- Email: [email protected]
Spiral computed tomography of the liver: Contrast agent pharmacokinetics and the potential for improved hepatic enhancement
Spiral Computed Tomography of the Liver: Contrast Agent Pharmacokinetics and the Potential for Improved Hepatic Enhancement Maria Polger, MD 1 , Steven E. Seltzer, MD 1, Bronwyn L. Head, AB 1, Gursel Savci, MD 2, Stuart G. Silverman, MD 1, Douglass F. Adams, MD 1
R a t i o n a l e a n d Objectives. We conducted a prospective study of 131 patients to evaluate the contrast agent dose-response relationship for liver spiral computed tomography (CT) and to test the hypothesis that spiral CT scanning provides greater enhancement than does dynamic CT scanning.
Methods. Patients were assigned to one of two control groups (dynamic CT) or to one of five experimental groups (spiral CT). Dynamic CT patients received 150 ml and spiral CT patients received either 75, 100, or 150 ml of diatrizoate meglumine. All groups had a monophasic injection rate of 2.5 ml/sec. Hepatic enhancement was compared among experimental and control groups. Results. In the experimental groups, there was a linear dose-response relationship (p < .0001) among the enhancements achieved for the three dosages. The enhancement of the last slice of liver for the spiral CT versus dynamic CT groups receiving 150 ml was significantly greater (p = .002). Peak, first liver slice, and average liver enhancement values were higher with spiral CT scanning, but the difference was not statistically significant (power > .55). C o n c l u s i o n . Using uniphasic injection rates and identical doses of contrast agent, spiral CT scanning has the advantage of improved enhancement of the last part of the liver to be imaged.
K e y W o r d s . Computed tomography; helical technology; contrast media. From the 1Department of Radiology, Brigham and Women's Hospital and Harvard Medical School, Boston, MA; and 2Department of Radiology, UIudag University Medical School, Gorukle, Turkey. This work was supported in part by a grant from Siemens Medical Systems. Address reprint requests to M. Polger, MD, Department of Radiology, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115. Received June 22, 1994, and accepted for publication after revision September 13, 1994. Acad Radio11995;2:19-25
© 1994, Association of University Radiologists
t
is generally accepted that the magnitude of hepatic contrast enhance-
I ment is important for liver lesion detection by using computed tomography (CT) scanning [1]. Much effort has been devoted to understanding the pharmacokinetics of intravenous contrast agents in order to optimize delivery protocols for dynamic CT scanning. The goal is to maximize hepatic opacification and liver tumor contrast differences [2-4]. Spiral CT scanning is a new tool for liver imaging that has been proposed as a means to optimize the "temporal window" for data acquisition of contrast-enhanced hepatic CT scanning [5]. The best strategy for contrast agent delivery for hepatic spiral CT scanning remains to be determined. Recent work, how-
19
POLGER E T A L .
Vot. 2, No. 1, January 1995
ever, has focused on the early evaluation of rapid hepatic enhancement associated with high flow rates [6] and comparison of monophasic and biphasic protocols [7]. Preliminary studies on the pancreas and thoracic vasculature suggest that smaller contrast agent doses may be used without loss of enhancement w h e n coupled with spiral CT data acquisition [8, 9]. If the same trends held for the liver, then use of spiral CT could yield greater hepatic enhancement than dynamic CT for a given dose of contrast, Additionally, spiral CT might reduce the amount of contrast media necessary for liver CT without loss of enhancement. However, because liver enhancement is complex and multifactorial, it is not possible to apply the observations made in other tissues to the liver. We sought to establish the value of spiral CT scanning for hepatic contrast enhancement. In this study we attempted to verify that a similar dose-response relationship exists for liver spiral CT as was previously shown for dynamic CT [10] and to test the hypothesis that spiral CT will allow more hepatic enhancement than standard dynamic CT.
shorter delay time that was acquired in a consecutive nonrandomized fashion after the initial investigation. Table 1 and Figure 1 illustrate the details of contrast agent injection and CT scanning time for the groups. Data collection and analysis were performed as follows. Experiment 1 was an investigation of dose-response curve for hepatic enhancement in spiral CT scanning. The relationships between the injected dose of contrast agent (75, 100, or 150 ml) and the magnitude of several hepatic enhancement parameters were tabulated to determine whether there would be a linear d o s e response function for hepatic spiral CT scanning.
Study Design A
751
B
75-
A
lO0-
B
100-
Group 1 ~
-
-
/
[]Injection [] wait [ ] Scan
Group 2
•
Inj + Scan
MATERIALS AND METHODS 150-
Group 3
Experimental Design
Our study was designed as a prospective, randomized, controlled investigation. Consecutive eligible patients were assigned according to their medical record number to one of five experimental groups (n = 92) or to one control group (n = 24). Because there is considerable controversy about the optimal delay time between injection and scanning for dynamic contrastenhanced hepatic CT scanning, we used a second dynamic CT scanning control group (n = 15) with a
1
150 D-
2
150 D-
Control
0
5t0
140
150
FIGURE 1. Graphic depiction of experimental design demonstrating duration of injection (Inj), delay prior to onset of scanning, and duration of data acquisition. D = dynamic.
TABLE 1: Hepatic Enhancement Liver Enhancement (HU) Group No,
Dose (ml)
Delay (see) Peak
First Slice
Last Slice
Average
la lb
75 75
30 60
16 15
38.5±4.1 39.4±3.0
18.4±3.6 a 28.5±3.4
29.4±3.8 37.3±3.0
28.9±3.2 35.4±2.9
2a 2b
100 100
40 60
20 16
44.7±3.5 46.3±3.4
26.9±4.1 34.8&3.8
3
150S
60
25
64.8±4.2
39.0±4.6
38.4±3.8 38.8±4.1 60.7±5.1 b
38.4±4.3 41.3±3.2 59.1±3.9
Control 1 Control 2
150 D 150 D
60 40
24 15
58.7 ± 5.4 59.9 ± 4.3
47.8 ± 5.2 44.0 ± 6.4
42.5!4.6 39.7±3.8
51.7±4.2 46.8±3.7
Liver enhancementvalues are in means ± standa~ errata. HU = Hounsfield units, S = spiral, D = dynamic. a75 (30-see delay) vs 75 (60-see delay); p = .05. b150 S vs Control 1 and Control 2; p = .002.
20
200
Vol. 2, No. 1, January 1995
SPIRAL CT OF THE LIVER
Experiment 2 was a comparison of hepatic enhancements between spiral and conventional dynamic CT scanning. The magnitude of several hepatic enhancement parameters for the experimental groups (spiral CT) were tabulated and compared with the two control groups (dynamic CT). In this way, we could determine whether spiral CT could achieve more hepatic enhancement than conventional CT scanning using 150 ml of contrast agent.
Patient Population The subjects were 62 men and 69 women (age range, 23-83 years; mean, 54.8 years). Data on patient weights were collected retrospectively for 93 patients (71% of the study population). There was no significant difference in weights among the various groups (p = .66; Fig. 2). All patients undergoing contrast-enhanced abdominal CT scanning were eligible for study participation, except for the following: (1) patients who were selected to receive a low-osmolar contrast agent; (2) patients in w h o m a 22gauge or smaller angiocath was needed to achieve venous access (this would not permit the flow rates specified in the study); or (3) patients who were intubated, were on oxygen therapy, or manifested respiratory compromise and therefore were expected to be unable to comply with the 24-sec breath-hold required
Patient Weights I00
75
T
T
I
for spiral CT scanning. All other patients were enrolled and assigned randomly into one of the five experimental groups and the first dynamic CT control group.
Contrast Agent Delivery Protocol Each patient had a 20-gauge or larger angiocath placed in an antecubital vein, and diatrizoate meglumine (Hypaque, Sanofi Winthrop, New York, NY, 282 mg of iodine per milliliter) was delivered via a power injector (Medrad, Pittsburgh, PA) at a monophasic rate of 2.5 ml/ sec. Each patient had a precontrast reference scan through the liver at a level also containing the pancreas. Experimental groups 1, 2, and 3 (undergoing spiral CT scanning) received 75, 100, or 150 ml, respectively, of contrast agent for total iodine doses of 21.2, 28.2, or 42.3 g. Because it was unclear what the optimal delay time between the start of injection and the start of scanning would be when using spiral CT scanning, the patients receiving a reduced contrast agent dose were divided into two subgroups. Patients in groups la, 2a, and 3 (who received the full 150-ml dose) had 30-, 40-, and 60-sec delays, respectively, between the start of injection and the onset of scanning; thus, the end of contrast administration coincided exactly with the initiation of scanning. Patients in groups lb and 2b had a 60-sec delay between injection and scanning, leading to 30- and 20-sec gaps, respectively, between the end of injection and the initiation of scanning. The control groups undergoing dynamic CT scanning received 150 ml of the same contrast agent at the same rate. Patients in control group 1 had a 60-sec delay, so that the end of injection coincided with the start of scanning; in control group 2, all contrast delivery parameters were identical, except the delay was shortened to 40 sec.
T
CT Scanning Protocol :.:,:.:,:,:.: 50
25
0 1A
1B
2A
Experimental Group#
2B
3
1
All scans were performed on a continuously rotating CT scanning system (Siemens Somatom Plus or Plus S, Siemens Medical Systems, Erlangen, Germany). The parameters for the spiral CT technique included 24-sec exposure, 10-mm slice collimation, 10 mm/sec table feed speed, 10-mm reconstruction increment, 125 kVp, 165 mA, and "standard" soft-tissue reconstruction algorithm. The patients were instructed to hyperventilate prior to the prolonged breath-hold, a maneuver that has been shown to help sustain the breath-hold [11]. From each 24-sec exposure, 22 images were obtained because the slices at the top and bottom of the scanned volume contained insufficient data to be reconstructed and were dis-
ii!i!i!i 2
Control Group#
FIGURE 2. Graphic depiction of the distribution of patient weights among the experimental and control groups.
21
POLGER E T AL.
voL 2, No. 1, January 1995
carded. The dynamic incremental CT technique consisted of contiguous 10-mm-thick sections, 1-sec exposure, and 5.5-sec interscan delay, amounting to a 6.5-sec cycle time. For both spiral and dynamic CT scanning, the scans were started at the diaphragm and progressed caudally. The amount of time it took to image an average-sized liver (16 slices) was 104 sec with our dynamic scanning protocol. Among the 92 patients who underwent spiral CT scanning, the liver was completely imaged in the 24-sec acquisition in 89 patients (96.7%). The liver was judged to be completely covered when the last slice no longer contained liver parenchyma.
ment of the last slice containing liver (last slice value minus precontrast value); and (4) the "normalized" area under the hepatic enhancement time--density curve (the area under the curve divided by the number of seconds that the liver was included in the scan; this was functionally equivalent to the average hepatic enhancement value over the course of the scan series). The last slice-to-peak enhancement ratio was also extracted. Data from each patient in each group were averaged. Each hepatic enhancement parameter was also calculated as a function of injected dose per kilogram for the 93 patients for w h o m weights were available.
Image Analysis
Statistical Analyses
To evaluate hepatic enhancements for each slice, we obtained three operator-defined electronic regions of interest (ROIs) from three different homogeneous areas of the liver, avoiding focal lesions, vessels, and beam-hardening or streak artifacts. We attempted to keep the ROI between 0.5 and 1.0 cm 2. From these ROIs, mean precontrast and postcontrast hepatic attenuation values (in Hounsfield units [HU]) were calculated for each section. Time-density curves were constructed for each patient and for each dose group. (Samples of these curves for groups la, 2a, and 3 are provided in Fig. 3.) From these curves, four main hepatic enhancement parameters were extracted: (1) the peak hepatic enhancement level (peak CT attenuation value in HU minus precontrast value); (2) the enhancement of the first slice containing liver (first slice value minus the precontrast value); (3) the enhance-
For Experiment 1, the contrast agent dose-response function was calculated by plotting each hepatic enhancement parameter both as a function of dose and dose per kilogram for the 93 patients whose weights were known and then fitting a regression line. The correlation coefficients for each parameter as a function of dose and as a function of dose per kilogram were calculated. Significance was assessed by using an analysis of variance. For Experiment 2, to test for the significance of differences of the peak, first liver slice, last liver slice, or average liver enhancement measured between experimental (spiral CT) and control (dynamic CT) groups, or among the experimental groups, or between the two control groups, we used an unpaired, two-tailed Student's t test. When differences between the experimental and control groups were not significant, we performed a power calculation to establish the probability of a Type II error.
Liver Time-Density Curves
RESULTS Hepatic Enhancement
T.~
120' 110"
E
100"
e~
90"
DOSe
• .... 7501 O - - 10
•
80"
T
70"
1
---
150 I
60
Time (see) 3. Time-density curves for experimental groups 1a (75 ml, 30-sec delay), 2a (100 ml, 40-sec delay), and 3 (150 ml, 60-sec delay). HU = Hounsfield units. FIGURE
22
The mean values for the peak, first liver slice, last liver slice, and average liver enhancement for the experimental and control groups are shown in Table 1. In Experiment 1, the dose-response curve for spiral CT hepatic enhancement was linear (p < .0001). This linear relationship held when the peak, first slice, last slice, and average liver enhancement values were expressed as a function of dose per kilogram or as a function of dose alone (i.e., milliliter of contrast agent injected). Figure 4 shows the average enhancement plotted on a dose-per-kilogram basis. The correlation coefficients relating the four enhancement parameters to dose per kilogram were slightly higher than the correlations relating them to dose alone, but these differences were not
Vol. 2, No. 1, January 1995
SPIRAL CT OF THE LIVER
statistically significant (p > .40; Table 2). Peak hepatic enhancement values ranged from 38.5 HU for a 75-ml dose to 64.8 HU for 150 ml. An average of 0.47 HU of peak hepatic enhancement was achieved per milliliter of contrast agent injected, corresponding to 1.67 HU per y = 17.497X + 15.635
= 0.526
r
100 []
[]
tD
o
o
[] [] 25
[]
gram of iodine. First liver slice enhancement ranged from 18.4 HU for a 75-ml dose (30-sec delay) to 39.0 HU for 150 ml. Enhancement remaining on the last slice containing liver ranged from 29.4 to 60.7 HU along the dose range. The average hepatic enhancement ranged from 28.9 to 59.1 HU. Although all four enhancement parameters were slightly higher for groups lb and 2b (the subgroups with 60-sec delays between injection and scanning) than for the corresponding groups la and lb (the subgroups with shorter delays), the only comparison that achieved statistical significance was the difference in first slice enhancement (p = .05). Therefore, the data for the subgroups la and l b were combined, as were subgroups 2a and 2b, to simplify graphic portrayal of the dose-response curves (Fig. 5). In Experiment 2, the main difference observed between the enhancement values achieved with spiral
TABLE 2: Correlation Coefficients
rl[] 0
Correlation as a Function of
'
i
,
,
i
I
0.5
1.0
1.5
2.0
2.5
3.0
Measure
3.5
Dose/kg FIGURE 4. Dose-response curve for the average hepatic enhancement on a dose-per-kilogram basis. AHU = difference in computed tomography value.
Peak Lastliverslice Average Firstliverslice
70
"~
Kilogram (ml/kg)
.45 .32 .46 .33
.57 .47 .53 .44
X2
P
1.16 1.40 0.44 0.71
>.2 >.2 >.4 =.4
60-
5"
50 4O
50 40
i== 3c ~-
Dose per
(ml)
70-
60-
G
Dose
3c
2( 1 0-S
iO-S
Average
A
Average
B
FIGURE 5. Graphic portrayal of the dose-response curves. A, Enhancement of the experimental groups undergoing spiral computed tomography (CT) scanning, combining subgroups laJl b and 2a/2b. B, Comparison of the enhancement of the 150-ml spiral CT experimental group and the two 150-ml dynamic CT control groups. ~HU = difference in CT value, S = spiral, D = dynamic.
23
POLGER ET AL.
CT and dynamic CT scanning was in the enhancement remaining on the last hepatic slice. For the 150-ml dose, this enhancement was 60.7 HU for spiral CT scanning and was significantly higher than the 42.5 HU for dynamic CT control group 1 (p = .011) and the 39.7 HU for dynamic CT control group 2 (p = .007), or both control groups combined (p = .002). For all three spiral CT doses, the enhancement remaining on the last slice averaged 87.7% of the peak enhancement compared with 72.4% for dynamic CT control group 1 and 66.0% for dynamic CT control group 2. Although the peak, first slice, and average liver enhancement values were higher for the 150-ml dose for spiral CT than for the two control groups, these differences did not reach statistical significance (p = .32, p o w e r = .83, for peak; p = .25, power = .79, for first slice; and p = .07, power = .55, for average). The same observations were valid w h e n enhancement was expressed as a function of dose per kilogram. Only the difference between the last slice enhancement was significant (p < .01). Extrapolating from the linear spiral CT dose-response curve, we predicted the dose of contrast agent needed in spiral CT scanning that would achieve the same amount of hepatic enhancement as for the average patient in our two dynamic CT protocols. These spiral equivalent doses were 136.8, 105.0, and 126.4 ml for the peak, last slice, and average enhancement parameters, respectively. These values corresponded to dose reductions of 8.8-30.0% from dynamic CT scanning. DISCUSSION
The magnitude of hepatic enhancement has been shown to depend on the dose of contrast agent administered, the rate of injection, and patient weight [12-14]. Our results confirm that the linear dose-response relationship already established for dynamic CT scanning applies to spiral CT scanning as well. Although the data showed a trend for higher hepatic enhancement for spiral CT scanning, the difference did not reach statistical significance except for the last slice enhancement, wherein a 30% greater enhancement was observed. There continue to be differences of opinion about whether contrast material used in abdominal CT scanning is best delivered as a standardized amount (e.g., 150 ml for all patients, which is our routine clinical practice) or whether it should be adjusted according to patient weight (e.g., 2 ml/kg) [15]. Because of this continued controversy, we chose to analyze our hepatic enhancement data both ways, evaluating enhancement
24
Vol. 2, No. 1, January 1995
as functions of both contrast agent volume and dose per kilogram. The data favoring spiral CT scanning were the same for both types of analyses. There is also controversy about the optimum delay time between contrast agent injection and the initiation of dynamic scanning. Therefore, we established two control groups with delay times spanning the range of what is considered appropriate. We found no differences in the enhancement values between the two control groups and no difference in the trends favoring the spiral CT groups over either control group. The optimal delay prior to onset of scanning with spiral CT technique could not be directly determined from our data. We were surprised to find no significant difference in the peak, last, or average hepatic enhancement values between 30- and 60-sec delays separating the start of injection and scanning for the 75-ml dose. Similarly, there were no significant differences in these parameters between the 40- and 60-sec delays for the 100-ml dose. However, because the 30-sec delay yielded first slice enhancements that were less than those achieved with a 60-sec delay for the groups receiving 75 ml of contrast agent, the 30-sec delays are probably suboptimal. In addition, Heiken et al. [16] found that when they used a contrast agent volume of 125 ml (320 mg of iodine per milliliter), it took a mean of 58 sec to achieve an enhancement threshold of 40 HU in their uniphasic low-flow pro-" tocol (2.5 ml/sec) and 64 sec to achieve an enhancement threshold of 50 HU. A 60-sec delay is therefore probably ideal to use with spiral hepatic CT scanning. The finding that spiral CT scanning allows more last slice enhancement is important because, w h e n using dynamic CT scanning and starting at the diaphragm, it is the caudal aspect of the liver that is more apt to be imaged during the equilibrium phase, which occurs at approximately 1.5-3 min after the start of injection of the contrast material bolus, depending on the contrast agent delivery protocol [7, 16-19]. During this late phase, when intravascular and extravascular contrast material equilibrate because of diffusion of intravascular contrast material into the extravascular space, live>to-lesion contrast is decreased. Isoattenuating metastases occur w h e n the extracellular iodine concentrations of liver and lesion reach similar levels. Use of a properly placed narrow data acquisition time window available with spiral CT scanning should allow hepatic scanning to be completed before the equilibrium phase. In our patients with a 24sec spiral CT data acquisition and 30- to 60-sec delays, the entire liver (in 89 of 92 patients) was imaged in 54-
VoI. 2, NO. 1, January 1995
84 sec, well in advance of the expected onset of equilibrium. Heiken et al. [16] showed that using an injection protocol almost identical to ours (uniphasic low flow, 125 ml ioversol, 320 mg of iodine per milliliter at 2.5 ml/ sec), the mean time to achieve equilibrium was 92 sec [16]. Foley et al. also reported that using a similar injection protocol (uniphasic, 180 ml diatrizoate, 300 mg iodine per milliliter at 3.0 ml/sec) equilibrium began at 95.1 sec [7]. Therefore, using these data, a 60-sec delay coupled with a 24-sec spiral CT scanning should allow the entire liver to be consistently imaged prior to the onset of equilibrium. Heiken et al. also showed that using enhancement thresholds of 50-60 HU, the optimal scanning interval, defined by the length of time between the onset of a desired level of hepatic enhancement and either the loss of that enhancement or the onset of equilibrium was about 30 sec for their uniphasic low-flow protocol. Theoretically, with proper timing of the 24-sec spiral data acquisition, one could obtain the entire scan above 50 HU of enhancement. Future work may focus on adjusting the contrast agent injection rate to optimize the dosing regimen specifically for spiral CT scanning. Heiken et al. [16] reported that the use of faster injection rates (5.0 ml/sec) boosted hepatic enhancement on dynamic CT scanning. The benefits of faster injection rates may be even more substantial for spiral CT scanning. Because of the narrow temporal window of spiral CT scanning, using it with a fast injection would allow one to garner all of the benefits of a fast rate and still complete the data acquisition before equilibrium begins. A disadvantage of using a fast injection rate is that the narrow shape of the resulting time-density curve means that there is little margin for error in setting the delay between injection and scanning. A faster injection rate will also likely require a shorter delay time. Our study has some limitations. First, we used highosmolar contrast agents exclusively. However, the same general results would be expected for low-osmolar contrast agents. Second, because we were limited to an exposure time of 24 sec, we could observe only a portion of the entire hepatic time-density curve. Third, we did not collect vascular enhancement data. These data are important determinants of the onset of the equilibrium phase. However, Heiken et al. [16] and Foley et al. [.7]studied vascular enhancement in detail and established clearly that using the contrast agent delivery parameters in our study, equilibrium begins 90 sec after contrast injection. Because we believed this fact to be well established, we chose to
SPIRAL CT OF THE LIVER
focus our attention on the liver. Fourth, we limited our investigation to a monophasic injection rate. The difference between biphasic and monophasic injection rates is likely to be less significant in spiral CT scanning [16]. Finally, because we evaluated only liver parenchymal enhancement, lesion conspicuity was not assessed. We conclude that a linear dose-response relationship exists for hepatic spiral CT scanning. An important advantage of spiral over conventional CT scanning for liver imaging is that it covers the entire liver faster, leading to improved enhancement of the last part of the liver to be imaged. REFERENCES 1. Anderson DJ, Berland L. CT techniques. In: Lee JK, Sagel SS, Stanley RJ, eds. Computed body tomography with MRI correlation, 2nd ed. New York: Raven Press, 1989:47-50. 2. Marchar GJ, Baert AL, Wilms GE. CT of ncncystic liver lesions: bolus enhancement. AJR 1980;135:57-65. 3. Young SW, Turner RJ, Castellino RA. A strategy for the contrast enhancement of malignant tumors using dynamic computed tomography and intravascular pharmacokinetics. Radiology1980;137:137-147. 4. Alpern MB, Lawson TL, Foley WD, et aL Focal hepatic masses and fatty infiltration detected by enhanced dynamic CT. Radiology 1986;158:45-49. 5. Cox IH, Foley WD, Hoffman RG. Right window for dynamic hepatic CT. Radiology 1991; 181:18-21. 6. Small WC, Nelson RC, Bernardino ME, Brummer LT. Contrast-enhanced spiral CT of the liver: effect of different amounts and injection rates of contrast material on early contrast enhancement. AJR 1994; 163:87-92. 7. Foley WD, Hoffmann RG, Quiroz FA, Kahn CE, Perret RS. Hepatic helical CT: contrast material injection protocol. Radiology 1994; 192:367-371. 8. Dupuy DE, Costello P, Ecker CP. Spiral CT of the pancreas. Radiology 1992; 183:815-818. 9. Costerlo P, Dupuy DE, Ecker CP, Tello R. Spiral CT of the thorax with reduced volume of contrast material: a comparative study. Radiology 1992;183:663-666. 10. Foley WD, Berland LL, Lawson TL, Smith DF, Thorsen MK. Contrast enhancement technique for dynamic hepatic computed tomographic scanning. Radiology1983;147:797-803. 11. Shaffer K, Pugatch RD. Small pulmonary nodules: dynamic CT with a single-breath technique. Radiology 1989;173:587-568. 12. Dean PB, Violante MR, Mahoney JA. Hepatic CT contrast enhancement: effect of dose, duration of infusion, and time elapsed following infusion. Invest Radiol 1980; 15:158-161. 13. Kormano M, Partanen K, Soimakalrio S, Kivimaki T. Dynamic contrast enhancement of the upper abdomen; effect of contrast medium and body weight. Invest Radio11983; 18:364-367. 14. Claussen CD, Banzer D, Pkretzschner C, Kalender WA, Schorner W. Bolus geometry and dynamics after intravenous contrast medium injection. Radiology 1984; 153:365-368. 15. Zeman RK, Clements LA, Silverman PM, et al. CT of the liver: a survey of prevailing methods for administration of contrast material. AJR 1988; 150:107-109. 16. Heiken JP, Brink JA, McCrennan BL, Sagel SS, Forman HP, DiCroce J. Dynamic contrast-enhanced CT of the liver: comparison of contrast medium injection rates and uniphasic and biphasic injection protocols. Radiology 1993; 187:327-331. 17. Foley WD. Dynamic hepatic CT. Radiology 1989;170:617-622. 18. Kormano M, Dean PB. Extravascular contrast material: the major component of contrast enhancement. Radiology 1976;121:379-382. 19. Burgener FA, Hamlin DJ. Contrast enhancement in abdominal CT: bolus vs infusion. AJR 1981; 137:351-358.
25
R a t i o n a l e a n d Objectives. We conducted a prospective study of 131 patients to evaluate the contrast agent dose-response relationship for liver spiral computed tomography (CT) and to test the hypothesis that spiral CT scanning provides greater enhancement than does dynamic CT scanning.
Methods. Patients were assigned to one of two control groups (dynamic CT) or to one of five experimental groups (spiral CT). Dynamic CT patients received 150 ml and spiral CT patients received either 75, 100, or 150 ml of diatrizoate meglumine. All groups had a monophasic injection rate of 2.5 ml/sec. Hepatic enhancement was compared among experimental and control groups. Results. In the experimental groups, there was a linear dose-response relationship (p < .0001) among the enhancements achieved for the three dosages. The enhancement of the last slice of liver for the spiral CT versus dynamic CT groups receiving 150 ml was significantly greater (p = .002). Peak, first liver slice, and average liver enhancement values were higher with spiral CT scanning, but the difference was not statistically significant (power > .55). C o n c l u s i o n . Using uniphasic injection rates and identical doses of contrast agent, spiral CT scanning has the advantage of improved enhancement of the last part of the liver to be imaged.
K e y W o r d s . Computed tomography; helical technology; contrast media. From the 1Department of Radiology, Brigham and Women's Hospital and Harvard Medical School, Boston, MA; and 2Department of Radiology, UIudag University Medical School, Gorukle, Turkey. This work was supported in part by a grant from Siemens Medical Systems. Address reprint requests to M. Polger, MD, Department of Radiology, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115. Received June 22, 1994, and accepted for publication after revision September 13, 1994. Acad Radio11995;2:19-25
© 1994, Association of University Radiologists
t
is generally accepted that the magnitude of hepatic contrast enhance-
I ment is important for liver lesion detection by using computed tomography (CT) scanning [1]. Much effort has been devoted to understanding the pharmacokinetics of intravenous contrast agents in order to optimize delivery protocols for dynamic CT scanning. The goal is to maximize hepatic opacification and liver tumor contrast differences [2-4]. Spiral CT scanning is a new tool for liver imaging that has been proposed as a means to optimize the "temporal window" for data acquisition of contrast-enhanced hepatic CT scanning [5]. The best strategy for contrast agent delivery for hepatic spiral CT scanning remains to be determined. Recent work, how-
19
POLGER E T A L .
Vot. 2, No. 1, January 1995
ever, has focused on the early evaluation of rapid hepatic enhancement associated with high flow rates [6] and comparison of monophasic and biphasic protocols [7]. Preliminary studies on the pancreas and thoracic vasculature suggest that smaller contrast agent doses may be used without loss of enhancement w h e n coupled with spiral CT data acquisition [8, 9]. If the same trends held for the liver, then use of spiral CT could yield greater hepatic enhancement than dynamic CT for a given dose of contrast, Additionally, spiral CT might reduce the amount of contrast media necessary for liver CT without loss of enhancement. However, because liver enhancement is complex and multifactorial, it is not possible to apply the observations made in other tissues to the liver. We sought to establish the value of spiral CT scanning for hepatic contrast enhancement. In this study we attempted to verify that a similar dose-response relationship exists for liver spiral CT as was previously shown for dynamic CT [10] and to test the hypothesis that spiral CT will allow more hepatic enhancement than standard dynamic CT.
shorter delay time that was acquired in a consecutive nonrandomized fashion after the initial investigation. Table 1 and Figure 1 illustrate the details of contrast agent injection and CT scanning time for the groups. Data collection and analysis were performed as follows. Experiment 1 was an investigation of dose-response curve for hepatic enhancement in spiral CT scanning. The relationships between the injected dose of contrast agent (75, 100, or 150 ml) and the magnitude of several hepatic enhancement parameters were tabulated to determine whether there would be a linear d o s e response function for hepatic spiral CT scanning.
Study Design A
751
B
75-
A
lO0-
B
100-
Group 1 ~
-
-
/
[]Injection [] wait [ ] Scan
Group 2
•
Inj + Scan
MATERIALS AND METHODS 150-
Group 3
Experimental Design
Our study was designed as a prospective, randomized, controlled investigation. Consecutive eligible patients were assigned according to their medical record number to one of five experimental groups (n = 92) or to one control group (n = 24). Because there is considerable controversy about the optimal delay time between injection and scanning for dynamic contrastenhanced hepatic CT scanning, we used a second dynamic CT scanning control group (n = 15) with a
1
150 D-
2
150 D-
Control
0
5t0
140
150
FIGURE 1. Graphic depiction of experimental design demonstrating duration of injection (Inj), delay prior to onset of scanning, and duration of data acquisition. D = dynamic.
TABLE 1: Hepatic Enhancement Liver Enhancement (HU) Group No,
Dose (ml)
Delay (see) Peak
First Slice
Last Slice
Average
la lb
75 75
30 60
16 15
38.5±4.1 39.4±3.0
18.4±3.6 a 28.5±3.4
29.4±3.8 37.3±3.0
28.9±3.2 35.4±2.9
2a 2b
100 100
40 60
20 16
44.7±3.5 46.3±3.4
26.9±4.1 34.8&3.8
3
150S
60
25
64.8±4.2
39.0±4.6
38.4±3.8 38.8±4.1 60.7±5.1 b
38.4±4.3 41.3±3.2 59.1±3.9
Control 1 Control 2
150 D 150 D
60 40
24 15
58.7 ± 5.4 59.9 ± 4.3
47.8 ± 5.2 44.0 ± 6.4
42.5!4.6 39.7±3.8
51.7±4.2 46.8±3.7
Liver enhancementvalues are in means ± standa~ errata. HU = Hounsfield units, S = spiral, D = dynamic. a75 (30-see delay) vs 75 (60-see delay); p = .05. b150 S vs Control 1 and Control 2; p = .002.
20
200
Vol. 2, No. 1, January 1995
SPIRAL CT OF THE LIVER
Experiment 2 was a comparison of hepatic enhancements between spiral and conventional dynamic CT scanning. The magnitude of several hepatic enhancement parameters for the experimental groups (spiral CT) were tabulated and compared with the two control groups (dynamic CT). In this way, we could determine whether spiral CT could achieve more hepatic enhancement than conventional CT scanning using 150 ml of contrast agent.
Patient Population The subjects were 62 men and 69 women (age range, 23-83 years; mean, 54.8 years). Data on patient weights were collected retrospectively for 93 patients (71% of the study population). There was no significant difference in weights among the various groups (p = .66; Fig. 2). All patients undergoing contrast-enhanced abdominal CT scanning were eligible for study participation, except for the following: (1) patients who were selected to receive a low-osmolar contrast agent; (2) patients in w h o m a 22gauge or smaller angiocath was needed to achieve venous access (this would not permit the flow rates specified in the study); or (3) patients who were intubated, were on oxygen therapy, or manifested respiratory compromise and therefore were expected to be unable to comply with the 24-sec breath-hold required
Patient Weights I00
75
T
T
I
for spiral CT scanning. All other patients were enrolled and assigned randomly into one of the five experimental groups and the first dynamic CT control group.
Contrast Agent Delivery Protocol Each patient had a 20-gauge or larger angiocath placed in an antecubital vein, and diatrizoate meglumine (Hypaque, Sanofi Winthrop, New York, NY, 282 mg of iodine per milliliter) was delivered via a power injector (Medrad, Pittsburgh, PA) at a monophasic rate of 2.5 ml/ sec. Each patient had a precontrast reference scan through the liver at a level also containing the pancreas. Experimental groups 1, 2, and 3 (undergoing spiral CT scanning) received 75, 100, or 150 ml, respectively, of contrast agent for total iodine doses of 21.2, 28.2, or 42.3 g. Because it was unclear what the optimal delay time between the start of injection and the start of scanning would be when using spiral CT scanning, the patients receiving a reduced contrast agent dose were divided into two subgroups. Patients in groups la, 2a, and 3 (who received the full 150-ml dose) had 30-, 40-, and 60-sec delays, respectively, between the start of injection and the onset of scanning; thus, the end of contrast administration coincided exactly with the initiation of scanning. Patients in groups lb and 2b had a 60-sec delay between injection and scanning, leading to 30- and 20-sec gaps, respectively, between the end of injection and the initiation of scanning. The control groups undergoing dynamic CT scanning received 150 ml of the same contrast agent at the same rate. Patients in control group 1 had a 60-sec delay, so that the end of injection coincided with the start of scanning; in control group 2, all contrast delivery parameters were identical, except the delay was shortened to 40 sec.
T
CT Scanning Protocol :.:,:.:,:,:.: 50
25
0 1A
1B
2A
Experimental Group#
2B
3
1
All scans were performed on a continuously rotating CT scanning system (Siemens Somatom Plus or Plus S, Siemens Medical Systems, Erlangen, Germany). The parameters for the spiral CT technique included 24-sec exposure, 10-mm slice collimation, 10 mm/sec table feed speed, 10-mm reconstruction increment, 125 kVp, 165 mA, and "standard" soft-tissue reconstruction algorithm. The patients were instructed to hyperventilate prior to the prolonged breath-hold, a maneuver that has been shown to help sustain the breath-hold [11]. From each 24-sec exposure, 22 images were obtained because the slices at the top and bottom of the scanned volume contained insufficient data to be reconstructed and were dis-
ii!i!i!i 2
Control Group#
FIGURE 2. Graphic depiction of the distribution of patient weights among the experimental and control groups.
21
POLGER E T AL.
voL 2, No. 1, January 1995
carded. The dynamic incremental CT technique consisted of contiguous 10-mm-thick sections, 1-sec exposure, and 5.5-sec interscan delay, amounting to a 6.5-sec cycle time. For both spiral and dynamic CT scanning, the scans were started at the diaphragm and progressed caudally. The amount of time it took to image an average-sized liver (16 slices) was 104 sec with our dynamic scanning protocol. Among the 92 patients who underwent spiral CT scanning, the liver was completely imaged in the 24-sec acquisition in 89 patients (96.7%). The liver was judged to be completely covered when the last slice no longer contained liver parenchyma.
ment of the last slice containing liver (last slice value minus precontrast value); and (4) the "normalized" area under the hepatic enhancement time--density curve (the area under the curve divided by the number of seconds that the liver was included in the scan; this was functionally equivalent to the average hepatic enhancement value over the course of the scan series). The last slice-to-peak enhancement ratio was also extracted. Data from each patient in each group were averaged. Each hepatic enhancement parameter was also calculated as a function of injected dose per kilogram for the 93 patients for w h o m weights were available.
Image Analysis
Statistical Analyses
To evaluate hepatic enhancements for each slice, we obtained three operator-defined electronic regions of interest (ROIs) from three different homogeneous areas of the liver, avoiding focal lesions, vessels, and beam-hardening or streak artifacts. We attempted to keep the ROI between 0.5 and 1.0 cm 2. From these ROIs, mean precontrast and postcontrast hepatic attenuation values (in Hounsfield units [HU]) were calculated for each section. Time-density curves were constructed for each patient and for each dose group. (Samples of these curves for groups la, 2a, and 3 are provided in Fig. 3.) From these curves, four main hepatic enhancement parameters were extracted: (1) the peak hepatic enhancement level (peak CT attenuation value in HU minus precontrast value); (2) the enhancement of the first slice containing liver (first slice value minus the precontrast value); (3) the enhance-
For Experiment 1, the contrast agent dose-response function was calculated by plotting each hepatic enhancement parameter both as a function of dose and dose per kilogram for the 93 patients whose weights were known and then fitting a regression line. The correlation coefficients for each parameter as a function of dose and as a function of dose per kilogram were calculated. Significance was assessed by using an analysis of variance. For Experiment 2, to test for the significance of differences of the peak, first liver slice, last liver slice, or average liver enhancement measured between experimental (spiral CT) and control (dynamic CT) groups, or among the experimental groups, or between the two control groups, we used an unpaired, two-tailed Student's t test. When differences between the experimental and control groups were not significant, we performed a power calculation to establish the probability of a Type II error.
Liver Time-Density Curves
RESULTS Hepatic Enhancement
T.~
120' 110"
E
100"
e~
90"
DOSe
• .... 7501 O - - 10
•
80"
T
70"
1
---
150 I
60
Time (see) 3. Time-density curves for experimental groups 1a (75 ml, 30-sec delay), 2a (100 ml, 40-sec delay), and 3 (150 ml, 60-sec delay). HU = Hounsfield units. FIGURE
22
The mean values for the peak, first liver slice, last liver slice, and average liver enhancement for the experimental and control groups are shown in Table 1. In Experiment 1, the dose-response curve for spiral CT hepatic enhancement was linear (p < .0001). This linear relationship held when the peak, first slice, last slice, and average liver enhancement values were expressed as a function of dose per kilogram or as a function of dose alone (i.e., milliliter of contrast agent injected). Figure 4 shows the average enhancement plotted on a dose-per-kilogram basis. The correlation coefficients relating the four enhancement parameters to dose per kilogram were slightly higher than the correlations relating them to dose alone, but these differences were not
Vol. 2, No. 1, January 1995
SPIRAL CT OF THE LIVER
statistically significant (p > .40; Table 2). Peak hepatic enhancement values ranged from 38.5 HU for a 75-ml dose to 64.8 HU for 150 ml. An average of 0.47 HU of peak hepatic enhancement was achieved per milliliter of contrast agent injected, corresponding to 1.67 HU per y = 17.497X + 15.635
= 0.526
r
100 []
[]
tD
o
o
[] [] 25
[]
gram of iodine. First liver slice enhancement ranged from 18.4 HU for a 75-ml dose (30-sec delay) to 39.0 HU for 150 ml. Enhancement remaining on the last slice containing liver ranged from 29.4 to 60.7 HU along the dose range. The average hepatic enhancement ranged from 28.9 to 59.1 HU. Although all four enhancement parameters were slightly higher for groups lb and 2b (the subgroups with 60-sec delays between injection and scanning) than for the corresponding groups la and lb (the subgroups with shorter delays), the only comparison that achieved statistical significance was the difference in first slice enhancement (p = .05). Therefore, the data for the subgroups la and l b were combined, as were subgroups 2a and 2b, to simplify graphic portrayal of the dose-response curves (Fig. 5). In Experiment 2, the main difference observed between the enhancement values achieved with spiral
TABLE 2: Correlation Coefficients
rl[] 0
Correlation as a Function of
'
i
,
,
i
I
0.5
1.0
1.5
2.0
2.5
3.0
Measure
3.5
Dose/kg FIGURE 4. Dose-response curve for the average hepatic enhancement on a dose-per-kilogram basis. AHU = difference in computed tomography value.
Peak Lastliverslice Average Firstliverslice
70
"~
Kilogram (ml/kg)
.45 .32 .46 .33
.57 .47 .53 .44
X2
P
1.16 1.40 0.44 0.71
>.2 >.2 >.4 =.4
60-
5"
50 4O
50 40
i== 3c ~-
Dose per
(ml)
70-
60-
G
Dose
3c
2( 1 0-S
iO-S
Average
A
Average
B
FIGURE 5. Graphic portrayal of the dose-response curves. A, Enhancement of the experimental groups undergoing spiral computed tomography (CT) scanning, combining subgroups laJl b and 2a/2b. B, Comparison of the enhancement of the 150-ml spiral CT experimental group and the two 150-ml dynamic CT control groups. ~HU = difference in CT value, S = spiral, D = dynamic.
23
POLGER ET AL.
CT and dynamic CT scanning was in the enhancement remaining on the last hepatic slice. For the 150-ml dose, this enhancement was 60.7 HU for spiral CT scanning and was significantly higher than the 42.5 HU for dynamic CT control group 1 (p = .011) and the 39.7 HU for dynamic CT control group 2 (p = .007), or both control groups combined (p = .002). For all three spiral CT doses, the enhancement remaining on the last slice averaged 87.7% of the peak enhancement compared with 72.4% for dynamic CT control group 1 and 66.0% for dynamic CT control group 2. Although the peak, first slice, and average liver enhancement values were higher for the 150-ml dose for spiral CT than for the two control groups, these differences did not reach statistical significance (p = .32, p o w e r = .83, for peak; p = .25, power = .79, for first slice; and p = .07, power = .55, for average). The same observations were valid w h e n enhancement was expressed as a function of dose per kilogram. Only the difference between the last slice enhancement was significant (p < .01). Extrapolating from the linear spiral CT dose-response curve, we predicted the dose of contrast agent needed in spiral CT scanning that would achieve the same amount of hepatic enhancement as for the average patient in our two dynamic CT protocols. These spiral equivalent doses were 136.8, 105.0, and 126.4 ml for the peak, last slice, and average enhancement parameters, respectively. These values corresponded to dose reductions of 8.8-30.0% from dynamic CT scanning. DISCUSSION
The magnitude of hepatic enhancement has been shown to depend on the dose of contrast agent administered, the rate of injection, and patient weight [12-14]. Our results confirm that the linear dose-response relationship already established for dynamic CT scanning applies to spiral CT scanning as well. Although the data showed a trend for higher hepatic enhancement for spiral CT scanning, the difference did not reach statistical significance except for the last slice enhancement, wherein a 30% greater enhancement was observed. There continue to be differences of opinion about whether contrast material used in abdominal CT scanning is best delivered as a standardized amount (e.g., 150 ml for all patients, which is our routine clinical practice) or whether it should be adjusted according to patient weight (e.g., 2 ml/kg) [15]. Because of this continued controversy, we chose to analyze our hepatic enhancement data both ways, evaluating enhancement
24
Vol. 2, No. 1, January 1995
as functions of both contrast agent volume and dose per kilogram. The data favoring spiral CT scanning were the same for both types of analyses. There is also controversy about the optimum delay time between contrast agent injection and the initiation of dynamic scanning. Therefore, we established two control groups with delay times spanning the range of what is considered appropriate. We found no differences in the enhancement values between the two control groups and no difference in the trends favoring the spiral CT groups over either control group. The optimal delay prior to onset of scanning with spiral CT technique could not be directly determined from our data. We were surprised to find no significant difference in the peak, last, or average hepatic enhancement values between 30- and 60-sec delays separating the start of injection and scanning for the 75-ml dose. Similarly, there were no significant differences in these parameters between the 40- and 60-sec delays for the 100-ml dose. However, because the 30-sec delay yielded first slice enhancements that were less than those achieved with a 60-sec delay for the groups receiving 75 ml of contrast agent, the 30-sec delays are probably suboptimal. In addition, Heiken et al. [16] found that when they used a contrast agent volume of 125 ml (320 mg of iodine per milliliter), it took a mean of 58 sec to achieve an enhancement threshold of 40 HU in their uniphasic low-flow pro-" tocol (2.5 ml/sec) and 64 sec to achieve an enhancement threshold of 50 HU. A 60-sec delay is therefore probably ideal to use with spiral hepatic CT scanning. The finding that spiral CT scanning allows more last slice enhancement is important because, w h e n using dynamic CT scanning and starting at the diaphragm, it is the caudal aspect of the liver that is more apt to be imaged during the equilibrium phase, which occurs at approximately 1.5-3 min after the start of injection of the contrast material bolus, depending on the contrast agent delivery protocol [7, 16-19]. During this late phase, when intravascular and extravascular contrast material equilibrate because of diffusion of intravascular contrast material into the extravascular space, live>to-lesion contrast is decreased. Isoattenuating metastases occur w h e n the extracellular iodine concentrations of liver and lesion reach similar levels. Use of a properly placed narrow data acquisition time window available with spiral CT scanning should allow hepatic scanning to be completed before the equilibrium phase. In our patients with a 24sec spiral CT data acquisition and 30- to 60-sec delays, the entire liver (in 89 of 92 patients) was imaged in 54-
VoI. 2, NO. 1, January 1995
84 sec, well in advance of the expected onset of equilibrium. Heiken et al. [16] showed that using an injection protocol almost identical to ours (uniphasic low flow, 125 ml ioversol, 320 mg of iodine per milliliter at 2.5 ml/ sec), the mean time to achieve equilibrium was 92 sec [16]. Foley et al. also reported that using a similar injection protocol (uniphasic, 180 ml diatrizoate, 300 mg iodine per milliliter at 3.0 ml/sec) equilibrium began at 95.1 sec [7]. Therefore, using these data, a 60-sec delay coupled with a 24-sec spiral CT scanning should allow the entire liver to be consistently imaged prior to the onset of equilibrium. Heiken et al. also showed that using enhancement thresholds of 50-60 HU, the optimal scanning interval, defined by the length of time between the onset of a desired level of hepatic enhancement and either the loss of that enhancement or the onset of equilibrium was about 30 sec for their uniphasic low-flow protocol. Theoretically, with proper timing of the 24-sec spiral data acquisition, one could obtain the entire scan above 50 HU of enhancement. Future work may focus on adjusting the contrast agent injection rate to optimize the dosing regimen specifically for spiral CT scanning. Heiken et al. [16] reported that the use of faster injection rates (5.0 ml/sec) boosted hepatic enhancement on dynamic CT scanning. The benefits of faster injection rates may be even more substantial for spiral CT scanning. Because of the narrow temporal window of spiral CT scanning, using it with a fast injection would allow one to garner all of the benefits of a fast rate and still complete the data acquisition before equilibrium begins. A disadvantage of using a fast injection rate is that the narrow shape of the resulting time-density curve means that there is little margin for error in setting the delay between injection and scanning. A faster injection rate will also likely require a shorter delay time. Our study has some limitations. First, we used highosmolar contrast agents exclusively. However, the same general results would be expected for low-osmolar contrast agents. Second, because we were limited to an exposure time of 24 sec, we could observe only a portion of the entire hepatic time-density curve. Third, we did not collect vascular enhancement data. These data are important determinants of the onset of the equilibrium phase. However, Heiken et al. [16] and Foley et al. [.7]studied vascular enhancement in detail and established clearly that using the contrast agent delivery parameters in our study, equilibrium begins 90 sec after contrast injection. Because we believed this fact to be well established, we chose to
SPIRAL CT OF THE LIVER
focus our attention on the liver. Fourth, we limited our investigation to a monophasic injection rate. The difference between biphasic and monophasic injection rates is likely to be less significant in spiral CT scanning [16]. Finally, because we evaluated only liver parenchymal enhancement, lesion conspicuity was not assessed. We conclude that a linear dose-response relationship exists for hepatic spiral CT scanning. An important advantage of spiral over conventional CT scanning for liver imaging is that it covers the entire liver faster, leading to improved enhancement of the last part of the liver to be imaged. REFERENCES 1. Anderson DJ, Berland L. CT techniques. In: Lee JK, Sagel SS, Stanley RJ, eds. Computed body tomography with MRI correlation, 2nd ed. New York: Raven Press, 1989:47-50. 2. Marchar GJ, Baert AL, Wilms GE. CT of ncncystic liver lesions: bolus enhancement. AJR 1980;135:57-65. 3. Young SW, Turner RJ, Castellino RA. A strategy for the contrast enhancement of malignant tumors using dynamic computed tomography and intravascular pharmacokinetics. Radiology1980;137:137-147. 4. Alpern MB, Lawson TL, Foley WD, et aL Focal hepatic masses and fatty infiltration detected by enhanced dynamic CT. Radiology 1986;158:45-49. 5. Cox IH, Foley WD, Hoffman RG. Right window for dynamic hepatic CT. Radiology 1991; 181:18-21. 6. Small WC, Nelson RC, Bernardino ME, Brummer LT. Contrast-enhanced spiral CT of the liver: effect of different amounts and injection rates of contrast material on early contrast enhancement. AJR 1994; 163:87-92. 7. Foley WD, Hoffmann RG, Quiroz FA, Kahn CE, Perret RS. Hepatic helical CT: contrast material injection protocol. Radiology 1994; 192:367-371. 8. Dupuy DE, Costello P, Ecker CP. Spiral CT of the pancreas. Radiology 1992; 183:815-818. 9. Costerlo P, Dupuy DE, Ecker CP, Tello R. Spiral CT of the thorax with reduced volume of contrast material: a comparative study. Radiology 1992;183:663-666. 10. Foley WD, Berland LL, Lawson TL, Smith DF, Thorsen MK. Contrast enhancement technique for dynamic hepatic computed tomographic scanning. Radiology1983;147:797-803. 11. Shaffer K, Pugatch RD. Small pulmonary nodules: dynamic CT with a single-breath technique. Radiology 1989;173:587-568. 12. Dean PB, Violante MR, Mahoney JA. Hepatic CT contrast enhancement: effect of dose, duration of infusion, and time elapsed following infusion. Invest Radiol 1980; 15:158-161. 13. Kormano M, Partanen K, Soimakalrio S, Kivimaki T. Dynamic contrast enhancement of the upper abdomen; effect of contrast medium and body weight. Invest Radio11983; 18:364-367. 14. Claussen CD, Banzer D, Pkretzschner C, Kalender WA, Schorner W. Bolus geometry and dynamics after intravenous contrast medium injection. Radiology 1984; 153:365-368. 15. Zeman RK, Clements LA, Silverman PM, et al. CT of the liver: a survey of prevailing methods for administration of contrast material. AJR 1988; 150:107-109. 16. Heiken JP, Brink JA, McCrennan BL, Sagel SS, Forman HP, DiCroce J. Dynamic contrast-enhanced CT of the liver: comparison of contrast medium injection rates and uniphasic and biphasic injection protocols. Radiology 1993; 187:327-331. 17. Foley WD. Dynamic hepatic CT. Radiology 1989;170:617-622. 18. Kormano M, Dean PB. Extravascular contrast material: the major component of contrast enhancement. Radiology 1976;121:379-382. 19. Burgener FA, Hamlin DJ. Contrast enhancement in abdominal CT: bolus vs infusion. AJR 1981; 137:351-358.
25