| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, Graduate School of Medical Sciences,
Kumamoto University, 1-1-1 Honjo, Kumamoto-city, Kumamoto 860-8556,
Japan.
2 Nuclear Medicine Department, Hitachi Medical Corporation, Tokyo, Japan.
Received June 7, 2004;
accepted after revision March 7, 2005.
Address correspondence to S. Shiraishi.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We evaluated 156 lymph nodes (66 metastatic, 90 nonmetastatic) from 29 patients with NSCLC. Using our combined SPECT/CT system, all patients underwent 201Tl SPECT and CT examinations immediately (early images) and 3 hr after (delayed images) the injection of 201Tl. SPECT images were reconstructed with and without attenuation correction. For the quantitative evaluation of lymph node metastasis, we calculated the early ratio, the delayed ratio, and the washout ratio for SPECT images and the short-axis diameter for CT images. Receiver operating characteristic (ROC) analysis was performed in each index for the differentiation between metastatic and nonmetastatic lymph nodes. Visual analysis was also performed by two experienced radiologists.
RESULTS. The area under the ROC curve (Az) showed that early ratio and delayed ratio were superior to short-axis diameter for the assessment of lymph node metastasis. In addition, early and delayed ratios on attenuation-corrected images were superior to those ratios on images without attenuation correction. However, the Az value for washout ratio was smaller than that for short-axis diameter. Early ratio on attenuation-corrected images was the most useful index (Az = 0.94). The sensitivity, specificity, and accuracy for early ratio on attenuation-corrected images were 78.8%, 94.4%, and 87.8% for the diagnosis of lymph node metastasis and 84.6%, 100%, and 93.1% for clinical staging (N0-N1 vs N2-N3), respectively. Fused images showed significantly higher diagnostic accuracy than CT images on visual analysis.
CONCLUSION. Quantitative assessment using fused SPECT/CT images is useful for the diagnosis of lymph node metastasis in patients with NSCLC.
Keywords: cancer • CT • lung • lymph node • SPECT • staging
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Chest CT, the most commonly used noninvasive imaging method for the clinical staging of NSCLC, is of limited value for lymph node staging [4, 5]. PET with 18F-FDG is a remarkable technical advance [6]: Results of a randomized clinical trial showed that preoperative PET of NSCLC patients reduced the number of futile operations by 50% [7]. Accurately fused functional and morphologic data sets can now be generated with dual-technique PET/CT systems [8], and the use of these instruments has resulted in a significant increase in the number of patients with correct nodal staging [9].
The value of some myocardial perfusion imaging tracers such as thallium-201 [10, 11], technetium-99m sestamibi [12, 13], and 99mTc tetrofosmin [14] for mediastinal lymph node detection in patients undergoing SPECT has been recognized. SPECT/CT hybrid systems yield coregistered dual-technique images. Functional anatomic mapping allows more precise interpretation of scintigrams, and fused images can improve the diagnostic accuracy of SPECT in various clinical situations [15]. However, because CT on the SPECT/CT hybrid system is time-consuming and the quality of anatomic images is inferior to that of conventional CT systems, CT is not suitable for the depiction of small lesions such as metastatic lymph nodes.
We installed a combined SPECT/CT system that is composed of a gantry-free gamma camera and an 8-MDCT scanner. To avoid positional differences between the SPECT and CT studies, imaging can be performed on the same platform. Our system makes possible both attenuation correction and image fusion of SPECT and high-performance CT images.
In the staging of NSCLC using 201Tl SPECT, neither quantitative analyses on a per-lymph node basis using indexes such as early ratio, delayed ratio, and washout ratio, nor studies of the effectiveness of attenuation correction have been reported. Using our combined SPECT/CT system, it is possible to target the region of interest (ROI) on small lesions on SPECT images using coregistered CT images. In this study, we assessed the value of these indexes and the effect of attenuation correction on the evaluation of lymph nodes in the staging of NSCLC using fused 201Tl SPECT/CT images.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The institutional review board of the Graduate School of Medical and Pharmaceutical Sciences of Kumamoto University approved the study, as did the institutional ethics committee. Prior informed written consent was obtained from all patients after orally reviewing with them the protocol and types of examinations and after answering all questions.
System Design
We used a combined SPECT/CT system that incorporates a commercially
available gantry-free SPECT scanner with dual-head detectors (Skylight, ADAC
Laboratories) and an 8-MDCT scanner (LightSpeed Ultra, GE Healthcare). The two
instruments were juxtaposed so that the CT table holding the patient could be
moved directly into the SPECT scanner before CT. Depending on the target site,
we also used an auxiliary plate. A supporting pillar compensated for any
sagging of the table due to patient weight. As a result, each patient was
positioned identically for SPECT and CT.
SPECT
The patients were scanned with their arms elevated. Early SPECT started at
10-15 min and delayed SPECT at 3 hr after the IV administration of 111 MBq of
201Tl. SPECT data acquisition was performed with a vertex general
purpose parallel-hole collimator. A 360° SPECT scan encompassing the
thorax and upper abdomen was acquired. SPECT images were obtained at a
magnification factor of 1 or 1.46 in a 64 x 64 or 128 x 128
matrix. We acquired 64 projections at 6° intervals in the step-and-shoot
mode. Between 45 and 60 sec was required for each projection; the full study
required 24-32 min.
CT
Two sets of CT images without contrast administration were obtained in this
study. First, with the patient resting and breathing freely, helical CT images
for attenuation correction of SPECT images were obtained at 120 kV, 20 mA,
20.0-mm table feed per rotation, 4.0-sec gantry rotation time, 2.5-mm
collimation, and 5.0-mm reconstruction. CT images were reconstructed using a
standard reconstruction algorithm with a 50-cm field of view to cover both the
patient and the table.
Second, CT images for the diagnosis of lymph node metastasis were obtained at 120 kV, 180 mA, 17.5-mm table feed per rotation, 0.7-sec gantry rotation time, 2.5-mm collimation, and 5.0-mm reconstruction. For these examinations, the patients held their breath between inspiration and exhalation while in the supine position. CT images were reconstructed using a standard reconstruction algorithm with a 35-40 cm field of view of the target sites.
Image Processing
Reconstructed CT images were processed into DICOM data and then transferred
to Pegasys (ADAC Laboratories), a workstation for SPECT processing. The matrix
of the transferred CT images was converted to 128 x 128 for attenuation
correction and to 256 x 256 for image fusion. The slice thickness was
adjusted to approximately 6 mm. For SPECT reconstruction, the
maximum-likelihood expectation maximization (ML-EM) algorithm was used with 12
iterations. Butterworth filtering (cutoff, 0.30-50 cycles per pixel; order,
5-6) was applied for prereconstruction filtering.
One lumen of a three-way cock (inner diameter, 4 mm; length, 10 mm) containing an aqueous solution of 201Tl and a contrast medium was used as an external fiducial marker. To obtain precise registration of both images, external fiducial markers were fixed to the common platform for SPECT and CT, and the two scans were obtained sequentially. Image fusion of SPECT and CT images was manually performed by registration of external fiducial markers of the two images on a workstation [16]. The usefulness of external fiducial markers has been reported by several researchers [17-19].
After registration of the SPECT and CT images, a CT-derived attenuation coefficient map was created with the calibration curve for 201Tl using targeted nuclide software (HYOGO CM Attenuation Correction, Hyogo College of Medicine) [16]. The SPECT images were then reconstructed with and without attenuation correction by means of the ML-EM algorithm. In the final step, image fusion of both the reconstructed attenuation-corrected and non-attenuation-corrected SPECT and CT images was performed by reregistration of the external fiducial markers on a workstation (Fig. 1).
|
Clinical-Pathologic Correlation in Lymph Nodes
A total of 181 lymph nodes measuring more than 5 mm in diameter were
identified on the CT images of our 29 patients; 84 of these were diagnosed on
the basis of histopathologic examination with H and E staining. The other 97
lymph nodes were diagnosed by clinical follow-up studies lasting more than 1
year. According to the American Joint Committee on Cancer (AJCC) and Union
Internationale Contre le Cancer (UICC)
[20] regional lymph node
classification system, in 84 lymph nodes the site of surgical dissection
matched that identified on CT images. In patients in whom metastatic lymph
nodes were proven by surgery, the nodes were identified in the corresponding
region on CT images based on their size and shape. Of the 84 histologically
diagnosed lymph nodes, 15 were metastatic and 69 were not metastatic. Of 97
lymph nodes diagnosed on follow-up studies, those showing an obvious increase
of 5 mm or more in short-axis diameter over time were defined as metastatic
and subjected to sequential CT studies
[21]. Lymph nodes that did not
increase in size on follow-up CT performed over a period of 1 year were
defined as nonmetastatic. Lymph nodes that did not fulfill either of these
criteria (n = 25) were considered to be indeterminate and were
excluded from assessment. Of these 97 lymph nodes, 51 were defined as
metastatic and 21 as nonmetastatic. Of the 181 lymph nodes available at the
inception of this study, 156 lymph nodes (66 with, 90 without metastasis) were
selected for evaluation.
Quantitative Analysis
To compare the diagnostic value of our quantitative indexes, we used
receiver operating characteristic (ROC) analysis that intraindividually
evaluated the diagnostic performance of each index. Binormal ROC curves were
generated with an ML-EM method. ROC analysis was performed using ROC-KIT
(Windows 95 [Microsoft], version 0.9.1, BETA) developed by Charles E. Metz
(University of Chicago). The Az and its SE (SE [Az])
were calculated as a measure of the likelihood of a correct test decision
using a global decision variable
[22].
Feasible threshold values for each index were set at a false-positive fraction of 0.05 (i.e., 95% specificity) to compare the diagnostic performance at low false-positive fractions. For each index, we calculated the sensitivity, specificity, positive and negative predictive values, and accuracy in the evaluation of lymph nodes (metastatic vs nonmetastatic) and in lymph node staging (N0-N1 vs N2-N3) [20]. The relevance of each quantitative index on SPECT images was compared with the relevance of the short-axis diameter on CT images using the McNemar test on statistical software (SPSS). A p value of less than 0.05 was considered to indicate a statistically significant difference.
Visual Analysis
Two experienced radiologists who had more than 10 years' experience in both
nuclear medicine and CT evaluated conventional CT images and fused SPECT/CT
images. The reviewers were unaware of the final pathology diagnoses, but they
were informed of the clinical reasons for the scintigraphy and of the site of
the suspected primary lesion. Reviewers initially interpreted CT images. Two
weeks later, to avoid recall bias, fused images were evaluated by the same
reviewers. Attenuation-corrected early phase SPECT images were used for fused
images because this parameter showed the highest diagnostic performance (see
Results, Quantitative Analysis).
For the interpretation of the CT images, the mediastinal lymph nodes with a diameter of less than 1 cm on the shortest axis were considered normal in size and those with a diameter of 1 cm or more were considered to be metastases [14, 23].
Fused images were classified as positive for lymph node metastases when there was one or more foci of well-defined increased 201Tl uptake in the mediastinum and hilum compared with background activity. Disagreements were resolved by consensus, with a third observer as a referee [14].
The diagnostic performance on fused images was compared with that on CT images using the McNemar test on statistical software (SPSS). A p value of less than 0.05 was considered to indicate a statistically significant difference.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
Visual Analysis
The early images with attenuation correction showed higher performance than
CT with a sensitivity, specificity, positive predictive value, negative
predictive value, and accuracy of 86.4%, 90.0%, 86.4%, 90.0%, and 88.5% in the
evaluation of each lymph node metastasis. The sensitivity, specificity,
positive predictive value, negative predictive value, and accuracy were 84.6%,
93.8%, 91.7%, 88.2%, and 89.7% for lymph node staging (N0-N1 vs N2-N3),
respectively (Tables 4 and
5).
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To our knowledge, the usefulness of quantitative indexes of 201Tl SPECT images for lymph node metastasis in NSCLC has not been reported. It is difficult to accurately identify small lesions such as lymph nodes because the spatial resolution of SPECT images is poor: The smallest lymph node detected by 201Tl SPECT was 1.5 cm [34]. Therefore, different or combined methods must be used for the quantitative evaluation of small lymph nodes. We used registered CT images for the placement of ROIs on SPECT images. When the ROIs were placed on the target lesions on CT images with the aid of software, they could be placed automatically on the corresponding sites on both the coregistered attenuation-corrected and the non-attenuation-corrected SPECT images. Because it is relatively easy to set ROIs for lymph nodes, unless severe misregistration occurs between SPECT and CT images the quantitative evaluation of lymph nodes is possible. On fused images, we could easily distinguish between lymph node accumulations and physiologic accumulations of 201Tl (Figs. 3A, 3B, and 3C).
|
|
|
The retention index is useful for evaluating primary lung cancer. Arbab et al. [35] showed that the retention index on 201Tl SPECT was particularly useful for differentiating malignant and benign lung lesions, and Takekawa et al. [36] found it useful for the prediction of prognoses. Mathematically, the retention index is expressed as a negative value of washout ratio (retention index = -washout ratio). In previous evaluations of primary lung lesions, ROIs for background were usually set on the lung fields, and the washout of tracer was evaluated relative to the lung. On the other hand, we chose fat tissue as the background for quantitative indexes because lymph nodes are located in fat tissue of the mediastinum. Therefore, we used washout ratio as a diagnostic index instead of retention index.
In our study, the Az value showed that for diagnosing lymph nodes, the performance of early ratio was superior to that of delayed ratio, washout ratio, or CT (Fig. 2). Moreover, for both the evaluation of lymph nodes for metastasis and for lymph node staging, the early ratio on SPECT images yielded the highest diagnostic accuracy in individual patients. This suggests that tracer accumulation in the metastatic lymph nodes is stronger in the early than in the delayed phase, resulting in the superior performance of the early ratio. Diagnostic accuracy at all quantitative indexes was higher on attenuation-corrected SPECT than on non-attenuation-corrected SPECT images (Figs. 2, 4A, 4B, 4C, and 4D, Table 3). Attenuation correction may improve diagnostic accuracy when quantitative indexes derived from SPECT images are used for clinical evaluations. Visual analysis also showed early fused images were useful for differentiating metastatic from nonmetastatic lymph nodes, which led to better performance in staging.
|
|
|
|
Although we performed image fusion manually, we confirmed the validity of our image fusion by registering external fiducial markers attached to the platform. This is a time-consuming procedure that might limit the routine use of this technique. Improved software algorithms are needed for automatic and robust image fusion. Considerable progress has been made in the development of fusion software to coregister different imaging techniques [37, 38]. Slomka et al. [38] developed a technique for automatic nonlinear registration of CT and PET whole-body images to common spatial coordinates. This technique may be applied in the automatic fusion of SPECT with CT acquired on stand-alone scanners during normal breathing or breath-holding data acquisition.
Our study has several limitations. First, quantitative assessment was performed only in patients with 201Tl accumulation in the primary lesion, and the study of additional patients with NSCLC may yield different results. Second, our study population included a small number of patients with nodal stages N1 and N2. Third, at present our combined SPECT/CT system is not widely used because it involves increased costs and is not commercially available. Image fusion of SPECT and CT frequently must be performed during separate imaging sessions using separate scanners, a situation dictated by clinical requirements or available facilities.
In conclusion, using a combined SPECT/CT system, we showed that fusion images that combine CT-based anatomic information with SPECT-based functional images are useful for quantitative analysis. ROC multiindex analysis of fused SPECT/CT images indicated that the early ratio on 201Tl SPECT images had the highest diagnostic accuracy. We also found that attenuation correction was effective for detecting lymph node metastasis in patients with NSCLC.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
