| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology and Nuclear Medicine, Charité-University
Hospital, Campus Benjamin Franklin, Hindenburgdamm 30, 12200 Berlin,
Germany.
2 Department of Biometry and Clinical Epidemiology, Charité-University
Hospital, Berlin, Germany.
3 Siemens Medical Solutions AG, Forchheim, Germany.
4 Present address: Department of Radiology and Radiological Science, The Johns
Hopkins Hospital, Baltimore, MD.
Received June 7, 2007;
accepted after revision October 7, 2007.
Supported by a research grant from Siemens Medical Solutions.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Fifteen patients with either hepatocellular carcinoma (HCC, n = 8) or hypervascular liver metastases from uveal melanoma (n = 7) underwent arterial and portal venous C-arm CT of the liver using intraarterial contrast media administration directly before TACE. The number and location of their hepatic malignancies were compared with those on MDCT. The scan coverage was documented and the liver diameter measured on MDCT.
RESULTS. Compared with MDCT, the sensitivity and specificity for segmental tumor involvement were 97% (76/78) and 85% (28/33), respectively, for reader 1, and 99% (77/78) and 79% (24/29), respectively, for reader 2. Complete scan coverage of the liver was obtained in five of the 15 patients with C-arm CT. In patients with incomplete scan coverage on C-arm CT, the craniocaudal liver diameter was significantly larger than in those patients with complete scan coverage (mean [95% CI], 22.7 [19.5–25.9] cm vs 20.2 [15.4–25.0] cm, p = 0.0193).
CONCLUSION. Biphasic arterial and portal venous C-arm CT showed a high sensitivity for the detection of malignant liver lesions. However, the liver could not be visualized completely in two thirds of the patients. Therefore, the current scan range limitations need to be overcome to make C-arm CT a valuable adjunct to MDCT for preprocedure evaluation and postprocedure follow-up imaging.
Keywords: C-arm CT • hepatocellular carcinoma • transarterial chemoembolization
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Before TACE, cross-sectional imaging tech niques such as MDCT served as the mandatory baseline study to assess tumor extent and to exclude contraindications such as thrombosis of the portal vein. During the TACE procedure, digital subtraction angiography of the celiac trunk is performed before administration of the chemotherapeutic and embolic agents to determine the origin and course of tumor-feeding vessels. This is important because a high variability of arterial supply to hypervascular liver lesions has been shown [3]. The angiographic images provide the baseline information required for selective positioning of the catheter in tumor-feeding vessels. In this way, collateral damage can be reduced to a minimum, which is optimal because of the palliative nature of the therapy.
The development of digital flat-detector angiographic systems with high frame rates facilitates 3D tomographic reconstructions, there by resulting in a new class of hybrid C-arm systems capable of producing conventional projectional angiographic images and CT-like images (i.e., C-arm CT images). For example, in patients undergoing TACE of the liver, the same equipment can be used to perform the procedure under fluoroscopic control and to acquire C-arm CT images during contrast injection into the superior mesenteric artery (SMA), known as CT during arterial portography (CTAP). Therefore, soft-tissue information is obtained during the interventional procedure.
The purpose of this study was to test the hypothesis that images acquired on biphasic C-arm CT during TACE of the liver provide the same information regarding hepatic tumor involvement as preinterventional, contrast-enhanced MDCT images. The secondary variables, the image quality and scan coverage of C-arm CT, were compared with those of MDCT.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Imaging Techniques
Biphasic CT of the liver was performed on a 16-MDCT scanner (Somatom
Sensation 16, Siemens Medical Solutions) using the following parameters: 120
kVp; 160 mAs; collimation, 16 x 1.5 mm; slice thickness, 2 and 5 mm;
table feed, 48 mm per rotation; and rotation time, 0.5 second. Via an
antecubital vein, 150 mL of IV contrast agent (iopromide [Ultravist, Bayer
HealthCare, formerly Schering], 300 mg I/mL; flow rate, 3.5 mL/s) followed by
50 mL of normal saline, was administered. Scanning of the liver in the
arterial phase was begun 4 seconds after a threshold attenuation of 
150
H was reached in the suprarenal aorta during a single breath-hold. After an
additional delay of 40 seconds, the liver was scanned in the portal venous
phase after the arterial scan, also during a single breath-hold.
TACE of the liver was performed on an angiography system (AXIOM Artis dBA, Siemens Medical Solutions) equipped with a 30 x 40 cm flat-panel detector.
Before TACE, selective digital subtraction angiography of the celiac trunk (5F Cobra C1 catheter [Cook], 20 mL of iomeprol [Iomeron, Bracco]; 300 mg I/mL; flow rate, 2–4 mL/s) and the superior mesenteric artery (5F Cobra C1 catheter; 20 mL of iomeprol, 300 mg I/mL; flow rate, 2–4 mL/s) was performed. The time between starting the contrast injection and the beginning of portal vein enhancement was measured to determine the scanning delay of the portal venous C-arm CT acquisition. Subsequently, portal venous C-arm CT of the upper abdomen was performed in the portal venous phase (C1 catheter in the SMA, 20 mL of iomeprol diluted with 20 mL of saline for an iodine concentration of 150 mg I/mL; flow rate, 3 mL/s). After repositioning the C1 catheter in the celiac trunk, arterial C-arm CT of the liver was acquired (11 mL of iomeprol diluted with 22 mL of saline for a iodine concentration of 100 mg I/mL; flow rate, 3 mL/s; and scanning delay, 2 seconds). In none of the patients was a variant of the arterial supply of the liver (e.g., replaced right hepatic artery, common gastrohepatic trunk) observed that necessitated additional C-arm CT for full liver coverage. The chronologic order of the procedure was chosen to allow in creasing dilution of the contrast material with C-arm CT. Based on results from phantom studies, an iodine concentration of 150 mg I/mL was used for the portal venous phase and a concentration of 100 mL I/mg was used for the celiac trunk injection.
Both C-arm CT sequences were acquired using the 10s1k preset (DynaCT, Siemens Medical Solutions) with an acquisition time of 10 seconds; a detector size of 30 x 40 cm; a fixed tube–detector distance of 0.9 m; a total scanning angle of 222°; projection increment, 0.8°; 1k-matrix; zoom factor, 0; field of view, 48 cm; and a system dose per pulse of 0.36 µGy during a singe breath-hold. With the detector we used, the scanning range had a cylindric shape with a height of 185 mm (craniocaudal coverage) and a diameter of 225 mm (transverse and sagittal scanning range). To maximize coverage of the liver, the center of the scanning field was on the upper right abdomen under fluoroscopic control. The raw data sets from the angiographic C-arm system were sent to a dedicated external workstation (X-Leonardo, Siemens Medical Solutions) and reconstructions were performed to generate the C-arm CT images. Depending on the local network performance, the time from the end of scanning until the availability of cross-sectional images on the workstation ranged from 4.2 to 6.3 minutes. The volume data set produced by the workstation had an isotropic voxel size of 0.4 mm. For all reconstructions of the abdominal C-arm CT images, the bone smooth reconstruction kernel with a large field of view and artifact reduction was used, as recommended by the manufacturer. For direct comparison of the C-arm CT images with the MDCT images, transversal multiplanar reconstructions (MPRs) with slice thicknesses of 2 and 5 mm were reconstructed.
Image Analysis
Image analysis was performed after complete data acquisition for the entire
patient population. The images of both imaging techniques were interpreted
independently in random order in two separate sessions. In the first session,
the C-arm CT images were analyzed, and MDCT images were reviewed 4 weeks
later.
Two radiologists, each with 7 years of experience in CT image evaluation, independently reviewed all 15 MDCT and C-arm CT studies separately and in random order on a commercial workstation (Leonardo VB30, Siemens Medical Solutions). Both observers were aware that patients were scheduled for TACE; however, both were blinded to the type of primary malignancy and to the extent of disease.
Both readers were asked to note whether the liver was completely covered by the scan on the basis of the liver segment classification of Couinaud [4]. Truncation segments were classified on a 3-point scale (T1, no or minor truncation of the liver segment border, not compromising image interpretation for segment involvement; T2, major truncation, substantially truncated segment, segment noninterpretable for segment involvement; and T3, all segments absent after surgical resection). In addition, the readers were asked to note and to classify potential artifacts in the liver segments with scores of A1, no artifacts; A2, minor artifacts, not compromising interpretation of the liver segment; A3, moderate artifacts, compromising interpretation of the liver segment but still diagnostic; and A4, massive, substantial artifacts, segment not diagnostic.
The maximum craniocaudal and maximum transverse liver diameters and the maximum sagittal and transverse abdominal diameters were measured in each patient and were correlated with the absence or presence of artifacts or segmental truncation.
Both readers were then asked to document location and maximum diameter of all hepatic lesions on schematic liver charts and to classify each lesion according to its enhancement pattern. Following established criteria [5], a lesion was classified as malignant if washout was observed on portal venous phase images after arterial contrast enhancement. A confidence score was assigned to each diagnosis on the basis of whether the lesion was hepatocellular carcinoma or a hepatic meta stasis of malignant ocular melanoma, sum marized as hepatic malignancy. Confidence scores ranged from 1 to 5 (1, not hepatic malig nancy; 2, probably not hepatic malignancy; 3, indeterminate; 4, pro bably hepatic malignancy; and 5, definitely hepatic malignancy). For all statistical analysis, only those lesions regarded as probably or definitely hepatic malignancy (confidence score of 4 or 5) were determined to be positive findings. All lesions with confidence scores of 1, 2, or 3 were considered neg ative for hepatic malignancy. If more than three positive findings per liver segment were recorded or the liver segment was completely interspersed with tumor, segmental involvement was classified as diffuse.
To define the standard of reference, both readers matched their positive findings on MDCT. In patients with discordant findings, the images were jointly reviewed and a consensus was defined. On the basis of this final consensus reading, a standard of reference liver chart was created, and a one-to-one correlation of the positive findings in C-arm CT and in MDCT as the standard of reference was performed. The original MDCT and C-arm CT data sets were available for review and could be used at the discretion of the readers.
Statistical Analysis
For statistical analysis, the MDCT reference liver chart was used as the
standard of reference. The sensitivity of C-arm CT was defined as the number
of hepatic malignancies correctly detected by each reader divided by the
number of hepatic malignancies in the MDCT standard of reference liver chart.
Only those segments with complete coverage on MDCT in both phases were
included. The false-positive rate was defined as the number of falsely
detected hepatic lesions divided by the total number of lesions (true-positive
and false-positive) identified as hepatic malignancy on C-arm CT. The positive
predictive value was defined as the number of correctly detected hepatic
malignancies on C-arm CT divided by the total number of lesions considered to
be hepatic malignancies on C-arm CT. The maximum diameters of true-positive
and false-positive lesions in C-arm CT were compared using Mann-Whitney's rank
sum test.
Descriptive statistical analysis—that is, sensitivity, specificity, and predictive values—was performed in relation to liver segments or lesions. Because of the statistical dependencies of measurements in segments of the same patients, confirmatory statistical analysis was done with relation to patients. Therefore, scores for the presence of artifacts and the scanning coverage of the liver were computed per patient. These scores were compared using Wilcoxon's signed rank test. The level of significance was 0.05 (two-sided). Liver diameters between patients were compared using the Mann-Whitney rank sum test. No adjustment for multiple testing was applied because of the limited number of patients. Therefore, significances are not strictly confirmatory. Cohen's kappa statistic was used to assess the interobserver variability regarding the segmental hepatic tumor involvement for MDCT and for C-arm CT in readers 1 and 2 and the intertechnique variability of MDCT and C-arm CT in readers 1 and 2.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Major artifacts (A4) causing nondiagnostic image quality were not observed on either technique. However, both readers rated significantly more segments as containing artifacts compromising image interpretation (A3) on C-arm CT than on MDCT (Table 2). Although on MDCT artifacts were observed in two patients in the portal venous phase (most likely because of their breathing during the examination), on C-arm CT, moderate artifacts occurred in eight (reader 1) and nine patients (reader 2), respectively. The distribution of segmental artifacts on MDCT and C-arm CT are shown in Table 2.
|
In patients with incomplete scan coverage on C-arm CT (n = 10), the craniocaudal liver diameter was significantly larger than in patients without incomplete scan coverage (n = 5) (mean [95% CI], 22.7 [19.5–25.9] cm vs 20.2 [15.4–25.0] cm, p = 0.0193). No significant difference was seen in the maximal transverse liver diameter or in the patient's maximum sagittal and transverse diameters in either group.
In patients with compromising artifacts on C-arm CT, the maximum transverse liver diameter was significantly larger than in patients without artifacts (mean [95% CI], 24.5 [21.3–27.7] cm, n = 8 [reader 1], and 23.7 [20.9–26.6] cm, n = 9 [reader 2]; without artifacts, 18.8 [16.5–21.2] cm, n = 7 [reader 1], and 19.1 [15.1–23.0] cm, n = 6 [reader 2]; p = 0.0059 [reader 1] and p = 0.0360 [reader 2]). No correlation was seen between the maximum craniocaudal liver diameter or the patient's sagittal and transverse abdominal diameters and the presence of artifacts on C-arm CT. We did not observe a correlation of artifacts or segmental coverage of MDCT with the maximum sagittal and transverse abdominal or liver diameters.
|
|
|
|
=
0.86) and C-arm CT ( = 0.85). The C-arm CT–MDCT intertechnique
agreement in terms of defining segmental tumor involvement was also very good
for both readers (reader 1, = 0.85; reader 2, = 0.82).
Including all segments completely covered by MDCT, the sensi tiv ity and
specificity for segmental tumor involvement were 97% and 85% for reader 1 and
99% and 79% for reader 2, respectively.
Table 3 summarizes the
intertechnique a gree ment data.
|
When excluding all segments with diagnostically relevant artifacts or truncation on C-arm CT, no relevant changes in intertechnique agreement, sensitivity, or specificity were observed.
Lesion-to-Lesion Correlation
Direct lesion-to-lesion agreement was determined for all lesions with a
confidence score of 4 or 5 for lesions located in segments without substantial
truncation on both phases of MDCT and C-arm CT (reader 1, 97/117 segments;
reader 2, 96/117 segments). Lesions with a confidence score less than 4 were
excluded (on MDCT, five lesions, three with a confidence score of 3 and two
with a confidence score of 2; on C-arm CT, reader 1 gave two lesions a
confidence score of 3, and reader 2 gave seven lesions a confidence score of
3). MDCT showed diffuse tumor spread in 41 segments; this was also seen on
C-arm CT by both readers.
In addition, C-arm CT showed diffuse tumor manifestation in seven and six segments for readers 1 and 2, respectively, whereas in one of those segments, no lesions were detected on MDCT. The remaining six and five segments with diffuse tumor manifestation on C-arm CT showed 10 and eight lesions on MDCT for readers 1 and 2. In the remaining 49 segments without diffuse manifestation on either technique, only two and one lesions were seen on MDCT by readers 1 and 2, respectively; 14 and 18 lesions were seen on C-arm CT (Figs. 1A, 1B, 1C, 1D, 2A, 2B, 2C, and 2D); and 24 lesions were seen on both techniques by both readers. Sensitivities for the detection of single lesions were 92% and 96% for readers 1 and 2. The maximum diameter of false-positive lesions was significantly smaller than the maximum diameter of true-positive lesions for both readers (reader 1, p = 0.0064; reader 2, p = 0.0001). The results are shown in Table 4.
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although the combination of CT and digital subtraction angiography can in general be used efficiently for various indications in an interventional environment, two machines including the required equipment and located in one room is not an ideal situation for economic reasons. Therefore, the two techniques were installed in separate but neighboring rooms. This allows independent use, but, if needed, patient transfer between the two rooms is possible using a sliding table [10]. However, this setup has potential shortcomings: First, there is an increased risk of contamination of the digital subtraction angiography equipment, which requires extensive cleaning before any combined intervention. Second, the need for cross-sectional imaging might arise unexpectedly during a procedure when the CT scanner is already in use. Third, in cases such as anatomic variants of the arterial liver supply (e.g., replaced right hepatic artery), multiple patient transfers are necessary to image the whole liver in the arterial phase.
These limitations have been overcome by the introduction of a flat-detector angiography system with C-arm CT capabilities. Angiography and C-arm CT can be performed interchangeably using the same flat-detector C-arm and without patient transfer. Compared with the non–flat-detector cone-beam volume CT, C-arm CT images have improved contrast and spatial resolution because of the absence of image distortion, the higher detection dynamics, lower sensitivity to overexposure, and higher detector efficiency [12]. The reconstruction mechanism of flat-detector conebeam volume CT is analogous to that of MDCT in which the CT detector has 16–256 detector rows, each providing an axial slice from one orbit. Flat-detector cone-beam CT typically uses a large-area detector with 1,000 or more detector rows. Therefore, 1,000 axial slices—that is, a complete volume, with almost isotropic submillimeter spatial resolution—can be reconstructed on the basis of a single orbit [13].
The usefulness of this new, combined technique has so far been described in only a few case reports of complication management in challenging neurointerventional procedures [14], embolization of pancreatic and small bowel tumors [15] and TIPS placement [16]. However, several further theoretic benefits of C-arm CT have so far not been evaluated. For example, when using intraarterial contrast injection in the liver, as performed in this study, the principal setting of C-arm CT is comparable to those of CT hepatic angiography and CTAP. Both showed extremely high sensitivities for the detection of hypervascular lesions [3, 17] but have undergone a sustained decline in use during the past decade because of their cumbersome patient handling; susceptibility to perfusion-related artifacts, especially in patients with coexisting chronic liver disease; and further improvements in MDCT technology [18].
During MDCT, however, even with the most aggressive IV bolus injection, because the contrast medium is substantially diluted and scattered by the time it reaches the hepatic artery, isolating a true arterial phase and a true portal venous phase may be difficult or even impossible. Therefore, MDCT has known limitations for the detection of liver lesions, especially small hypervascular lesions. Valls et al. [19] evaluated the diagnostic accuracy of contrast–enhanced helical CT in 84 patients with liver cirrhosis before liver transplantation, using a contrast injection rate of 5 mL/s. Compared with the histopathologic results, the overall detection rate was 79%; however, the detection rate of lesions with a diameter less than 2 cm was 61% [19]. This may offer an explanation for the higher detection rate of hypervascular masses on digital subtraction angiography compared with that on contrast-enhanced CT [20].
Because of its local contrast material administration, C-arm CT during hepatic arteriography maintains the high lesion-to-background contrast of digital subtraction angiography and offers 3D spatial resolution with isotropic 0.4-mm voxels, which is superior to most MDCT scanners. C-arm CT should thus be able to provide delineation of tumor-feeding vessels and information about tumor location and extent, both of which could significantly influence the course of the intervention.
In this study, we evaluated whether C-arm CT provides the same information regarding hepatic tumor spread in patients with hypervascular tumors as does state-of-the-art MDCT. Both methods were used to assess the extent of hepatic tumor involvement before TACE. Interobserver agreement for segmental tumor involvement was very good for C-arm CT and compared favorably with that of MDCT. Regarding segmental tumor manifestation, a sensitivity ranging from 97% (76/78) to 100% (46/46) was obtained on C-arm CT; the specificity ranged from 79% (24/29) to 85% (28/33) and was slightly lower. In our TACE setting, however, the slightly lower specificity of C-arm CT may be negligible because some degree of overtreatment is more acceptable than undertreatment in which a tumor might be missed. Therefore, C-arm CT may even increase the oncologic safety of the procedure because some false-positive lesions seen on C-arm CT might have been missed on MDCT. Although the high false-positive rate might have been caused in part by a higher detection rate, as is seen on CT hepatic angiography and CTAP [3], C-arm CT is also, at least theoretically, prone to producing more false-positives because of flow-related effects.
One limitation of the currently available flat-detector angiography system used in this study is its limited field of view—a field of view of 225 x 225 x 185 mm. In two thirds of our cases, therefore, the liver was incompletely covered by C-arm CT. This might be acceptable for planning and for follow-up of targeted TACE in a predefined part of the liver, but it is problematic when visualization of the entire liver is required. Furthermore, artifacts in the cross-sectional images were more often observed in patients with the coronal liver diameter exceeding the possible maximum transverse scan diameter. Both limitations hampered the performance of C-arm CT in this study. However, these shortcomings can be overcome by increasing the scanning range and the size of the flat detector [12].
Further limitations of our study are the time-consuming processes of data transfer to the image reconstruction workstation and image reconstruction. However, continuous developments in computer technology will lead to improvements in the near future, thereby enabling successful real-time image reconstruction of C-arm CT.
Another limitation of this study is the lack of a true standard of reference for all lesions. Although pathologic correlation was obtained in each patient, it was not available for each lesion. MDCT as the common baseline examination for therapy planning of TACE was defined as the standard of reference. On the basis of the poor performance of CT for detecting small hypervascular lesions [19, 21], some of the false-positive findings on C-arm CT might actually have indicated true lesions that were not detected on MDCT. Similar issues have been raised in studies evaluating CT hepatic angiography and CTAP [3].
In conclusion, biphasic arterial and portal venous C-arm CT showed a high sensitivity for the detection of hypervascular liver lesions compared with that of state-of-the-art MDCT. To evaluate the effects of segmental tumor spread, C-arm CT showed a high agree ment with MDCT. Therefore, the hypothesis that C-arm CT provides the same information on hepatic tumor manifestation as MDCT can only be accepted with regard to the segments covered by C-arm CT. Currently, C-arm CT cannot replace MDCT for pretherapeutic workup.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
