| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Breast Imaging |
1 From the Departments of Clinical Oncology (D.K.W.Y.), Diagnostic Radiology and Organ Imaging (W.T.Y.), and Anatomical and Cellular Pathology (G.M.K.T.), Prince of Wales Hospital, 30-32 Ngan Shing St, Shatin, Hong Kong, China. Received September 11, 2001; revision requested October 22; revision received January 7, 2002; accepted February 26. Address correspondence to D.K.W.Y. (e-mail: dkyeung@cuhk.edu.hk).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Breast and lymph node MR spectroscopy of lesions identified at contrast material–enhanced MR imaging was performed in 39 patients with breast cancer. Spectroscopic and histopathologic findings were determined and compared. The sensitivity, specificity, and accuracy of the MR spectroscopic technique in the detection of axillary lymph node metastases were determined.
RESULTS: There were four cases of ductal carcinoma in situ (DCIS) and 34 invasive carcinomas, including three with an extensive in situ component. Twenty-six breast lesions were positive for choline at MR spectroscopy; nine, negative; and three, failed cases (ie, determination of positive or negative for choline could not be made). No data were available for one lesion. Four of the nine negative findings were DCIS; three, infiltrating ductal carcinoma (IDC) with an extensive in situ component; and two, IDC. Fourteen axillary lymph nodes were positive for choline; 17, negative; and four, failed cases. No data were available for four nodes. Comparison of the preliminary diagnostic indexes of the MR spectroscopic technique with the ultrasonographically guided fine-needle aspiration biopsy findings in lymph nodes revealed a sensitivity of 82%, specificity of 100%, and accuracy of 90%.
CONCLUSION: Choline is consistently detected in IDC. DCIS and IDC with an extensive in situ component are likely to be negative for choline at MR spectroscopy. In vivo proton MR spectroscopy of axillary lymph nodes in patients with breast cancer is feasible and has encouraging preliminary results.
© RSNA, 2002
Index terms: Breast neoplasms, 00.32, 00.33 • Breast neoplasms, MR, 00.121411, 00.121413, 00.121415, 00.121416, 00.12143, 00.12145 • Lymphatic system, neoplasms, 997.33, 00.33 • Magnetic resonance (MR), spectroscopy, 00.12145
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
During the past 10 years, the results of both in vitro (13–15) and in vivo (16–20) 1H MR spectroscopic studies of the breast have confirmed that high levels of choline-containing compounds at 3.2 ppm accumulate mostly in malignant lesions and that low levels accumulate mostly in benign lesions or normal breast tissue. Although the use of the presence of choline at in vivo 1H MR spectroscopy to evaluate for breast cancer has shown promising results, a number of false-negative findings involving various in situ and invasive lesions also have been reported (17,18,20). It is unclear whether the failure to detect choline in some breast cancers is related to insufficient sensitivity or is an indication of less aggressive lesion types that have lower invasive or metastatic potential. If in vivo 1H MR spectroscopy is to be successfully used with other MR imaging modalities, the issue of whether there are subtypes of breast cancer that are likely to be negative for choline at spectroscopy must be examined.
Improvements in the diagnostic accuracy of breast MR imaging would be beneficial to patients because unnecessary biopsies would be avoided. For many patients with confirmed breast cancer who would have to undergo complete axillary lymph node dissection, however, the development of noninvasive procedures that enabled reliable detection of axillary lymph node metastases would not only help to reduce the risks of axillary dissection complications (21) but also help to determine the best treatment strategy. Although gadolinium-enhanced MR imaging has a good positive predictive value in the detection of lymph node metastases, it cannot be used to exclude positive nodes (22). In addition, to our knowledge, no study has been performed to examine whether in vivo 1H MR spectroscopy of axillary nodes is clinically feasible.
The purpose of our study was to assess the relationship between breast cancer subtypes and choline detection by using in vivo 1H MR spectroscopy and to assess the feasibility of 1H MR spectroscopy in the study of axillary lymph node metastases.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Of the 43 patients recruited, four with benign lesions at histopathologic analysis were excluded from the study cohort: One patient had diabetes-related mastopathy; one patient, sclerosing adenosis; and two patients, proliferative fibrocystic changes. Thus, thirty-nine patients (mean age, 52.5 years; age range, 26–82 years) who had histopathologically confirmed breast cancers were included in the study.
MR Imaging Examinations and Analysis
The imaging examinations were performed with a 1.5-T whole-body MR system (Gyroscan ACS-NT; Philips, Best, the Netherlands). A standard receive-only bilateral breast coil was used to perform both MR imaging and MR spectroscopy of the breasts. Additionally, a 14-cm circular receive-only surface coil that was placed against the affected breast was used to improve the signal-to-noise ratio of MR imaging and MR spectroscopy of the axillary lymph nodes. The patients were examined while in the prone position and with their breasts suspended in the breast coil. The body coil was used to generate a homogeneous radio-frequency induction field (ie, B1) in all MR examinations.
MR imaging was performed in the transverse and sagittal planes. Transverse images were obtained by using a T1-weighted spin-echo sequence (repetition time msec/echo time [TE] msec, 450/12; section thickness, 4 mm with no intersection gap; field of view, 350 mm; matrix, 256 x 256; two signals acquired) with spectral presaturation and inversion recovery for fat saturation. Thirty transverse images of both breasts were obtained before the administration of contrast material. After the patient was given a bolus intravenous injection of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany), 0.2 mmol per kilogram of body weight, a contrast-enhanced transverse MR image was acquired. Image subtraction was then performed to identify enhancing lesions on the subtracted images, and the largest dimension of the lesions was measured (W.T.Y.).
With the 14-cm circular surface coil selected as the signal receiver, unenhanced sagittal images of the affected breast and axillary lymph nodes were obtained by using a T2-weighted turbo spin-echo with spectral presaturation and inversion-recovery fat saturation sequence (2,000/100, 4-mm section thickness with 10% intersection gap, 256 x 256 matrix, three signals acquired). Then, contrast-enhanced sagittal images were obtained by using a T1-weighted spin-echo sequence (450/12, 4-mm section thickness with no intersection gap, 256 x 256 matrix, two signals acquired). The diameter of the largest lymph node was measured on the T1-weighted postcontrast MR image (W.T.Y.).
With MR imaging guidance, the radiologist (W.T.Y.) carefully positioned the volume of interest in each enhancing breast lesion (mean volume, 14.4 cm3; range, 3.4–45.0 cm3) and in a single (largest visible) axillary lymph node (mean volume, 11 cm3; range, 2.2–125.0 cm3). When there was more than one enhancing suspicious breast lesion, the largest lesion was selected for study. With use of the point-resolved spectroscopic, or PRESS, sequence (2,000/38, 135, 270) with either the bilateral breast coil or the circular surface coil selected as the receiver, three water-suppressed spectra were acquired for each volume of interest. Preacquisition automated parameter optimization consisted of frequency and receiver gain adjustment, shimming, and gradient tuning. Water suppression was achieved by means of selective inversion recovery; the measurement was started at the zero crossing of the water signal. Data were acquired at a spectral bandwidth of 1,000 Hz, and 64 signals were averaged for each water-suppressed spectrum. The time required to complete the spectroscopic examinations of both breasts and the axillary nodes was approximately 30 minutes.
Spectra were analyzed with the time-domain fitting routine variable projection, or VARPRO, method (23,24) that was incorporated in a computer software package (MR User Interface [MRUI]; developed by A. van den Boogaart, Katholieke Universiteit Leuven, Belgium) (25). Residual water was first removed from the measured free induction decay by means of time-domain Hankel-Lanczos singular value decomposition filtering (26). The resonance frequency and line width of choline were selected manually; these parameters were used as the starting values in the nonlinear least-squares fitting algorithm. The signal-to-noise ratio of the apparent peak at 3.2 ppm was measured, and a minimal signal-to-noise ratio of 2 was chosen to indicate that choline was present in a spectrum. A lesion was determined to be a positive MR spectroscopic finding if this signal-to-noise ratio criterion was met in at least two spectra acquired at different TEs.
Histopathologic Specimens and Comparisons
All 39 breast lesions in this study were enhancing and briefly clinically examined for palpability prior to core-needle biopsy with a 16-gauge needle (Monopty; Bard, Covington, Ga). Wide local excision and/or mastectomy was performed in 35 of the 39 patients. Of the remaining four patients, three were undergoing chemotherapy at the time this article was written, and one refused to undergo surgery.
The axillary lymph nodes were palpated. US-guided fine-needle aspiration biopsy (FNAB) was performed on all palpable and nonpalpable lymph nodes following 1H MR spectroscopy for direct comparison of findings in 35 of 39 patients who completed spectroscopy of the axilla. Four of the 39 patients did not have axillary node data: In one case the imaging study had to be aborted owing to machine breakdown, and in three cases imaging could not be completed because the patient had progressive claustrophobia or poor general health. Further comparison between MR spectroscopic and histopathologic findings was also performed at axillary lymph node dissection when specimens were available. All clinical examinations and tissue sample collections were performed by one radiologist (W.T.Y.), and all of the samples were analyzed by one pathologist (G.M.K.T.). For histopathologic analysis, the breast tumors were graded according to the World Health Organization classification system. The lymph nodes, if present, were evaluated for presence of tumor and for evidence of post-FNAB changes (ie, hemorrhage, reactive spindle cell reaction, and crush artifact). The physicist (D.K.W.Y.) who read the MR spectroscopic measurements in the breast and axilla was blinded to the MR imaging and histopathologic findings.
MR Spectroscopic Data Analysis
The MR spectroscopic results obtained by using different TEs were analyzed to show the relationship between choline detection and breast cancer classification. The MR spectroscopic lymph node results were compared with clinical (ie, palpability) examination, FNAB, and histopathologic (ie, dissection) findings. The possible reasons for false-negative results in the breast and lymph node studies, including those with noninterpretable spectra, were evaluated. True-positive, true-negative, false-positive, and false-negative detections, as well as sensitivity, specificity, and accuracy, were determined.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
On the basis of the criterion of a choline-containing compound that was identifiable in at least two spectra, 26 breast lesions were positive at MR spectroscopy, nine were negative, and three cases of invasive cancer were regarded as failed—that is, a determination of positive or negative for choline could not be made. Of the nine negative cases, four were DCIS; three, IDC with an extensive in situ component; and two, IDC not otherwise specified. Regardless of the breast cancer subtype, the overall rate of true-positive lesion detection was 74% (26 of 35 tumors). The observed mean signal-to-noise ratio of choline resonance in the breasts was 9.04 (range, 2.3–25.0). The MR spectroscopic and histopathologic (ie, core-needle biopsy) findings are summarized in Table 1.
|
When we compared the 1H MR spectroscopic results with the histopathologic findings at axillary lymph node dissection, we observed three more cases in which the 1H MR spectroscopic and FNAB results were negative but the nodes were positive at axillary dissection. These three cases included those of two patients (patients 20 and 21) each with a single 4-mm metastatic node containing a 2-mm focus of tumor. The third patient (patient 30) had two metastatic nodes that were 6 and 10 mm in diameter. Review of the histopathologic (ie, dissection) findings in the two false-negative cases (patients 20 and 21) with single 4-mm metastatic nodes revealed the presence of larger benign nodes that were 20 and 15 mm in diameter, respectively, and had intranodal hemorrhage that was consistent with post-FNAB changes. The sizes of these two nodes at histopathologic analysis corresponded to the sizes of the nodes measured and studied at 1H MR spectroscopy in both patients. These findings suggested that the larger benign nodes were detected at 1H MR spectroscopy and US-guided FNAB in both of these patients. No false-positive findings were observed in the axilla. The MR spectroscopic and histopathologic findings in the axillary lymph nodes are summarized in Table 2.
|
MR Spectroscopic Analysis
Figure 1 shows two true-positive MR spectra acquired at a 135-msec TE in patient 38, who had a diagnosis of IDC. The volume of interest was placed in the largest contrast-enhanced breast lesion and in the 2-cm axillary lymph node. Figure 2 shows a true-positive breast lesion spectrum and a true-negative axillary lymph node spectrum acquired at 1H MR spectroscopy with a 135-msec TE in patient 17, who had a diagnosis of medullary carcinoma. The spectra were acquired in the contrast-enhanced lesion and in the 1-cm axillary lymph node. Figure 3 shows an absence of detectable choline-containing compounds at 3.2 ppm in the MR spectra acquired at TEs of 38, 135, and 270 msec in patient 9, who had a diagnosis of pure DCIS.
|
|
|
|
|
|
|
|
|
When the 1H MR spectroscopic findings were compared with the histopathologic results obtained at surgical lymph node dissection, with the exclusion of those cases in which the patients were undergoing chemotherapy, the true-positive detection rate for 1H MR spectroscopy was 65% (11 of 17 cases); the true-negative detection rate, 100% (10 of 10 cases); the false-negative detection rate, 29% (five of 17 cases); and the false-positive detection rate, 0% (zero of 10 cases). Therefore, the mean sensitivity of 1H MR spectroscopy with lymph node dissection as the reference standard was 69% ± 8.9; the mean specificity, 100%; and the mean accuracy, 78% ± 8.0.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Because the minimal detectable amount of choline-containing compounds can be found in either a large lesion with a low internal choline concentration or a small lesion with a high choline concentration, the variety of voxel sizes used in our study may have been a factor in the observed differences in choline detection between the invasive and noninvasive cancers. However, our study results showed that for the DCIS breast cancer subtype, the voxel sizes were relatively large (7.5, 12.0, 36.0, and 45.0 cm3), but choline was not detected in these lesions. For the subtype of IDC with an extensive in situ component, the voxel sizes were 3.4, 15.0, and 17.5 cm3, and choline was not detected in these lesions. In addition, when the results for four lesions that were studied by using the same voxel size (3.4 cm3) were compared, the findings in patient 10, who had a diagnosis of IDC with an extensive in situ component, were negative for choline, whereas the findings in patients 17, 24, and 25, all of whom had a diagnosis of invasive cancer, were positive for choline. Similarly, when the results for four lesions that were studied by using similar voxel sizes (12.0–12.6 cm3) were compared, patient 7, who had a diagnosis of DCIS, had a negative MR spectroscopic result, whereas patients 15, 38, and 39, all of whom had a diagnosis of IDC, had positive spectroscopic results.
Differences in the 1H MR spectroscopic detection of choline between invasive and in situ cancers may reflect the distinct histopathologic and biologic subtypes of breast cancer. DCIS has little or no known metastatic potential, whereas invasive cancers have greater metastatic potential both regionally and systemically. All cases with a major in situ component were negative for choline despite a tumor size range of 1.5–8.0 cm; these included two cases of lesions that were 5 cm after MR contrast enhancement. These findings indicate that noninvasive cancers may have much lower levels of choline-containing compounds. It would be interesting to speculate on the reasons for these negative results. Although the exact biologic basis for elevated choline levels in malignancy is not fully understood, the presence of choline-containing metabolites reflects active cytoplasmic membrane synthesis.
Our in vivo finding is consistent with a previous report on ex vivo FNAB specimens from 218 breast lesions that were analyzed by using an 8.5-T MR spectrometer (13). In that study, the choline-to-creatine peak ratios were derived by using a semiquantitative approach, and a cutoff value of 1.7 was used to differentiate between benign and malignant lesions. In most cases of DCIS, the choline-to-creatine ratio was less than 1.7 at 1H MR spectroscopy, which is similar to the choline-to-creatine ratio in normal breast tissue and benign lesions. In six cases of DCIS with comedonecrosis or microinvasion, the patients were found to have choline-to-creatine ratios marginally higher than 1.7. In our series, only one patient with IDC with an extensive in situ component (patient 16) had a detectable level of choline at MR spectroscopy performed with the shortest TE (38 msec). The failure to detect choline in the same lesion with longer TEs (Table 1) suggests that the concentrations of choline in invasive lesions that have an extensive in situ component may be low. In contrast, choline was readily detected in 23 of 28 spectra in predominantly invasive breast cancers at 270-msec TE MR spectroscopy, despite considerable relaxation losses that occurred with the use of the longest TE.
All cases of the special subtypes of invasive cancer, including two cases of mucinous carcinoma, two cases of medullary carcinoma, and one case of infiltrating lobular carcinoma, were positive for choline at 1H MR spectroscopy in this study. This is more encouraging than the recent report by Yeung et al (20), in which a single case of medullary carcinoma that was negative at 1H MR spectroscopy is described. Two false-negative IDC cases (patients 18 and 32) in this study were of lesions that were 1.8 and 2.0 cm, respectively. In patient 18, the small breast lesion was close to the chest wall, where physiologically induced motion might have been a contributing factor. In patient 32, the subtraction MR images had considerable misregistration artifacts that suggested that she had moved between image acquisitions. Three breast 1H MR spectroscopic studies were not interpretable, and these failures may have been related to severe patient movement during data acquisition.
We extended the study protocol to evaluate the feasibility of applying in vivo 1H MR spectroscopy in the study of axillary lymph nodes in patients with breast cancer. Our preliminary observations are that 1H MR spectroscopy is a feasible technique that is capable of depicting choline in metastatic nodes. Because we studied only one node in each axilla in each patient, we compared the 1H MR spectroscopic results for axillary nodes with the US-guided FNAB findings and with the final histopathologic findings at axillary lymph node dissection as well. The sensitivity of MR spectroscopy was 82% when it was compared with US-guided FNAB and 69% when it was compared with histopathologic (dissection) analysis. Although a comparison of the MR spectroscopic and histopathologic findings was necessary to reflect the reality of clinical practice, we believe that two of the three lymph nodes that were false-negative at histopathologic (dissection) analysis in this study did not reflect the failure of 1H MR spectroscopy to depict metastasis, but rather they resulted from the selection of a nonmetastatic node for study.
We recognize the limitations of MR spectroscopy in the evaluation of the axilla. This is reflected in our study design, in which only a single node in each patient was studied and which emphasizes the importance of appropriate node selection for study. A potential application of this technique may be the evaluation of the sentinel node if it can be accurately identified at MR imaging. The aim of this study, however, was to evaluate the feasibility of in vivo 1H MR spectroscopy in the detection of choline in metastatic axillary nodes in patients with breast cancer. To our knowledge, this has not been attempted in vivo previously, but it has been performed in vitro in humans (27) and rats (28,29). Our aim is not to replace axillary node dissection or histopathologic analysis at this time.
A second limitation was that all of the lymph nodes in this study were 1 cm or larger. This precluded the evaluation and detection of metastases in small subcentimeter nodes, as was reflected in two of the false-negative cases. Although using a smaller surface coil improves the signal-to-noise ratio at in vivo 1H MR spectroscopy of lymph nodes, the sensitivity of the examination is limited owing to the inherent sensitivity of the MR system that operates at 1.5 T. With the introduction of clinical systems that operate at 3.0 T and the development of better coils, it will be interesting to see whether smaller breast lesions and lymph nodes can be studied by using 1H MR spectroscopy.
Another limitation was the extended examination time that was necessary for MR imaging and MR spectroscopy of the axilla. Several patients could not keep from moving at the end of the examination because of fatigue, and this contributed to the relatively higher rate of motion-related failure compared with the rate of motion-related failure in the studies of the breast alone. Reducing the number of spectra acquired for each volume of interest might help to shorten the overall examination time. Our study results suggest that a TE of around 135 msec is a good compromise for the sensitivity and intense fat signal intensity interference associated with shorter TEs. However, motion-related false-negative findings might affect the accuracy of the technique if choline is detected definitively on the basis of a single spectrum. Therefore, the development of 1H MR spectroscopic sequences in which motion correction procedures have been incorporated in the data acquisition is important (30).
In conclusion, we reported our preliminary observations in the measurement of choline levels at in vivo 1.5-T 1H MR spectroscopy of different breast cancer subtypes. In this study, invasive breast cancers consistently had elevated choline levels that could be reliably detected in lesions larger than 1 cm, whereas noninvasive cancers such as DCIS were negative for choline at 1H MR spectroscopy. Invasive cancers with an extensive in situ component also produced negative spectra in this preliminary study. Our study results suggest that consideration of the differences in choline detection and breast cancer subtypes may be useful in any combined breast MR examinations in which in vivo 1H MR spectroscopy is incorporated as an additional study for the characterization of breast lesions. We have found in vivo 1H MR spectroscopy of axillary lymph nodes to be feasible and that choline-containing compounds can be reliably detected in metastatic nodes in patients with breast cancer. A potential application of in vivo 1H MR spectroscopy may be the noninvasive evaluation of the sentinel node for prognostic and surgical planning purposes.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, D.K.W.Y., W.T.Y., G.M.K.T.; study concepts and design, D.K.W.Y., W.T.Y., G.M.K.T.; literature research, D.K.W.Y., G.M.K.T.; clinical studies, W.T.Y.; data acquisition, D.K.W.Y., W.T.Y.; data analysis/interpretation, D.K.W.Y., W.T.Y., G.M.K.T.; statistical analysis, D.K.W.Y.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, D.K.W.Y., W.T.Y., G.M.K.T.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Wang, P.-L. Torng, T.-P. Liu, K.-L. Chen, and T. T.-F. Shih Proton MR Spectroscopy in Normal Breasts Between Pre- and Postmenopausal Women: A Preliminary Study Am. J. Roentgenol., February 1, 2008; 190(2): 505 - 510. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Bartella, S. B. Thakur, E. A. Morris, D. D. Dershaw, W. Huang, E. Chough, M. C. Cruz, and L. Liberman Enhancing Nonmass Lesions in the Breast: Evaluation with Proton (1H) MR Spectroscopy Radiology, October 1, 2007; 245(1): 80 - 87. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Bartella and W. Huang Proton (1H) MR Spectroscopy of the Breast RadioGraphics, October 1, 2007; 27(suppl_1): S241 - S252. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. C. Genson, C. E. Blane, M. A. Helvie, S. A. Waits, and T. L. Chenevert Effects on Breast MRI of Artifacts Caused by Metallic Tissue Marker Clips Am. J. Roentgenol., February 1, 2007; 188(2): 372 - 376. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Bartella, E. A. Morris, D. D. Dershaw, L. Liberman, S. B. Thakur, C. Moskowitz, J. Guido, and W. Huang Proton MR Spectroscopy with Choline Peak as Malignancy Marker Improves Positive Predictive Value for Breast Cancer Diagnosis: Preliminary Study Radiology, June 1, 2006; 239(3): 686 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Huang, P. R. Fisher, K. Dulaimy, L. A. Tudorica, B. O'Hea, and T. M. Button Detection of Breast Malignancy: Diagnostic MR Protocol for Improved Specificity Radiology, August 1, 2004; 232(2): 585 - 591. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. K. Tse, H. S. Cheung, L.-M. Pang, W. C. W. Chu, B. K. B. Law, F. Y. L. Kung, and D. K. W. Yeung Characterization of Lesions of the Breast with Proton MR Spectroscopy: Comparison of Carcinomas, Benign Lesions, and Phyllodes Tumors Am. J. Roentgenol., November 1, 2003; 181(5): 1267 - 1272. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. S. Siegelman Invited Commentary RadioGraphics, July 1, 2003; 23(4): 981 - 981. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
