| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology, Gifu University School of Medicine, Japan (M.K., M.M., H.K., M.E., S.G., H.H.); Department of Radiology, CB 7510, University of North Carolina, Chapel Hill, NC 27599-7510 (R.C.S.); and Department of Diagnostic Radiology, National Cancer Center Hospital, Tokyo, Japan (N.M.). Received April 27, 2001; revision requested June 9; final revision received March 14, 2002; accepted April 29. Supported in part by the Grant for Scientific Research Expenses for Health, Labour, and Welfare Programs; the Foundation for the Promotion of Cancer Research; and the 2nd-Term Comprehensive 10-Year Strategy for Cancer Control. Address correspondence to M.K. (e-mail: masa-gif@umin.ac.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: MR images were obtained after intravenous bolus injection of gadopentetate dimeglumine in 100 patients with chronic liver damage. Test bolus imaging was performed to determine the aortic transit time. A 26-second spoiled gradient-recalled-echo sequence was used. Patients were randomized into four groups so that the middle of k space was acquired at 5, 10, 15, and 20 seconds for the first phase and 45, 50, 55, and 60 seconds for the second phase, respectively, from the time of arrival of contrast material in the abdominal aorta. Mean signal intensities of the liver, spleen, and abdominal aorta were measured, and images were reviewed prospectively by three radiologists in consensus. Analysis of variance, the Scheffé criterion for continuous data, and the Kruskal-Wallis test for categorical data were used for statistical evaluation.
RESULTS: Intense splenic enhancement with the moiré pattern without intense hepatic enhancement occurred at 10–15 seconds. Aortic and splenic enhancement significantly decreased from 45 to 50 seconds (P < .05). Spleen-to-liver contrast-to-noise ratio began to decrease at 20 seconds and decreased constantly over time. Qualitative results correlated well with quantitative results.
CONCLUSION: Biphasic imaging with k space centered at 10–15 and 50 seconds or later after arrival of contrast material in the abdominal aorta may be the optimal technique to obtain ideal contrast enhancement. Empirically, delays of 28–34 and 68 seconds or later after initiating contrast material injection may be effective for biphasic imaging.
© RSNA, 2002
Index terms: Hepatic arteries, MR, 952.91 • Liver, cirrhosis, 761.288 • Liver, diseases, 761.288 • Liver, MR, 761.121412, 761.12143 • Portal vein, MR, 957.91
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In particular, the detection of hepatocellular carcinoma, which frequently arises in cases of chronic liver damage, requires optimal hepatic arterial phase imaging, as shown by results of previous CT studies (12,13). In prior articles (3,7,14), with the use of an empirically determined imaging delay of a fixed 15–20 seconds from the initiation of manual contrast material injection, researchers have described the initiation of data acquisition for hepatic arterial phase imaging. Earls et al (15) reported the wide range of contrast material transit time in the abdominal aorta (8–32 seconds) and main portal vein (20–55 seconds) following a test bolus injection. They reported that successful hepatic arterial phase images were obtained in no more than 61% of patients in whom an empirical imaging delay was used, and they advocated the routine use of a test bolus injection with a power injector.
Meanwhile, results of a study on gadolinium-enhanced MR angiography (16) have shown the clinical usefulness of triggering software to obtain optimal timing to produce image contrast as intense as possible; this software device automatically initiates data acquisition with a certain delay time after detection of contrast material arrival in the aorta. Although the device has not been compatible with two-dimensional GRE imaging of the liver, we believe this technique will be applied to dynamic MR imaging of the liver in the near future.
When using triggering software, or even when performing test bolus imaging, we need to determine optimal imaging delays from contrast medium arrival in the aorta to obtain optimal hepatic arterial and portal venous phase images. The purpose of our study, therefore, was to investigate the optimal imaging delays for hepatic arterial and portal venous phases of gadolinium-enhanced dynamic spoiled GRE MR imaging of the liver in patients with chronic liver damage.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Twenty patients were excluded from the study for the following reasons: 12 patients had previously undergone partial hepatectomy for hepatocellular carcinoma, three had previously undergone splenectomy for hypersplenism due to cirrhosis, one with Wilson disease had undergone splenectomy, one had severe hemochromatosis, one had severe heart failure, and two experienced technical failure during MR imaging. These patients were excluded from the study population because the vascular alteration after surgery or the susceptibility effect due to metal deposition may have adversely affected the determination of timing in the hepatic circulation.
The remaining 100 patients formed the study population and included 70 men and 30 women, aged 42–85 years (mean age, 66.5 years). The cause of chronic liver damage and the clinical severity and progression of cirrhosis evaluated according to the Child-Pugh classification (grades A, B, and C, on the basis of a score from 5 to 15, for the variables of ascites, encephalopathy, bilirubin and albumin values, and prothrombin time) (18,19) and the presence of a malignant hepatic tumor were assessed on the basis of patient records.
MR imaging was performed with a 1.5-T superconducting MR system (Signa Horizon; GE Medical Systems, Milwaukee, Wis). The system provided a maximum gradient strength of 23 mT/m, with a peak slew rate of 120 mT/m/msec. All MR images were obtained with a phased-array body multicoil. The MR imaging protocol consisted of in-phase T1-weighted GRE imaging (150/4.2 [repetition time msec/echo time msec], 512 x 224 matrix, one signal acquired, and two 20-second breath-hold acquisitions for coverage of the entire liver), fat-suppressed respiratory-triggered fast spin-echo T2-weighted imaging (3,750–7,500 [effective]/80 [effective], echo train length of 12–18, 512 x 256 matrix, three signals acquired, and 3.2–5.0-minute acquisition time), and breath-hold gadolinium-enhanced triphasic dynamic fast multiplanar spoiled GRE imaging with steady-state free precession (150/1.6, 90° flip angle, 512 x 224 matrix, received bandwidth of plus or minus 62.5 kHz, one signal acquired, 18 locations per 26 seconds, no chemical-shift–selective fat-suppression technique). The k-space lines for the GRE sequence were filled with echo data from top to bottom in the phase-encoding direction (sequential view ordering). The center of k space was obtained at 13 seconds in the 26-second sequence.
For all transverse MR imaging, the section thickness was 8 mm with a 2-mm intersection gap, and spatial presaturation pulses were applied superiorly and inferiorly to the imaging volume. The imaging volume was determined to provide sufficient coverage of the entire liver.
Test Bolus Imaging
To measure aortic transit time, defined as the time from initiation of intravenous contrast material injection to peak enhancement of the abdominal aorta at the level of the first lumbar vertebral body, transverse single-section fast multiplanar spoiled GRE images (19/1.3, 90° flip angle, 256 x 128 matrix, 32 x 24-cm field of view, received bandwidth of ±62.5 kHz, one signal acquired, 2-second acquisition time) were obtained every 2 seconds after initiation of an intravenous bolus injection of 1 mL of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany), followed by a flush with sterile saline solution. The volume of the flush was calculated according to patient body weight by using the following equation: V = 15 + 0.2 x Wbody - 1, where V is the volume of the saline solution in milliliters and Wbody is patient body weight in kilograms. A commercially available power injector (Auto-enhance A50; Nemotokyorindo, Tokyo, Japan) was used to inject first contrast material and then saline solution into an antecubital vein through a 22-gauge 25-mm-long catheter at an injection rate of 3 mL/sec.
Signal intensities of the abdominal aorta on test bolus GRE images were measured by using operator-defined region-of-interest measurements of mean signal intensity by using a circular cursor 15–20 mm (mean, 17.4 mm) in diameter, which was operated by one of three radiologists (M.K., M.M., M.E.), and the aortic transit time was determined as an odd number.
Dynamic MR Imaging
Breath-hold dynamic gadolinium-enhanced GRE images were obtained with a rectangular field of view of 22–29 x 26–35 cm. Dynamic GRE images were obtained before and after an intravenous bolus injection of 0.1 mmol per kilogram of body weight of gadopentetate dimeglumine, followed by a 15-mL flush of sterile saline solution. The amount of sterile saline solution flushed was fixed in all patients. The same power injector used in test bolus imaging was used for injecting contrast material and saline solution at a rate of 3 mL/sec.
The practical imaging delay (D, the time in seconds from initiation of contrast material injection to initiation of GRE image acquisition) for first-phase dynamic GRE imaging was determined in individual patients by using the following equation: D = TV-A + t - 13, where TV-A was the aortic transit time (from the antecubital vein [V] to the abdominal aorta [A]) determined by means of test bolus imaging, and t was the experimental imaging delay (time in seconds between arrival of contrast material in the abdominal aorta and middle of GRE image acquisition). Thirteen seconds, one-half the acquisition time of the fast multiplanar spoiled GRE sequence, was subtracted because the central k-space lines determined the bulk of signal intensity on MR images. Figure 1 illustrates the timing scheme. Second-phase dynamic GRE imaging was initiated 14 seconds after completion of hepatic arterial phase GRE imaging in all patients to allow time for patient respiration.
|
Quantitative Image Analysis
The quantitative measurements were conducted by one of three radiologists (M.K., M.M., M.E.). Operator-defined region-of-interest measurements were obtained on unenhanced, first-phase, and second-phase dynamic gadolinium-enhanced GRE MR images, with mean signal intensity measured in the right and left lobes of the liver, spleen, abdominal aorta, and background. A circular cursor of 18–25 mm (mean, 20.5 mm) in diameter was used for the liver and spleen, and an oval cursor of 18–22 x 100–150 mm (mean, 20.5 x 108.5 mm) was used for the background. Signal intensities in the liver and spleen were measured in regions without large vessels, dilated intrahepatic biliary ducts, or prominent artifacts. Signal intensities of the background were measured in the phase-encoding direction outside the anterior abdominal wall. The cursors were placed so that the vertical distance from the anterior abdominal wall was consistent among the different imaging sequences with regard to the near-field effect with the surface coil.
In each patient, the two hepatic measurements were averaged to obtain the mean signal intensity of the liver. The signal-to-noise ratios for the liver, spleen, and abdominal aorta were calculated by dividing the signal intensity in the respective organ by 1 SD of background signal intensity. Quantitative degrees of contrast enhancement in the liver, spleen, and abdominal aorta were expressed as contrast enhancement indexes (CEIs), which were calculated by subtracting the signal-to-noise ratios on unenhanced GRE images from those on first- and second-phase dynamic GRE images. The spleen-to-liver contrast-to-noise ratios (CNRs) were calculated by dividing the difference in signal intensity between the spleen and liver by 1 SD of background signal intensity. One SD of splenic signal intensity was recorded to quantitatively assess the degree of moiré pattern enhancement of the spleen.
Qualitative Image Analysis
Three experienced MR image readers (M.K., M.M., H.K.) in consensus reviewed the first-phase (hepatic arterial) and second-phase (portal venous) dynamic GRE images separately, with reference to the unenhanced GRE images. The images were evaluated prospectively by means of subjective assessment by the three readers, who were blinded to patient clinical information. The readers evaluated the first- and second-phase images in terms of degree of enhancement of four items: the main portal vein, hepatic parenchyma, spleen, and splenic moiré pattern enhancement. The readers used a three-point scale: A score of 1 was assigned when the organ was not enhanced at all or minimally enhanced, a score of 2 when the organ was enhanced but not fully enhanced, and a score of 3 when the organ was maximally enhanced.
Finally, the readers recorded their impression of a fifth item, the overall image value, for hepatic arterial and portal venous phase images by using a three-point scale, with a score of 1 assigned for no or little value, a score of 2 for moderately informative value, and a score of 3 for optimal and highly informative value. Thus, there were a total of 10 scores for each patient (five for the hepatic arterial phase images and five for the portal venous phase images).
Determining Effective Imaging Delays
To determine the effective imaging delays applied for biphasic imaging without test bolus imaging, resultant imaging delays from initiation of contrast material injection to middle of GRE image acquisition were calculated from the practical imaging delays recorded in the four groups of patients.
Statistical Analysis
Analysis of variance and multiple comparisons with the Scheffé criterion (20) were performed to evaluate the following determinants in the four groups: patient age and body weight; aortic transit time; CEI for the liver, spleen, and abdominal aorta; spleen-to-liver CNR; and 1 SD of splenic signal intensity. The Kruskal-Wallis test and multiple comparisons with the Scheffé criterion were performed for evaluation of qualitative scores obtained as categorical data (21).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
-fetoprotein or PIVKA II and follow-up imaging findings in the remaining six.
Quantitative Assessment
There was no significant difference in patient age, body weight, or aortic transit time among the four patient groups (Table 1). The mean CEIs for the liver, spleen, and abdominal aorta in the four patient groups are summarized in Table 2. The experimental imaging delay versus mean CEI curves for the liver, spleen, and abdominal aorta plotted for the four patient groups are shown in Figure 2. The mean CEI of the liver in the first phase increased constantly from 5 to 15 seconds and in the second phase had a peak at 45 seconds, decreased at 50 seconds, and fluctuated afterward (Fig 2). The mean CEI of the liver in the first phase was significantly higher (P < .005) at 15–20 seconds than at 5 seconds and in the second phase was significantly higher (P < .05) at 45 seconds than at 50 seconds (Fig 2).
|
|
|
The mean CEI of the abdominal aorta in the first phase showed a peak at 5 seconds and decreased at 10–20 seconds and in the second phase had a peak at 45 seconds and decreased and fluctuated at 50–60 seconds (Fig 2). The mean CEI of the abdominal aorta in the first phase was significantly higher (P < .001) at 5 seconds than at 10–20 seconds and in the second phase was significantly higher (P < .05) at 45 seconds than at 50–60 seconds (Fig 2).
The experimental imaging delay versus mean spleen-to-liver CNRs plotted for the four patient groups are shown in Figure 3. The spleen-to-liver CNR in the first phase had a peak at 5 seconds and decreased minimally with fluctuation over time and in the second phase had a peak at 45 seconds and decreased constantly over time. There was no significant difference among the mean spleen-to-liver CNRs in the first or second phase.
|
|
|
|
|
|
|
|
|
|
Effective Imaging Delay
Resultant imaging delays from initiation of contrast material injection to middle of GRE image acquisition in the four groups are summarized in Table 4. The mean resultant imaging delays in the four groups of experimental imaging delays at 5, 10, 15, and 20 seconds were 22.6, 28.3, 34.0, and 36.6 seconds, respectively, and those at 45, 50, 55, and 60 seconds were 62.6, 68.3, 74.0, and 76.6 seconds, respectively.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In these previous CT studies, however, the difference in aortic transit time in individual patients was not taken into consideration. The aortic transit time ranged widely from 11 to 37 seconds in our current study (Table 1), which suggests that the imaging delay should be optimized on the basis of the time between arrival of contrast material in the abdominal aorta and data acquisition, rather than the time between the initiation of contrast material injection and data acquisition. This delay may be particularly applicable for dynamic MR imaging, since the volume of contrast material is much less for MR imaging than that for CT. Therefore, the aortic transit time and duration of peak enhancement are more dependent on the individual patient’s circulation time with dynamic MR imaging than with CT.
Our quantitative results showed that splenic enhancement at 5 seconds was already as intense as that at 10–15 seconds, and the liver was not significantly enhanced at 5 seconds. This observation may suggest the effectiveness of obtaining hepatic arterial phase images even at 5 seconds. At this timing, however, splenic enhancement can be premature in some patients and intense in others. Our qualitative results indicated that there was either no splenic enhancement or weak enhancement in six (24%) of 25 patients with first-phase images obtained at 5 seconds. This observation may indicate that acquisition of hepatic arterial phase images at 5 seconds has a substantial risk of premature hepatic arterial phase imaging.
Both quantitative and qualitative results in our study showed that liver parenchyma began to enhance via the portal veins at 20 seconds, and imaging at this time may be late for hepatic arterial phase imaging. Other observations with images obtained at 20 seconds, such as quantitative decrease in spleen-to-liver CNR, quantitative decrease in the mean of 1 SD of splenic signal intensity, and qualitative decrease in the degree of splenic moiré pattern enhancement, may suggest the inappropriateness of obtaining hepatic arterial phase images at 20 seconds.
Earls et al (15) reported that the test bolus reached the main portal vein between 20 and 55 seconds after initiating test bolus injection. Results of our qualitative image analysis showed that the main portal vein was moderately to intensely enhanced in 10 (40%) of 25 patients at 5 seconds, in 19 (76%) of 25 patients at 10 seconds, and in 24 (96%) of 25 patients at 15 seconds. The mean CEI of hepatic parenchyma significantly increased at 15–20 seconds and plateaued from 15 to 20 seconds. The CEI of hepatic parenchyma at 45 seconds was 2.5 times greater than that at 15–20 seconds. These observations suggest that the "portal venous phase," defined as a phase when the hepatic parenchyma is most intensely enhanced, does not commence until 20 seconds. The protocol in our current study did not specifically allow time between contrast material arrival in the abdominal aorta and peak hepatic parenchymal enhancement.
We (25) recently evaluated hepatic and pancreatic enhancement that occurs at different delays after contrast material arrival in the abdominal aorta in a study design similar to that in the current study. Our previous results showed that hepatic enhancement plateaued at a peak level at 25–45 seconds. Now, we deduce that the hepatic parenchymal enhancement rapidly reaches a peak at 25 seconds, maintains its intensity from 25 to 45 seconds, and significantly decreases at 50 seconds.
Our quantitative and qualitative results may suggest that optimal portal venous phase images are obtained at 45 seconds. Our quantitative results, however, showed that the abdominal aortic and splenic enhancements were still intense at 45 seconds and significantly decreased from 45 to 50 seconds. The persistent aortic and splenic enhancement at 45 seconds implies persistent hepatic arterial perfusion at this timing, leading to persistent enhancement of hepatocellular carcinoma or hypervascular metastasis in the enhanced surrounding liver parenchyma and possibly decreasing lesion conspicuity. Optimal portal venous phase images may be obtained at 50 seconds or later.
Limiting factors to the feasibility of biphasic imaging of the liver consist of (a) imaging delays for the hepatic arterial and portal venous phases, (b) acquisition time during respiratory suspension, and (c) breathing interval between the two phases. For example, by using a pulse sequence with a 20-second acquisition time, hepatic arterial and portal venous phase images were obtainable at delays of 10–15 and 50 seconds, respectively, while ensuring a 15–20-second breathing interval. However, by using a pulse sequence with a larger image matrix than that used in the current study, it may be feasible to delay portal venous phase imaging (ie, imaging at 55–60 seconds) to maintain a minimum but required breathing interval.
Our rationale for evaluation of only patients with chronic liver damage are twofold: First, hepatocellular carcinomas frequently arise in livers with chronic hepatitis or cirrhosis associated with viral hepatitis, alcoholism, primary biliary cirrhosis, or congenital metabolic disorder such as Wilson disease, and optimally obtained hepatic arterial phase images should improve detection of hypervascular hepatocellular carcinoma. Second, in cirrhosis accompanied by portal venous hypertension, the optimal imaging delay for the portal venous phase may be more delayed than that in livers without cirrhosis, although a distinct portal venous phase may not occur in cases of severe cirrhosis with scarce portal venous flow in the liver.
Bolus-tracking triggering software or test bolus imaging may be an important aspect of contrast-enhanced studies. However, commercially available triggering software compatible with two-dimensional GRE sequences is not available at the present time. Also, a test bolus is not performed at many centers because of patient throughput consideration, and an empiric imaging delay is often used. In our study, the patients imaged at 10 and 15 seconds after arrival of contrast material in the abdominal aorta, in other words, were imaged at 28.3 seconds ± 5.2 and 34.0 seconds ± 4.2, respectively, after the initiation of contrast material injection. This observation may imply that an effective imaging delay (from the initiation of contrast material injection to the middle of GRE image acquisition) of 28–34 seconds is recommended for obtaining hepatic arterial phase images. Likewise for the portal venous phase, an effective imaging delay of 68–77 seconds is recommended. Nevertheless, it should be noted that the resultant imaging delay for first-phase imaging had a fairly wide range (>12 seconds) (Table 2), which suggests that the use of the effective imaging delay we now advocate still has a considerable risk of suboptimal hepatic arterial phase imaging and that the use of a tailored imaging delay is recommended.
There are some limitations to our study. We did not perform a region-of-interest measurement study for hepatic tumors because the vascularity varied considerably among lesions and even within a lesion because of mosaic enhancement or internal necrosis. It is commonly recognized that moiré pattern splenic enhancement is indicative of optimal hepatic arterial phase imaging, and contrast enhancement of hypervascular hepatic tumors and that of the spleen are comparable (26), although splenomegaly may elongate the duration of moiré pattern enhancement in the spleen. Although two of the three blinded readers had performed region-of-interest measurements, the knowledge bias was considered minimal because the readers measured signal intensity in a total of 15 locations in one patient, they had been instructed not to pay attention to the values during the measurement procedures, and they never handled the measurement values before image review. Finally, our results are not necessarily applicable to patients with healthy livers, particularly when the imaging delay for portal venous phase imaging is considered, because portal venous flow velocity is commonly higher in healthy livers than in livers with portal hypertension.
In conclusion, optimal enhancement is achieved when images are obtained so that k-space line data acquisition occurs at 10–15 seconds after arrival of contrast material in the abdominal aorta for the hepatic arterial phase and at 50 seconds or later for the portal venous phase. This imaging delay should apply under the conditions that gadopentetate dimeglumine (0.1 mmol/kg) and sterile saline solution flush (15 mL) are injected intravenously at a rate of 3 mL/sec, and the actual delay setting may depend on certain factors such as breath-hold acquisition time or breathing interval. Empirically, with conditions similar to those used in our study, imaging delays of 28–34 seconds for hepatic arterial phase imaging and 68 seconds or later for portal venous phase imaging, from initiating contrast material injection to the middle of GRE image acquisition, may be effective for biphasic imaging in patients with chronic liver damage.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, H.H.; study concepts, M.K.; study design, N.M., M.K.; literature research, M.K.; clinical studies, M.K., M.M., M.E., H.K., S.G.; data acquisition, M.K., M.M., M.E., H.K., S.G.; data analysis/interpretation, N.M., M.K., M.M., H.K.; statistical analysis, N.M., M.K.; manuscript preparation, M.K.; manuscript definition of intellectual content and editing, M.K., R.C.S.; manuscript revision/review, N.M., M.K., R.C.S.; manuscript final version approval, H.H.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. F. Hanna, N. Kased, S. W. Kwan, A. C. Gamst, A. C. Santosa, T. Hassanein, and C. B. Sirlin Double-Contrast MRI for Accurate Staging of Hepatocellular Carcinoma in Patients with Cirrhosis Am. J. Roentgenol., January 1, 2008; 190(1): 47 - 57. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. L. Chu, B. N. Joe, A. C. A. Westphalen, E. M. Webb, F. V. Coakley, and B. M. Yeh Patient-specific Time to Peak Abdominal Organ Enhancement Varies with Time to Peak Aortic Enhancement at MR Imaging Radiology, December 1, 2007; 245(3): 779 - 787. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Mori, H. Yoshioka, N. Takahashi, M. Yamaguchi, T. Ueno, T. Yamaki, and Y. Saida Triple Arterial Phase Dynamic MRI with Sensitivity Encoding for Hypervascular Hepatocellular Carcinoma: Comparison of the Diagnostic Accuracy Among the Early, Middle, Late, and Whole Triple Arterial Phase Imaging Am. J. Roentgenol., January 1, 2005; 184(1): 63 - 69. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. G. Armato III, G. McLennan, M. F. McNitt-Gray, C. R. Meyer, D. Yankelevitz, D. R. Aberle, C. I. Henschke, E. A. Hoffman, E. A. Kazerooni, H. MacMahon, et al. Lung Image Database Consortium: Developing a Resource for the Medical Imaging Research Community Radiology, September 1, 2004; 232(3): 739 - 748. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
