity, and accuracy of STIR MR imaging for the detection of metastases in mediastinal lymph nodes on a per-lymph-node basis were 100%, 96%, and 96%, respectively, although the total number of patients in that study was small and the signal intensity of the lymph nodes was affected by lymph node size. Therefore, in the present study, we planned to include a larger prospective cohort. The purpose of our study was to evaluate STIR turbo spin-echo (TSE) MR imaging for detection of lymph nodes with metastasis by using quantitative and qualitative analyses.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
MR Imaging Examination
MR imaging was performed with one of two 1.5-T superconducting magnets (Gyroscan PowerTrak 6000 or Gyroscan Intera; Philips Medical Systems, Best, the Netherlands) by using a body coil. In all patients, transverse electrocardiographically and respiratory-triggered STIR TSE MR images (repetition time, one to two R-R intervals; effective echo time, 15 msec; inversion time, 150 msec; echo train length, five; number of signals acquired, two; section thickness, 8 mm; section gap, 0.8 mm; matrix, 256 x 208; reconstruction matrix, 512 x 416; field of view, 320 mm) were obtained; a 0.9% saline phantom was placed alongside the chest wall at imaging of each patient. The saline phantom consisted of 60 mL of 0.9% saline within a glass tube covered by a plastic cap (Kenis, Osaka, Japan).
CT Examination
All CT examinations were performed by using a multi–detector row CT scanner, the Somatom Plus 4 Volume Zoom (Siemens, Forchheim, Germany). Scans were obtained from the lung apex to the diaphragm during suspended respiration. The instrument parameters at multi–detector row CT were 140 kV, 330 mA, 4 x 1-mm collimation, 6:1 pitch, and 5-mm section thickness reconstruction. During scanning, 100 mL of iopamidol (Iopamiron 300; Schering Japan, Osaka) was administered to patients intravenously through an antecubital vein at 2–3 mL/sec by using a power injector (Auto Enhance-50; Nemoto, Tokyo, Japan). All CT scans were obtained with a scan delay of 30–40 seconds. The CT scans were reviewed with a window width of 400–500 HU and a window level of 30–40 HU.
Lymph Node Sampling and Histopathologic Examination
Cervical mediastinoscopy (supplemented by left anterior mediastinotomy if the primary tumor was in the left upper lobe of the lung) was selectively performed if CT findings suggested invasion of the superior mediastinum by cancer or enlarged nodes (see below for criteria for enlarged nodes). Complete standard mediastinal node sampling was performed at mediastinoscopy.
At thoracotomy, ipsilateral hilar and mediastinal sampling was systematically performed by the same surgeon (M.Y.) according to the American Joint Committee on Cancer and the Union International Contre le Cancer (AJCC-UICC) regional lymph node classification system (25–27) at the following locations: On the right side, L1–L4 (ie, the highest mediastinal, upper paratracheal, prevascular and retrotracheal, and lower paratracheal lymph nodes) were dissected; posterior-to-inferior dissection around the hilum was then performed to remove L7 (a subcarinal node), L8 (a paraesophageal node), and L9 (a node near the pulmonary ligament). On the left side, dissection was begun at the aortic arch by locating the vagus nerve and its recurrent laryngeal branch and excising L4, L5 (a subaortic node), and L6 (a paraaortic node); dissection then proceeded around the hilum as on the right side.
The sites of these surgically dissected lymph nodes were matched to the lymph nodes identified at CT and MR imaging according to the AJCC-UICC regional lymph node classification system for lung cancer staging (26).
All resected lymph nodes were fixed in 10% buffered formalin and cut into 1–2 mm slices after macrosopic examination was performed by a pulmonary pathologist (C.O.). Each histologic specimen was sliced in the largest area of each section of lymph node and diagnosed as containing metastasis or not containing metastasis by the same pathologist.
Analysis of Respiratory-triggered STIR TSE MR Images
Quantitative analysis.—On STIR TSE images, all signal intensities were measured in circular or oval regions of interest drawn over each mediastinal lymph node and the phantom of 60 mL of 0.9% saline by a chest radiologist with 10 years of experience (Y.O.). The regions of interest drawn over the lymph node encompassed the entire cross-sectional area of the lymph node and saline phantom (3–18 mm in diameter).
So that we could quantitatively evaluate the signal intensity of lymph nodes at STIR TSE MR imaging, all signal intensities of lymph nodes were normalized by comparing them with the signal intensities of the 0.9% saline phantom to produce the lymph node–saline ratio (LSR). In general, signal intensity at MR imaging is relative. However, at STIR TSE MR imaging, the highest signal intensity in gray scale is that of water and the lowest signal intensity is that of fat. Therefore, the LSR in each subject represented the index of signal intensity of each lymph node as a quantitative measurement.
The LSR was determined with the formula LSR = SILN/SISP, where SILN is the signal intensity of the lymph node and SISP is the signal intensity of the saline phantom.
Qualitative analysis.—To determine the ability of qualitative analysis of STIR TSE MR images to substitute for LSRs in distinguishing lymph nodes with metastasis from those without, results of all examinations were interpreted by two chest radiologists (T.H. and D.T., who had 8 and 13 years of experience, respectively) who did not have knowledge of the surgical results. The probability that a lymph node contained metastasis was evaluated on a per-node basis by using a five-point visual scoring system in which a score of 1 meant that the signal intensity of the node could not be evaluated; a score of 2, that the signal intensity of the node was less than or equal to that of mediastinal fat; a score of 3, that the signal intensity of the node was greater than that of mediastinal fat and less than or equal to that of muscle; a score of 4, that the signal intensity of the node was greater than that of muscle and less than or equal to that of the primary lesion; and a score of 5, that the signal intensity of the node was greater than that of the primary lesion. Higher scores were considered to indicate an increased probability that metastasis was present. The final diagnosis of lymph node metastasis was rendered by both radiologists (T.H. and D.T.) working in consensus.
Diagnosis of lymph node metastasis at contrast-enhanced CT.—All CT images were evaluated according to the criteria of Webb et al (9) because, to our knowledge, there is no definite evidence that characteristics other than size (such as shape or the appearance of borders) are of value in staging lung cancer (28). Lymph nodes were considered abnormal if they were larger than 10 mm in short-axis diameter, except in the upper paratracheal stations, where the threshold size for an enlarged node was 7 mm in short-axis diameter, and in the subcarinal station, where nodes larger than 11 mm in short-axis diameter were considered enlarged (5,9). (The threshold size of 7 mm for the upper paratracheal stations was selected according to the recommendations of Glazer et al [5].)
Data and Statistical Analysis
So that we could evaluate the difference in signal intensity between lymph nodes with metastasis and those without and determine the effect of short-axis diameter on LSR, the lymph nodes were classified into the following three size groups: Size group A included lymph nodes with a short-axis diameter of less than 6 mm; size group B, lymph nodes with a short-axis diameter that was 6 mm or greater but less than 11 mm; and size group C, lymph nodes with a short-axis diameter of 11 mm or greater. The LSRs of lymph nodes with metastasis and those of nodes without metastasis were compared among short-axis diameter groups by using analysis of variance followed by Tukey honestly significant difference multiple comparison testing.
To evaluate the ability of LSRs to enable the differentiation of lymph nodes with metastasis from those without metastasis, the feasible LSR threshold of each group on a per-node basis was determined by using a receiver operating characteristic–based positive test.
To determine the observer performance at qualitative analysis of STIR TSE MR images on a per-node basis, a 
statistic was used. Because the P values were exploratory in nature, no Bonferroni correction was performed. Interobserver agreement was considered to be slight when was less than 0.21, fair when ranged from 0.21 to 0.40, moderate when ranged from 0.41 to 0.60, substantial when ranged from 0.61 to 0.80, and almost perfect when ranged from 0.81 to 1.00 (29).
The feasible threshold of qualitative analysis for distinguishing lymph nodes with metastasis from those without metastasis on a per-node basis was also determined by using a receiver operating characteristic–based positive test.
Receiver operating characteristic analysis was used to evaluate the effectiveness of the LSR and the rated signal intensity score for revealing lymph nodes with metastasis. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated for each level of LSR and signal intensity score by varying the LSR and signal intensity score that signified a positive test (ie, the threshold value) (24,30). Sensitivity was defined as the percentage of lymph nodes with metastasis that had an LSR and rated signal intensity score greater than or equal to the given threshold level, and specificity was defined as the percentage of lymph nodes without metastasis that had an LSR and signal intensity score less than the threshold level.
Feasible threshold values at quantitative and qualitative analyses of STIR TSE MR images were tested for their ability to enable lymph nodes with metastasis to be distinguished from lymph nodes without metastasis on a per-patient basis. Overlapping lymph nodes evaluated with the feasible threshold value in each size group were also pathologically diagnosed by the same pulmonary pathologist (C.O.) by using hematoxylin-eosin staining. In the group of lymph nodes with metastasis, overlapping lymph nodes were considered to be present when LSRs were lower than feasible threshold values in each size group. In the group of lymph nodes without metastasis, overlapping lymph nodes were considered to be present when LSRs were greater than or equal to the feasible threshold values in each size group.
The abilities of quantitative and qualitative analyses of STIR TSE MR images to enable a correct diagnosis were compared with the ability of contrast-enhanced CT on a per-patient basis by using the McNemar test.
For all statistical analyses, a P value of less than .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
|
|
|
11 mm) (P < .05). In lymph nodes without metastasis, results of analysis of variance revealed significant differences among the mean LSRs of each size group (F = 48.72, P < .001). The mean LSR for nodes in size group A was significantly different from that for nodes in the other two size groups (P < .05), and the mean LSR for size group B nodes (ie, those with a short-axis diameter of 6 mm or greater but less than 11 mm) was significantly different from that for size group C nodes (P < .05).
Results with the receiver operating characteristic–based positive test for each size group on a per-node basis are shown in Figure 3. An LSR of 0.6 was adopted as the threshold for a positive test (ie, an LSR of 0.6 or greater indicated that a lymph node contained metastasis) for all size groups. For size group A lymph nodes, the sensitivity and specificity for differentiating lymph nodes with metastasis from those without metastasis by using the threshold LSR of 0.6 were 76% (13 of 17 nodes) and 99% (223 of 226 nodes), respectively. The LSRs of four (24%) of 17 lymph nodes with metastasis were overlapped with those of lymph nodes without metastasis (ie, the LSRs of four lymph nodes were less than 0.6). These overlapped lymph nodes had micrometastasis within anthracosilicotic, silicotic, or fibrotic nodes. The LSRs of three (1%) of 226 lymph nodes without metastasis were overlapped with those of lymph nodes with metastasis (ie, the LSRs of three lymph nodes were 
0.6). These overlapped lymph nodes were normal lymph nodes whose LSRs affected the artifact due to cardiac motion.
|
For size group C lymph nodes, the sensitivity and specificity for differentiating lymph nodes with metastasis from those without metastasis by using the threshold LSR of 0.6 were 100% (38 of 38 nodes) and 93% (70 of 75 nodes), respectively. Five (7%) of 75 size group C lymph nodes without metastasis overlapped lymph nodes with metastasis. These overlapped lymph nodes were lymph nodes with inflammation.
Results of visual scoring of nodal signal intensity by the two readers are shown in Table 2. Interobserver agreement at qualitative analysis of STIR TSE MR imaging data was substantial ( = 0.77). Results with the receiver operating characteristic–based positive test are shown in Figure 4. At qualitative analysis, a rated signal intensity score of 4 was adopted as the threshold for a positive test (ie, a score greater than or equal to 4 indicated that a lymph node contained metastasis). Only eight lymph nodes with micrometastases were evaluated; each of these nodes had a signal intensity less than or equal to that of muscle. Twelve lymph nodes with inflammation were evaluated; each of these nodes had a signal intensity greater than that of muscle.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
In the present study, our results show that mean LSR of lymph nodes with metastasis was significantly higher than that of lymph nodes without metastasis in every lymph node size group. At STIR TSE MR imaging, inversion time is approximately 80–150 msec. At these inversion times, longitudinal magnetization for virtually all tissues is negative when a 90° pulse is applied. Recovery is just beginning for most tissues. After the second 90° pulse, the T1 contrast and the T2 contrast are additive; that is, increasing the T1 of a tissue increases the tissue’s relative signal intensity, and so does increasing its T2. Some investigators have shown that there are significant differences between malignant and benign nodes in terms of their T1 and T2 relaxation times (31–34). Because many pathologic lesions show an increase in both T1 and T2, the addition of these two types of contrast with the STIR sequence produces a higher net tissue contrast (35). Therefore, significant differences between lymph nodes with metastasis and lymph nodes without metastasis were observed in the present study in each size group.
On the other hand, the mean LSR for lymph nodes with metastasis in size group A was significantly lower than that for lymph nodes with metastasis in size group C. In addition, there were significant differences in the mean LSRs for lymph nodes without metastasis among the three size groups. When 0.6 was adopted as the feasible LSR threshold value, LSRs of six (7%) of 92 lymph nodes with metastasis were overlapped with those of lymph nodes without metastasis. These overlapped lymph nodes had micrometastasis within anthracosilicotic, silicotic, fibrotic, and normal nodes. LSRs of 15 (2%) of 710 lymph nodes without metastasis were overlapped with those of lymph nodes with metastasis. These overlapped nodes were lymph nodes with inflammation and normal lymph nodes whose LSRs affected the artifact due to cardiac motion. Previous studies have revealed a significant correlation between the signal intensity of a lymph node and its short-axis diameter (24). Therefore, these results suggest that lymph node size and the degree of such pathologic changes as metastatic cell nests, inflammation, anthracosilicosis, calcification, silicosis, fibrosis, and coagulation necrosis within a lymph node affect the changes in T1 and T2 relaxation times and result in some overlap between lymph nodes with metastasis and those without.
In this study, at qualitative analysis of STIR TSE MR imaging data, interobserver agreement was substantial (>0.61). In addition, on the basis of the results with the receiver operating characteristic–based positive test, the feasible signal intensity score threshold value was determined to be 4 (ie, signal intensity greater than that of muscle and less than or equal to that of the primary lesion). According to reports in the literature, T1 and T2 relaxation times of muscle are shorter than those of lymph nodes with metastasis, although inflammation prolongs T1 and T2 relaxation times and appears as areas of high signal intensity similar to the signal intensity of malignant tumor (33). In addition, with the STIR TSE sequence, when magnitude reconstruction with an inversion time in the range of 80 to 150 msec is used, the fat tissue signal is nulled or suppressed (35). Suppression of the fat signal is useful for detecting small lymph nodes because mediastinal lymph nodes reside within mediastinal fat tissue (24,36). Therefore, STIR imaging makes it possible to detect small lymph nodes in patients with lung cancer. When interobserver agreement and the relaxation time analysis are considered, qualitative analysis of STIR TSE MR images may enable the diagnosis of metastasis in lymph nodes about as well as quantitative analysis of STIR TSE MR images.
At comparison of the results of contrast-enhanced CT and quantitative and qualitative analyses of STIR TSE MR images on a per-patient basis, we found that for distinguishing lymph nodes with metastasis from those without metastasis, both methods of MR image analysis had significantly higher sensitivity and accuracy than did contrast-enhanced CT. In addition, there were no significant differences between the results of quantitative and those of qualitative analysis of STIR TSE MR images. Therefore, on a per-patient basis, both quantitative analysis and qualitative analysis of STIR TSE MR images can enable evaluation of metastasis to mediastinal and hilar lymph nodes more accurately than can contrast-enhanced CT. In addition, qualitative analysis of STIR TSE MR images can substitute for quantitative analysis of STIR TSE MR images for clinical purposes.
Results of a number of studies in which the accuracy and utility of CT for diagnosing mediastinal node involvement by lung cancer were evaluated indicate that the sensitivity values of CT range from 52% to 79%; the specificity values, from 60% to 69%; and the accuracy values, from 61% to 84% (4–10). Researchers who evaluated the accuracy of conventional T1- and T2-weighted MR imaging for diagnosing mediastinal node involvement by lung cancer have reported sensitivities ranging between 48% and 93%, specificities ranging between 64% and 91%, and accuracies ranging between 61% and 84% (7–9). In previous reports of results for differentiating lymph nodes with metastasis from lymph nodes without metastasis by using FDG PET, sensitivities, specificities, and accuracies ranged from 67% to 80%, from 97% to 100%, and from 87.5% to 88%, respectively (4–17). Therefore, quantitative and qualitative analyses of STIR TSE MR images enable differentiation of lymph nodes with metastasis from lymph nodes without metastasis with sensitivity values that are greater than or equal to those of FDG PET.
There were limitations to our study. First, although we histologically examined all lymph nodes at every 1 or 2 mm of section thickness by using hematoxylin-eosin staining, some microscopic metastatic nodes may have been missed (37,38). Second, we performed radiologic-pathologic correlation of 802 resected lymph nodes from 110 patients with CT and MR imaging but not with FDG PET. Therefore, a larger prospective comparative study involving FDG PET, as well as a cost analysis, would be required to determine the true value of STIR TSE MR imaging for the diagnosis of metastasis in lymph nodes. A prospective study in which CT, FDG PET, and STIR TSE MR imaging are compared on the basis of our results in the present study is planned for the near future.
In conclusion, both quantitative analysis and qualitative analysis of STIR TSE MR images enable lymph nodes with metastasis to be differentiated from those without. Qualitative analysis can be substituted for quantitative analysis of STIR TSE MR imaging data.
|
STATISTICAL CONSULTANT COMMENTARY |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
11 mm in short-axis diameter), the authors performed a one-way analysis of variance followed by the Tukey honestly significant difference multiple comparison test. The response variable here is LSR, the factor that may have an effect on the response variable is short-axis diameter, and there are three levels of this factor. Thus, analysis of variance "tests the hypothesis that means of all levels are equal" (Choi SC. Introductory applied statistics in science. Englewood Cliffs, NJ: Prentice-Hall, 1978). If the resultant F statistic is statistically significant, and "the hypothesis of equal means is rejected by the analysis of variance in a one-way classification" (Choi SC. Introductory applied statistics in science), we would then like to know where these differences in means occur. To do this, we employ a multiple comparison test, a post hoc test that examines differences between any two levels. Among the more widely used multiple comparison tests are the Fisher least significant difference test, the Sheffe test, and the Tukey honestly significant difference test. While the Tukey honestly significant difference test is used to look for reasonable similarity in the numbers of observations in each level, this test can be appropriate in this instance. Please note that the authors first present the results that the analysis of variance tests revealed statistically significant differences, then follow with the results of the Tukey honestly significant difference multiple comparison tests. Specifically, the authors report that, for lymph nodes with metastasis, results of analysis of variance were significant, and the LSR for size group A nodes was significantly different from the LSR for size group C nodes. Also, for lymph nodes without metastasis, results of the analysis of variance test were statistically significant: The LSR for size group A nodes was significantly different from the LSR for size group B and size group C nodes, and the LSR for size group B nodes was significantly different from the LSR for size group C nodes. This is a useful manner in which to present these results.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, Y.O.; study concepts, Y.O., H.H.; study design, Y.O.; literature research, Y.O., D.T.; clinical studies, Y.O., D.T., T.H., H.W., C.O., M.Y., M.S., Y.N., K.S.; data acquisition, Y.O., D.T., T.H.; data analysis/interpretation, Y.O.; statistical analysis, Y.O.; manuscript preparation, Y.O.; manuscript definition of intellectual content, editing, and final version approval, Y.O., H.H.; manuscript revision/review, Y.O., H.H., K.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Y. Kim, C. A Yi, K. S. Lee, M. J. Chung, Y. K. Kim, B. K. Choi, H. Kim, and O J. Kwon Nodal Metastasis in Non Small Cell Lung Cancer: Accuracy of 3.0-T MR Imaging Radiology, December 4, 2007; (2007) 2461061907. [Abstract] [Full Text]
|
|
|
|
 |
|
 C. A Yi, T. Y. Jeon, K. S. Lee, J. H. Lee, J. B. Seo, Y. K. Kim, and M. J. Chung 3-T MRI: Usefulness for Evaluating Primary Lung Cancer and Small Nodules in Lobes Not Containing Primary Tumors Am. J. Roentgenol., August 1, 2007; 189(2): 386 - 392. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Kawai, M. Sumi, and T. Nakamura Turbo Short {tau} Inversion Recovery Imaging for Metastatic Node Screening in Patients with Head and Neck Cancer AJNR Am. J. Neuroradiol., June 1, 2006; 27(6): 1283 - 1287. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. K. Shankar and D. C. Sullivan Functional Imaging in Lung Cancer J. Clin. Oncol., May 10, 2005; 23(14): 3203 - 3211. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= sensitivity for size group A nodes, 
= specificity for size group A nodes, 
= sensitivity for size group B nodes, 
= specificity for size group B nodes, 
= sensitivity for size group C nodes, 
= specificity for size group C nodes. An LSR of 0.6 was adopted as the threshold for a positive test for all size groups.