点击显示 收起
1 From the Department of Radiology, Vancouver Hospital and Health Science Center, Vancouver, British Columbia, Canada (A.L.S.); Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710 (R.C.N., C.R.L., G.A.J., D.H.S., E.K.P.); Department of Radiology, Northeast Medical Center, Concord, NC (D.H.S.); and GE Medical Systems, Milwaukee, Wis (G.S.). Received December 27, 2000; revision requested January 26, 2001; revision received July 9; accepted July 27.
| ABSTRACT |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|
Index terms: Computed tomography (CT), helical, 761.12115 • Liver, CT, 761.12114, 761.12115
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|
For reasons that are poorly understood, tumor enhancement patterns are widely variable, even in a single patient with multiple lesions from the same primary neoplasm. Characteristic enhancement patterns have been ascribed to certain primary and metastatic hepatic tumors, such as focal nodular hyperplasia, hepatocellular carcinoma, neuroendocrine tumors, and melanoma, which tend to be relatively hypervascular and hyperattenuating (3,4). Other hepatic metastases may be hypoattenuating (eg, colorectal carcinoma) or variable in vascularity and enhancement (eg, breast carcinoma, which is frequently hypervascular) (5–7).
With helical CT, it is possible to image the entire liver during each of two somewhat distinct phases of contrast media enhancement, the earlier and fleeting hepatic arterial–dominant phase, and the later and more prolonged portal venous–dominant phase (8). Detection of hepatic lesions often requires a tailored protocol to optimize the phase of enhancement to improve lesion conspicuity. This is particularly important for hypervascular lesions, where scanning timing and contrast media flow rates may be crucial to optimize detection in the often fleeting hepatic arterial phase (9,10).
Recently, helical CT scanners with multiple detector rows have been introduced that dramatically increase acquisition speeds (11). Since as many as four sections are obtained during each gantry rotation, the entire liver can be examined in as little as 5–7 seconds, depending on the gantry rotation and table speed (12). This increase in speed not only allows optimization of dual phase imaging but also provides some unique opportunities. Digital subtraction of CT images has been possible in the past, although the technique was limited by respiratory misregistration that resulted from multiple breath holds. With multi–detector row helical CT, it is now possible to acquire two phases of hepatic contrast enhancement during the same breath hold, which potentially negates the effects of respiratory misregistration. This facilitates digital subtraction of the precontrast CT scan from the hepatic arterial CT scan to emphasize the various features of early contrast enhancement.
The purpose of our study was to determine the technical feasibility of performing dynamic subtraction CT of the liver during a single breath hold with multi–detector row helical CT.
| Materials and Methods |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|
Each patient underwent our standard triple-phase liver CT protocol, which was modified slightly to allow acquisition of the precontrast and hepatic arterial CT scans in a single breath hold. Triphasic liver CT was performed with a helical CT scanner (QX/i Lightspeed; GE Medical Systems, Milwaukee, Wis), which obtains four sections per gantry rotation, with a rotation speed of 0.8 second. The triphasic helical acquisition was performed with 175 mL of contrast material (iopamidol, Isovue 300 [300 mg of iodine per milliliter]; Bracco Diagnostics, Princeton, NJ) administered intravenously at a rate of 4 mL/sec by using a mechanical power injector (PercuPump, E-Z-Em, Westbury, NY; or Envision, MedRad, Pittsburgh, Pa). After a 7-second scanning delay, the precontrast CT scan was acquired with 5-mm-thick sections (5.0-mm full width at half maximum), table speed of 28.13 mm/sec, 140 kVp, and 100 mA.
Timing for the precontrast CT acquisition (7-second delay) allowed us to shorten the breath hold necessary for the combined precontrast and hepatic arterial acquisitions while remaining without contrast material, since the contrast material had not reached the liver. After a 34-second scanning delay, the hepatic arterial CT scans were acquired with 5-mm-thick sections (5.0-mm full width at half maximum), table speed of 28.13 mm/sec, 140 kVp, and 210–220 mA. The detector configuration for the precontrast and hepatic arterial CT acquisitions was such that four 3.75-mm-thick sections were acquired during each gantry rotation. The precontrast and hepatic arterial data sets were reconstructed at 2-mm intervals. After a 65-second scanning delay, portal venous CT scans were acquired with 5-mm-thick sections (5.2-mm full width at half maximum), table speed of 18.75 mm/sec, 140 kVp, and 170–220 mA. The detector configuration for the portal venous phase was such that four 5.0-mm-thick sections were acquired during each gantry rotation. The portal venous data were not used in the subtraction study.
The precontrast and hepatic arterial CT scans were acquired during the same breath hold in an attempt to minimize respiratory misregistration (Fig 1). Scanning times for precontrast and hepatic arterial acquisitions were identical in each patient (mean, 6.8 seconds ± 1.0 [SD]; range, 4.6–9.2 seconds); the interval between the two phases was a mean 21.9 seconds ± 2 (range, 15.0–30.2 seconds). Overall, the breath-hold length for combined precontrast and hepatic arterial CT, including the interscan delay, was a mean 35.4 seconds ± 2.5 (range, 29.3–45.8 seconds). The breath-hold length varied somewhat depending on the longitudinal dimension of the liver.
|
One investigator (A.L.S.) performed digital subtraction of the precontrast CT scan from the hepatic arterial CT scan by using a workstation (Advantage Windows, version 3.1p; GE Medical Systems). The attenuation values for paired images from the two data sets, which corresponded to a similar physical location in the patient’s body, were subtracted on a voxel-by-voxel basis. Negative values were automatically set to zero. Measurements were made in regions of interest placed by one investigator (A.L.S.) in liver parenchyma, the right paraspinal muscle, and any lesions that were present. Care was taken to avoid blood vessels, perfusion abnormalities, and artifacts. The area of the regions of interest was between 75 and 200 mm2. Liver-to-muscle and lesion-to-liver contrast ratios were calculated for the precontrast, hepatic arterial, and subtracted images. Because muscle demonstrates relatively little enhancement on postcontrast CT scans, the liver-to-muscle contrast ratio was used as an estimate of relative hepatic enhancement. Perilesion perfusion abnormalities were also observed.
At the time this study was performed, the local institutional review board representative determined that formal approval and informed consent were not required because the study protocol did not differ substantially from our routine multi–detector row CT protocol.
The mean values of the contrast ratio variables were compared by means of a signed rank test. Kendall rank correlation coefficients, a nonparametric measure of correlation between two ordinal variables, were computed. After Bonferroni adjustment for multiple comparisons, differences with a P value less than .025 were considered significant.
| Results |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|
Subtracted images revealed vivid enhancement of the liver parenchyma. The mean liver-to-muscle contrast ratios for the precontrast, hepatic arterial, and subtracted images were 1.3 ± 0.3, 1.4 ± 0.3, and 2.3 ± 1.1, respectively. The mean contrast ratio was significantly higher for the subtracted images compared with the precontrast and hepatic arterial CT images (P < .001). In the 13 patients with focal lesions, the mean liver-to-lesion contrast ratios for the precontrast, hepatic arterial, and subtracted images were 0.9 ± 0.5, 1.5 ± 0.4, and 3.7 ± 1.8, respectively. In patients who demonstrated liver lesions, there was a mean 2.5-fold increase in mean lesion-to-liver contrast ratio on the subtracted images compared with the hepatic arterial images (P < .001) (Figs 2, 3). Anecdotally, in a number of the patients with liver lesions, perilesion perfusion abnormalities were revealed on the subtracted images that were not visualized on the hepatic arterial images (Figs 4, 5). The perfusion abnormalities were seen as areas of hyperattenuation around the lesion; one of these areas had classic imaging features of a simple hepatic cyst.
|
|
|
|
|
|
|
|
|
|
| Discussion |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|
In this study, multi–detector row CT allowed rapid imaging of the entire liver in 5–9 seconds. Multi–detector row CT has the potential to improve hepatic arterial imaging, which, owing to its fleeting presence, requires rapid and accurately timed image acquisition. Multi–detector row CT also affords acquisition of both precontrast and hepatic arterial CT scans during a single breath hold, which potentially allows the performance of high-quality subtraction CT. Owing to the long interphase scanning interval, however, the mean breath hold required in our study was long (35 seconds). While it is reasonable to assume long breath holds should result in more misregistration artifacts, breath-hold length demonstrated no significant correlation with misregistration. Perhaps this relates, in part, to our method of measuring the misregistration. Nevertheless, 21 (40%) of 52 patients had 6 mm or more of misregistration. Acceptable levels of misregistration to optimize lesion detection and characterization have yet to be determined.
To our knowledge, the use of a subtraction technique for liver evaluation with multi–detector row CT has not been previously explored. The subtraction technique is easily performed in less than 5 minutes and can be accomplished by a CT technologist with minimal instruction and without direct guidance by a radiologist. The mean liver-to-muscle contrast ratio, as an estimate of liver contrast, and mean lesion-to-liver contrast ratio were significantly higher on the subtracted images (P < .001). Regardless of the type of tumor, enhancement of hypervascular tumors during the hepatic arterial phase can be subtle.
Our preliminary experience with subtraction CT suggests that the contrast of hypervascular lesions to liver during the hepatic arterial phase can be increased. Whether this manifests as increased conspicuity is yet to be determined. Furthermore, the subtle perfusion abnormalities observed in a number of cases may contribute to greater lesion detection and characterization. It is possible that the improved lesion visualization may allow the contrast media requirements to diminish in both total volume and rate of injection without decreasing lesion conspicuity or detection. This alteration has the potential to improve patient tolerance and safety, especially in the very sick oncology population, and should result in cost savings over time.
A limitation of the study and a selection bias, owing to our referral pattern, was the large number of female patients evaluated for metastatic breast carcinoma. Not only did breast cancer metastases exhibit a variable enhancement pattern, some with internal necrosis, but also relatively few scans were positive for lesions. Although there were only a few cases of hypervascular liver lesions, this study was performed to assess the technical feasibility of the digital subtraction technique with multi–detector row CT. Also, interval growth or pathologic confirmation was not possible in all cases. One could argue that simple subtraction of the mean precontrast attenuation value of the liver from every voxel may be as effective as the voxel-by-voxel subtraction technique investigated in this study. This was not explored, although we believe this technique provides greater lesion-to-liver contrast and allows better evaluation of perfusion abnormalities. Another possible drawback to this study is the fact that the subtracted images, which were acquired with soft-tissue window settings, were not visually compared with images obtained with the dedicated liver window settings for hepatic arterial CT. The difference in window settings would not, however, account for the higher contrast ratios for the subtracted images.
There are a number of issues with respect to our technique of subtraction CT that remain unresolved and require further evaluation. The results of this feasibility study suggest that a much larger comparative trial in patients with a greater variety of lesions, preferably biopsy proved, is warranted to assess the clinical usefulness of this technique. Are lesion conspicuity (not just contrast) and reader confidence increased? A larger study should compare hepatic arterial plus portal venous CT scans versus subtracted plus portal venous CT scans versus subtracted plus hepatic arterial plus portal venous CT scans. The frequency and origin of perilesion perfusion abnormalities needs to be addressed. The factors involved in misregistration also need to be explored along with factors to reduce the breath hold, such as faster gantry rotation and acquisition of more than four sections during each rotation.
Further directions of study include evaluation of the role of this technique in cases of liver metastases from hypovascular tumors. It is possible that subtraction CT may be a useful imaging tool in patients who are treated with one of the new antiangiogenesis agents. Before these agents are used, the technique may help identify hypervascular liver lesions and predict which patients are likely to respond. After treatment with these agents, the technique may help quantify lesion enhancement as a measure of treatment response, in addition to allowing changes in tumor volume to be followed up.
In this preliminary study, we evaluated a technique for subtracting a precontrast CT scan from a CT scan obtained during the hepatic arterial phase of liver enhancement, which were both acquired in a single breath hold with multi–detector row helical CT; the technique is technically feasible, and the images demonstrate substantially greater lesion-to-liver contrast during the hepatic arterial phase. This technique shows promise for elucidating and quantifying the arterial phase enhancement of hypervascular liver tumors. Further evaluation of subtraction CT is needed to validate the initial finding of improved lesion detectability and to determine clinical usefulness.
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION Materials and Methods Results Discussion REFERENCES |
|---|