| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Departments of Radiology (H.T.W.M., S.G.J., C.A.M., Y.L., A.M.A., R.D.T.) and Radiation Oncology (R.C.M.), Indiana University School of Medicine, Indianapolis. Received July 11, 2002; revision requested August 29; final revision received February 11, 2003; accepted March 3. Address correspondence to H.T.W.M., 11224 Clarkston Rd, Zionsville, IN 46077 (e-mail: hwinermu@iupui.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: Spherical simulated tumors of varying diameters were imaged with varying CT section widths, the images were displayed on a workstation, the cross-sectional area of the tumor on each section was measured by using elliptical and perimeter methods, and the areas were integrated to compute tumor volume. The actual and measured tumor volumes for differing section widths and tumor diameters were compared, and compensatory equations were derived. The equations were applied to contemporaneous chest CT images obtained in patients with stage I lung cancer, and the difference between thick- and thin-section–derived volumes before and after application of the equations was determined.
RESULTS: All simulated tumor volumes were overestimated 11%–278%; overestimation varied directly with section width and inversely with tumor diameter. With both measurement methods, mean thin-section volumes of clinical tumors in 55 patients were significantly smaller (P < .01) than mean thick-section volumes: Mean elliptical measurements were 15,025 mm3 (thin) and 18,037 mm3 (thick), with a 20.0% difference; mean perimeter measurements were 16,164 mm3 (thin) and 20,718 mm3 (thick), with a 22.2% difference. The thin-section–to–thick-section volume difference was larger for the smallest tumors. Thin-section volumes were smaller than thick-section volumes in 53 patients with the elliptical method and in 51 patients with the perimeter method. Applying the equations decreased the difference between thick- and thin-section volumes in 37 (67%) of the 55 patients with the elliptical method and in 41 (74%) patients with the perimeter method. The mean thin-section–to–thick-section volume difference became nonsignificant with the perimeter method but remained significant with the elliptical method.
CONCLUSION: Measured lung tumor volumes vary significantly with varying CT section width; overestimation varies directly with section width and inversely with tumor size. Compensatory equations that are somewhat effective in reducing these effects can be derived.
© RSNA, 2003
Index terms: Computed tomography (CT), experimental studies, 60.12111, 60.12115 • Lung neoplasms, 60.32 • Lung neoplasms, CT, 60.12111, 60.12115 • Thorax, CT, 60.12111, 60.12115
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
The amount of volume averaging may also be affected by changes in the section width used and/or in the tumor size between examinations. At our facility, all patients suspected of having lung cancer are imaged with the same CT scanner, exposure parameters, and pitch and reconstruction algorithm, and the images are displayed at the same window and level settings. However, these patients are often imaged with different section widths and/or section increments. The initial CT examination is usually performed with a thick section width (8–10 mm) to minimize scanning time and thus allow imaging of the entire chest during a single breath hold. In patients with a nodule that is indeterminate for cancer at the initial CT examination, a follow-up examination of the nodule that includes thin sections (2–3 mm) may be performed to evaluate for calcification.
If the nodule remains indeterminate, serial follow-up CT examinations to determine whether the nodule is growing or otherwise changing in appearance may be performed with thick and/or thin section widths. In addition, CT examinations for fine-needle aspiration biopsy or radiation therapy planning may be performed with intermediate section widths (3–5 mm).
To our knowledge, there are no published reports that attempt to quantify or predict the effects of varying section width and tumor size on volumetric measurement of lung tumors of various sizes viewed with standard lung windows. Thus, the purpose of our study was to determine how volume measurements of simulated and clinical lung tumors at standard CT lung window and level settings vary with varying section width and to derive and apply compensatory equations.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Simulated Lung Tumors
Eleven nylon thermoplastic spheres (McMaster-Carr, Elmhurst, Ill) were used in the study: Six spheres were 12.7 mm (
inch), three were 25.4 mm (1 inch), and two were 38.1 mm (1 inches) in diameter. These diameters were chosen as being representative of the range of sizes of stage I lung cancers. These spheres are slightly denser than actual lung tumors (1.10 g/cm3 vs 1.05 g/cm3; attenuation, 100 HU vs approximately 40 HU). However, the combination of the typical noise level for our scanner (3–5 HU) and the use of a very wide window setting (2,000 HU) should result in a minimal visible difference between 40- and 100-HU objects.
The calculated volumes of the simulated tumors were 1,073 mm3 (12.7-mm diameter), 8,580 mm3 (25.4-mm diameter), and 28,958 mm3 (38.1-mm diameter). Manufacturer-reported volume tolerances were ±0.3%. The 1,073-mm3 simulated tumors were embedded in a staggered pattern in a 38.0 x 33.0 x 3.8-cm polystyrene plastic block; the 8,580- and 28,958-mm3 simulated tumors were embedded in a 9.6 x 7.0 x 7.0-cm polystyrene plastic block. Each block was then imaged by using the imaging parameters listed earlier with the following combinations of section width and section increment, respectively: 10 and 10 mm, 8 and 6 mm, 5 and 5 mm, 3 and 3 mm, and 2 and 2 mm. These values were chosen because they either have been or are being used regularly to image clinical lung nodules with the scanner that we used. A total of 10 CT examinations (two blocks, five combinations each) of the simulated tumors were performed.
Each CT image was downloaded from the archives to a video display system (Barcoview, Kortrijk, Belgium) with 2,500 x 2,000-pixel monitors. All images were displayed at our standard lung settings (window width, 2,000 HU; window level, -700 HU) with x2 magnification. Image viewing and manipulation were controlled by using computer software (RadWorks 5.1; Applicare Medical Imaging, Zeist, the Netherlands). This software allows the reader to draw lines through and perimeters around regions of interest. The software then automatically calculates the line/perimeter length and the area enclosed by a perimeter.
To reduce bias, the following procedure was used to display the images: Each CT image was displayed twice, in random order, for one board-certified chest radiologist (H.T.W.M.), who is experienced in using the image viewing and manipulation software. On each image section that depicted one or more simulated tumors, the radiologist was initially directed to use either the elliptical measurement method—that is, measure the major axis of the simulated tumor and its in-plane perpendicular (ie, minor) axis—or the perimeter measurement method—that is, draw a line around the perimeter of the simulated tumor. After all images had been viewed, the radiologist viewed them again, and the measurement method (ie, elliptical or perimeter) not used in the first viewing was applied. Thus, on each section, the dimensions of each simulated tumor were measured once with each method.
The details of our volumetric calculation methods have been published previously (10). In brief, for each simulated tumor, (a) the areas defined by the major and minor axes were calculated on each section, summed, and multiplied by the section increment (elliptical method), and (b) the areas inscribed by the perimeter were calculated on each section by using the RadWorks software, summed, and multiplied by the section increment (perimeter method).
For each simulated tumor diameter (six tumors 12.7 mm [1,073 mm3], three 25.4 mm [8,580 mm3], two 38.1 mm [28,958 mm3]), the volume mean and range were then determined for each section width–section increment combination and each measurement method. These results were then compared (by S.G.J.) with the known volumes of the simulated tumors, and the differences were used to derive error-reduction equations for each measurement method to compensate for errors associated with varying section widths and tumor sizes.
Clinical Lung Tumors
From our tumor registry (Richard L. Roudebush VA Medical Center, Indianapolis, Ind), we obtained a list of patients who had received a diagnosis of stage I lung cancer and were being followed up at our institution from February 1996 (the initiation of digital image archiving) to August 2001. The medical records of these patients were reviewed by one of the authors (S.G.J.). The patient information documented included age at diagnosis, date of diagnosis, cancer location and stage, and date treatment was initiated. Records were also reviewed for other pertinent data, including method of lung cancer diagnosis (eg, fine-needle aspiration biopsy, sputum cytology, or bronchoscopy), comorbidities, and smoking history.
Any patient who had previously received a diagnosis of primary lung cancer or cancer metastasis to the lung within the preceding 5 years was excluded. For each remaining patient, we (S.G.J., H.T.W.M.) together identified all chest CT examinations that were performed prior to the initiation of treatment and were available in our digital image archives. The examination date, nominal section width (in millimeters), and section increment (in millimeters) were noted. CT examinations that were performed with the patient in the prone position were excluded from the study, because prone positioning may influence the measured volume of the tumor. CT examinations were also excluded from the study if the spiral technique was not used, the examination was not performed during a single breath hold, the entire tumor was not imaged, or the entire tumor margin could not be visualized.
For each patient, one of the authors (S.G.J.) identified all remaining sequential examinations that (a) varied in section width by a ratio of at least 2:1 (thick and thin section widths) and (b) were performed fewer than 25 days apart. For example, sequential examinations performed with 10- and 5-mm section widths might have qualified for study entry, but those obtained with 5- and 3-mm section widths would not. The maximum time between examinations was based on previous work (10–12) in which the results indicated that for a tumor growing at the mean rate for all lung tumors, the growth between scans obtained fewer than 25 days apart was unlikely to introduce error into the volume measurements. Therefore, for measurement purposes, sequential examinations performed fewer than 25 days apart were considered to be contemporaneous. For each patient, the first pair of sequential chest CT examinations that met all of the above criteria formed the study set.
Each CT image was downloaded from the archives to a video display system, as previously described herein, and displayed at our standard lung settings without magnification. On each CT section depicting tumor, a radiologist (H.T.W.M.) measured the tumor once with each method (elliptical and perimeter) by using the same procedure that was used for the simulated tumors, and the volume was calculated. The maximum cross-sectional area for each tumor was noted as well. For each tumor, "roundness" was calculated by using the average result of the following formula: roundness = 4
A/L2, where for each section depicting tumor, A is the cross-sectional area (with perimeter method) and L is the measured perimeter length (13). Tumor irregularity was calculated by using the average result of the following formula: irregularity = L/P, where L is the measured perimeter length and P is the perimeter length of an ideally elliptical cross section, as defined by the Maertens and Rousseau formula (14):
|
Statistical Analysis
The mean tumor volumes calculated by using thick and thin section widths were compared by using the two-tailed t test. Pearson correlation coefficients were computed to compare the percentage difference in measured volume between thick and thin section widths and to assess (a) the ratio of thick to thin section width, (b) tumor roundness, and (c) tumor irregularity (with perimeter method only; by definition, irregularity equals 1 with the elliptical method).
After the error-reduction equations were applied to the clinical tumor data, the difference in mean thick- and thin-section–derived volumes was evaluated by using the paired two-tailed t test. For each method, the number of patients in whom the percentage difference between the thick- and thin-section–derived volumes decreased after the equations were applied was also noted. For all comparisons, P < .05 was considered to indicate a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
|
|
|
|
Images of the tumors were acquired with the following thick and thin section widths, respectively: 10 and 2 mm in nine patients, 10 and 3 mm in 22 patients, 10 and 5 mm in eight patients, 8 and 2 mm in 14 patients, 8 and 3 mm in one patient, and 5 and 2 mm in one patient. The section width and section increment were identical for all CT examinations except those performed with a section width of 8 mm and a section increment of 6 mm. The time between the acquisition of sequential CT images ranged from 0 to 24 days; the images were acquired on the same day in 45 (82%) of the 55 patients.
With use of the elliptical method of tumor volume measurement, the measured clinical tumor volumes ranged from 764 to 187,179 mm3 (median, 6,893 mm3) for the thick CT sections and from 391 to 179,450 mm3 (median, 4,578 mm3) for the thin sections (Table 3). The mean measured volume was 18,037 mm3 ± 32,018 (SD) for thick sections and 15,025 mm3 ± 29,883 for thin sections; the difference (20.0%) in volume was significant (P < .01). Fifty-three of the 55 clinical tumors appeared smaller on the thin sections. For the smallest clinical tumors (volume < 2,000 mm3 on thin sections [n = 13]), the mean volume difference was 35.9%. The largest difference was observed with the smallest clinical tumor: 3,391 mm3 on the thick section and 391 mm3 on the thin section.
|
With the elliptical method, the mean maximum cross-sectional area was 573 mm2 for thick sections and 600 mm2 for thin sections. The maximum cross-sectional area was greater on the thin sections in 34 (62%) of the 55 patients. With the perimeter method, the mean maximum cross-sectional area was 615 mm2 for thick sections and 627 mm2 for thin sections; the maximum cross-sectional area was larger on the thin sections in 30 (54%) patients.
The mean volume differences between the different thick and thin section width pairs are listed in Table 4. There was no significant association between the thick-section-width–to–thin-section-width ratio and the percentage volume difference with use of either the elliptical (R = .170, P = .21) or perimeter (R = .155, P = .25) method. Tumor roundness ranged from 0.42 to 0.93 (mean, 0.73) for thick sections and from 0.43 to 0.89 (mean, 0.71) for thin sections. There was no significant association between the thick-section–to–thin-section volume difference and either the tumor roundness on the thick (R = .050, P = .72) and thin (R = -.065, P = .64) sections with use of the elliptical method or the tumor roundness on the thick (R = .124, P = .37) and thin (R = -.239, P = .08) sections with use of the perimeter method. Tumor irregularity ranged from 1.05 to 1.93 (mean, 1.23) for thick sections and from 1.01 to 2.04 (mean, 1.27) for thin sections. There was a significant association between the thick-section–to–thin-section volume difference and the tumor irregularity on the thin sections (R = .273, P = .04) but not on the thick sections (R = -.172, P = .21) with use of the perimeter method.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
We found evidence that the partial volume effect accounts for at least some of the described results. A voxel that actually contains both tumor and normal lung parenchyma may be interpreted as containing only tumor and thus lead to an overestimation of the tumor volume (Fig 2). Because large attenuation differences between cancerous and normal tissue may accentuate this effect in the lung, a tumor occupying only a tiny portion of a voxel may be thought to be filling the entire voxel (15). The partial volume effect increases as the surface-to-interior voxel ratio increases. Therefore, this effect should vary directly with section width. This effect should also vary inversely with tumor size. The ratio of partially to completely filled voxels increases with decreasing tumor size, and each partially filled voxel that is interpreted as positive for tumor results in volume overestimation (Fig 3). We found that for the smallest clinical tumors (volume < 2,000 mm3), there was a substantially greater difference between the thick- and thin-section volumes, as compared with this difference for the larger tumors.
|
|
|
Changing the section width may lead to a greater difference in partial volume effects in "horizontal" objects (ie, those longer on the x and y axes than on the z axis) than in "vertical" objects (ie, those longer on the z axis) (8). With almost all CT scanners, voxels are anisotropic; the dimension of a voxel on the z axis is much greater than the dimension in the x,y plane. Thus, the largest errors in cross-sectional area measurement should occur in the sections at the top and bottom edges of the tumor, where partial volume effects are the greatest. Because the data used for the reconstruction algorithm were not saved at our institution, however, three-dimensional reconstructions were not available in the present study, and, thus, we could not determine whether there was a smaller difference in partial volume effects between section widths for the vertical tumors.
The use of a pitch greater than 1 (we used a pitch of 1.5) also may introduce a type of partial volume effect error owing to broadening of the nominal section width. With a constant pitch of greater than 1, volume overestimation increases as section width increases and/or tumor size decreases. We do not know how our use of a pitch greater than 1 affected the data; one of the authors (Y.L.) is currently investigating this topic.
In addition to partial volume effects, there are other potential sources of measurement error. Misregistration artifacts may result from calibration error in the CT reconstruction algorithm or from patient-related factors such as respiratory or cardiac motion (2). Varying CT acquisition and/or display parameters (eg, kilovolt peak, milliampere second, window width and level, and viewing workstation features) may also affect the apparent tumor size (2,5,7,8). These parameters should be specified and kept constant between serial CT examinations.
Even with relatively large tumors, varying the window level from 600 to -600 HU alters the measured volume by approximately 50% (5,8). A structure is most accurately represented when the window level is midway between the CT numbers of the structure of interest and the background (2). If the window level is less than midway between the background (eg, normal lung parenchyma) and the target structure (eg, cancerous lung tissue), the tumor margin will "blossom" and result in volume overestimation, regardless of the section width used (5,7). However, a low window level is commonly used in clinical settings because it helps to optimize the detection of small nodules. We displayed the CT images at a low window level (-700 HU) to ensure that all tumors would be visible. Reduced contrast between normal and cancerous lung tissue might have caused some of the tumors to "disappear" (ie, not be depicted) if a theoretically optimal level for accurate measurement had been used, but we did not assess whether this actually would have occurred in our study patients (6,15).
We are unaware of the existence of other clinical studies to assess how varying section width affects the CT volumetric measurement of clinical lung tumors. However, in a study (8) in which phantoms were used, all tumor volumes were overestimated. This overestimation decreased from 10.7% to 0.7% as the section width was decreased from 10.0 to 1.5 mm. A comparison between that study and ours is difficult, however: The phantom nodules in the other study were substantially larger (27,000 mm3) than the clinical lung tumors in our study (median volume, approximately 5,000 mm3) and were surrounded by water instead of air. Moreover, the display parameters used in the other study were not described.
According to Van Hoe et al (5), for clinical tumors that are similar in size to those that we studied, the range of measured volume–to–actual volume ratios should be quite large: 0.2 (window level, 600 HU) to 1.4 (window level, -600 HU) with an 8-mm section width. Because we used a lower window level (-700 HU), we expected the ratio to be even greater than 1.4 in our study; with the perimeter method, we observed a ratio of median thick-section (7,504 mm3) to thin-section (4,639 mm3) derived volume of 1.6. Even thin-section volumes are overestimated, so the ratio of median thick-section–derived volume to actual volume must be even greater.
We have often encountered thoracic radiologists who believe that tumor volumes measured with thin sections are usually larger than those measured with thick sections. With thick sections, there is greater tumor margin blurring, which may cause the tumor cross section to appear smaller—that is, a negative partial volume effect. In most of our study patients—in 62% of patients with the elliptical method and in 54% with the perimeter method—the maximum cross-sectional area of the tumor was larger in the thin-section CT examinations, whereas the volume was almost always smaller.
For the simulated lung tumors in our study, the measured volume varied directly with the section width and inversely with the actual volume in a nearly linear manner. Thus, we expected the compensatory equations to be both easy to derive and effective. Yet when the equations were applied to contemporaneous CT examinations of clinical tumors that were performed with different section widths, they were less effective than expected. Although the percentage difference between the thick- and thin-section–derived volumes decreased in most patients, a significant difference remained with use of the elliptical method (P = .04); the difference with the perimeter method approached significance (P = .09). With both measurement methods, application of the equations led to "overcorrected" differences between the thick- and thin-section–derived volumes. Although this was unexpected, it has been shown that among phantoms with the same diameter, cylindrical phantoms that are oriented on the z axis exhibit fewer partial volume effects than spherical phantoms (16). Perhaps many of our clinical tumors were oriented on the z axis.
In the compensatory equations, the values for K (a constant to be derived from the simulated tumor data) are dependent on the window and level settings. Moreover, K may also vary according to the CT scanner used and the display system. Although multiple readers were not used in this study, we appreciate that K may vary between readers as well. Thus, we are uncertain how applicable our K values are to other combinations of CT scanners and display systems. Yet our method for deriving compensatory equations should be universally applicable.
Although deriving these equations is not easy—it requires a few hours for phantom measurement and equation derivation—and these equations are not entirely effective, the actual tumor growth may be underestimated if the equations are not applied. For example, our results show that if an equation is not applied, a clinical tumor that is similar in size to our largest simulated tumor (volume of 28,958 mm3 [38.1 mm in diameter], equivalent to a small stage IB tumor) and has a doubling time of 200 days will appear to have the same volume on both an 8-mm section width CT scan obtained initially and the 2-mm section width CT scan obtained 94 days later when the perimeter method is used. For tumors that are smaller and/or are growing more slowly, this underestimation will be even greater.
Our study data demonstrate that even if the same section width is used for serial CT examinations, tumor size will affect the volume magnification. Without application of a compensatory equation, tumor growth will be underestimated because small tumors are magnified more than large tumors. For example, with an 8-mm section width, a tumor that increases from 1,000 mm3 to 2,000 mm3 in 1 year has an apparent volume increase of 100% and a doubling time of 365 days. However, when the equation is applied (with the perimeter method), the revised volumes are 321 mm3 initially and 746 mm3 1 year later; the revised volume increase is 132%, and the revised doubling time is 299 days. Thus, although measurement errors caused by changing tumor size are somewhat less severe than those caused by changing section width, they are not unimportant.
A limitation of this study was that we did not assess interobserver variability—there was only one reader—and only partially evaluated intraobserver variability—we obtained multiple measurements of the simulated lung tumors but only two (elliptical and perimeter) measurements of the clinical lung tumors. At our institution, it was logistically impossible for multiple readers to measure the sections at one workstation at one site. We are not aware of any studies in which the variability of repeated manual volume measurements of lung tumors has been evaluated. For repeated area measurements of body organs that were similar in size to the clinical lung tumors measured in the present study, Staron and Ford (17) found the intraobserver coefficient of variation to be 4%–7%. When measuring tumor diameters, Hopper et al (18) found interobserver variability to be less than 15%.
Automated three-dimensional volumetric measurement was not used in this study. Automated systems may enhance measurement reproducibility, and use of the reduced section widths that are available with the newer CT scanners will result in less volume overestimation. Application of compensatory equations will still be needed when knowledge of actual tumor volumes is necessary—for example, in cancer screening, radiation therapy planning, and evaluation of response to therapy. Our study results indicate that even with our largest simulated tumor, the use of a 1-mm section width would result in a 7%–8% overestimation of tumor volume.
In conclusion, our study results show that the volumes of both simulated lung tumors and clinical stage I lung tumors are increasingly overestimated with increasing CT section width and decreasing tumor size. Although compensatory equations that are somewhat effective in helping to reduce overestimation can be derived, it is best to compare tumor volumes on serial CT images with the same section width, if possible.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
K was approximated by using a reiteration technique in which the magnification was assumed to vary between the actual simulated tumor volumes and the section widths, as just described. By using each measured magnification value in Table 1, a mean magnification value for each combination of measurement method, section width, and simulated tumor volume was determined. (See example with the elliptical method in Table A1.) With each measurement method, a mean magnification value for each section width was then determined. (See example in Table A2.) To determine K for each measurement method, the corresponding mean magnification value for a 10-mm section width and a 1,073-mm3 actual simulated tumor volume was substituted into the magnification equation above and solved for K.
|
|
|
|
|
|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Zhang, S. K. Kang, L. Wang, A. Touijer, and H. Hricak Distribution of Renal Tumor Growth Rates Determined by Using Serial Volumetric CT Measurements Radiology, January 1, 2009; 250(1): 137 - 144. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. G. Ravenel, W. M. Leue, P. J. Nietert, J. V. Miller, K. K. Taylor, and G. A. Silvestri Pulmonary Nodule Volume: Effects of Reconstruction Parameters on Automated Measurements--A Phantom Study Radiology, May 1, 2008; 247(2): 400 - 408. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Suzuki, H. Jacobsson, T. Hatschek, M. R. Torkzad, K. Boden, Y. Eriksson-Alm, E. Berg, H. Fujii, A. Kubo, and L. Blomqvist Radiologic Measurements of Tumor Response to Treatment: Practical Approaches and Limitations RadioGraphics, March 1, 2008; 28(2): 329 - 344. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. K. Gould, J. Fletcher, M. D. Iannettoni, W. R. Lynch, D. E. Midthun, D. P. Naidich, and D. E. Ost Evaluation of Patients With Pulmonary Nodules: When Is It Lung Cancer?: ACCP Evidence-Based Clinical Practice Guidelines (2nd Edition) Chest, September 1, 2007; 132(3_suppl): 108S - 130S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Petrou, L. E. Quint, B. Nan, and L. H. Baker Pulmonary Nodule Volumetric Measurement Variability as a Function of CT Slice Thickness and Nodule Morphology Am. J. Roentgenol., February 1, 2007; 188(2): 306 - 312. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. G. Jennings, H. T. Winer-Muram, M. Tann, J. Ying, and I. Dowdeswell Distribution of Stage I Lung Cancer Growth Rates Determined with Serial Volumetric CT Measurements Radiology, November 1, 2006; 241(2): 554 - 563. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. T. Winer-Muram The solitary pulmonary nodule. Radiology, April 1, 2006; 239(1): 34 - 49. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. R. Goodman, M. Gulsun, L. Washington, P. G. Nagy, and K. L. Piacsek Inherent Variability of CT Lung Nodule Measurements In Vivo Using Semiautomated Volumetric Measurements. Am. J. Roentgenol., April 1, 2006; 186(4): 989 - 994. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
U. Tateishi, H. Uno, K. Yonemori, M. Satake, M. Takeuchi, and Y. Arai Prediction of Lung Adenocarcinoma Without Vessel Invasion: A CT Scan Volumetric Analysis Chest, November 1, 2005; 128(5): 3276 - 3283. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Goo, T. Tongdee, R. Tongdee, K. Yeo, C. F. Hildebolt, and K. T. Bae Volumetric Measurement of Synthetic Lung Nodules with Multi-Detector Row CT: Effect of Various Image Reconstruction Parameters and Segmentation Thresholds on Measurement Accuracy Radiology, June 1, 2005; 235(3): 850 - 856. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Zhao, L. H. Schwartz, C. S. Moskowitz, L. Wang, M. S. Ginsberg, C. A. Cooper, L. Jiang, and J. P. Kalaigian Pulmonary Metastases: Effect of CT Section Thickness on Measurement--Initial Experience Radiology, March 1, 2005; 234(3): 934 - 939. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. G. Jennings, H. T. Winer-Muram, R. D. Tarver, and M. O. Farber Lung Tumor Growth: Assessment with CT--Comparison of Diameter and Cross-sectional Area with Volume Measurements Radiology, June 1, 2004; 231(3): 866 - 871. [Abstract] [Full Text] [PDF]
|
|
| ||||||
