| OCR Text |
Show ORIGINAL ARTICLE Utility of Source Images of Three- Dimensional Time- of- Flight Magnetic Resonance Angiography in the Diagnosis of Indirect Carotid- Cavernous Sinus Fistulas Yuh- Feng Tsai, MD, Liang- Kong Chen, MD, Cheng- Tau Su, MD, Ta- Nien Lu, MD, Chin- Chu Wu, MD, and Chu- Jen Kuo, MD Background: We sought to assess the relative contribution of magnetic resonance imaging ( MRI), maximum intensity projection ( MIP), and source images of three-dimensional ( 3D) time- of- flight ( TOF) magnetic resonance angiography ( MRA) to the diagnosis of indirect ( dural) carotid- cavernous sinus fistulas ( CCFs). Methods: MRI and 3D TOF MRA were obtained in eight consecutive patients with indirect CCFs confirmed by conventional catheter angiography. Two radiologists masked to the angiographic results reviewed images retrospectively to evaluate the efficacy of MRI and 3D TOF MRA source and MIP images in the diagnosis of CCF. Results: MRI disclosed CCF in five of eight cases; MIP images of TOF MRA disclosed CCF in four cases; source images of TOF MRA disclosed all eight CCF cases. Conclusions: The MRA source images are indispensable for a confirmatory diagnosis of indirect ( dural) CCF. Un-derdiagnosis may occur by relying on MRI or 3D TOF MIP images alone. ( JNeuvo- Ophthalmol 2004; 24: 285- 289) The gold standard in diagnosis of carotid- cavernous sinus fistula ( CCF) is conventional catheter angiography. However, it is an invasive procedure that subjects the patient to the potential risks of procedure- related vascular and radiation injuries, as well as contrast- induced anaphylaxis and renal damage. Moreover, it carries a risk of thromboembolic events as high as 3% in a large series ( 1). The advent of magnetic resonance angiography ( MRA), using Department of Diagnostic Radiology, Shin Kong Wu Ho- Su Memorial Hospital, Shih Lin, Taipei, Taiwan, Republic of China, and College of Medicine, Fu Jen Catholic University, Taipei, Taiwan, Republic of China. Address correspondence to Cheng- Tau Su, MD, Department of Diagnostic Radiology, Shin Kong Wu Ho- Su Memorial Hospital, 95, Wen- Chang Road, Shih Lin, Taipei, Taiwan, Republic of China; E- mail: yuhfeng. tsai@ msa. hinet. net blood flow as the physical basis for producing contrast between moving spins and stationary tissue ( 2), has been applied in the diagnosis of CCF ( 3,4), but the reported literature is limited. We therefore sought to evaluate the efficacy of MRI, maximal intensity projection ( MIP) and source images of three- dimensional ( 3D) time- of- flight ( TOF) MRA in the diagnosis of indirect ( dural) CCF. METHODS From March 2001 to May 2003, eight consecutive patients ( four women and four men; mean age, 64.3 years; range, 43- 80 years) with indirect ( dural) CCFs were examined with MRI and MRA. The diagnosis was then confirmed by digital subtraction conventional catheter angiography. MRI and MRA were performed with a 1.5- T unit ( Magnetom; Siemens, Symphony, Germany). A standard MRI protocol was performed in all patients and consisted of axial/ coronal Tl- weighted spin- echo ( 525/ 15/ 90), axial, and/ or coronal T2- weighted fast spin- echo ( 3,500/ 96 ms), and sagittal fluid attenuated inversion recovery sequences. An intravenous contrast injection was not routinely performed. If so, the protocol consisted of coronal and axial Tl- weighted images. Three- dimensional TOF MRA was performed with tilted, optimized, nonsaturating excitation and magnetization transfer techniques. Acquisition parameters of this sequence were fast imaging with steady- state precision 3D; time to recovery, 36 ms; time to echo, 7 ms; flip angle, 25°; matrix, 192 x 512; field of view, 200 mm; three slabs with 32- mm thickness for each slab, 32 partitions of 1- mm thickness. The total acquisition time was 5 minutes 33 seconds. The MRA slabs were axially placed on the skull base by using the sagittal image as reference. All acquisitions included presaturation of venous blood flow. The axial ( source) images thus acquired were used for the construction of projection images with a MIP algorithm. For all patients, 12 projections were obtained vertically and horizontally at 15° increments over a 180° range. The MRI, MIP, and source images were filmed for review. J Neuro- Ophthalmol, Vol. 24, No. 4, 2004 285 JNeuro- Ophthalmol, Vol. 24, No. 4, 2004 Tsai et al Digital subtraction angiography ( DSA) was performed with a 512 x 512 matrix angiographic unit ( An-giostar; Simens). All patients underwent selective catheterization of vertebral, external carotid, and internal carotid arteries with a 4- or 5- French catheter via a femoral artery approach. The DSA images were also filmed for review. The angiographic results were grouped according to the Barrow classification of CCFs ( 5) as follows: type A, direct high- flow shunt between the internal carotid artery ( ICA) and the cavernous sinus ( CS); type B, dural shunt between the meningeal branches of ICA and the CS; type C, dural shunt between the meningeal branches of the external carotid artery and the CS; and type D, dural shunt between the CS and the meningeal branches of both the ICA and external carotid artery. MRI and MRA MIP and source images were reviewed independently by two radiologists who were aware of the clinical history but masked to the catheter angiographic results. MRI, MRA MIP, and MRA source images from the same patient were evaluated on separate dates. The final decision was reached by consensus in cases of discrepancy between the two reviewers. RESULTS The efficacy of MRI and MIP and source MRA images in depicting CCF using DSA as the gold standard is illustrated in Table 1. There were seven Barrow type D and one Barrow type B CCFs. MRI alone made the correct diagnosis in five cases; MRA MIP alone made the diagnosis in four cases; and MRA source images alone made the diagnosis in all eight cases by demonstrating flow- related enhancement in the cavernous sinus. In four cases ( patients 1, 2,3, and 8), MRI, MRA MIP, and MRA source images were concordant in making the diagnosis ( Figure 1, Patient 8). In five cases ( Patients 1,2, 3, 7, and 8), CCFs were diagnosed by a combination of MRI and MRA MIP, without reviewing the MRA source images. In one case ( Patient 1), a type A fistula caused by rupture of an intracavernous aneurysm was diagnosed by MRI and both the MRA MIP and source images, whereas a type D fistula without an associated aneurysm was diagnosed only after DSA ( Fig. 2). DISCUSSION In this article, we have reported the clinical applications of MRI and MRA to the diagnosis of indirect CCF. In terms of diagnosing CCF, MRI relies on demonstration of abnormal flow voids in the CS and demonstration of complications related to venous hypertension, such as an enlarged CS, engorged draining veins, and swollen extraocular muscles ( 6- 8). Uchino et al ( 6) reported that they were able to detect MRI flow voids in the CS in 11 of 12 CCFs and a dilated superior ophthalmic vein in 9 of 12 CCFs, and therefore drew the conclusion that MRI would be relatively useful in the diagnosis of indirect CCF. However, we detected MRI flow voids in the CS in only five of eight cases, among which an engorged CS was noted in only four cases, an enlarged cortical draining vein in one case, and swollen extraocular muscles in no cases. Besides, we encountered dilemmas in reviewing MRI flow voids in the CS. First, flow artifacts resulting from pulsation of the cavernous ICA led to signal " smearing" along the phase- encoded direction and corrupted the details of the CS ( Fig. 1 A). This problem could be eliminated by changing phase- encoded direction, applying a flow compensation technique, or referring to TABLE 1. Comparison of the diagnoses made by MRI, MRA maximal intensity projection ( MIP) and MRA source images in eight indirect ( dural) carotid- cavernous sinus fistulas using digital subtraction conventional ( catheter) cerebral angiography as the gold standard Patient No. 1 2 3 4 5 6 7 8 Magnetic Resonance Imaging ( findings that led to diagnosis) CCF** ( FVCS, SOVE, CVE, CSE) CCF ( FVCS, CSE) CCF ( FVCS) Normal Normal Normal CCF ( FVCS, SOVE, CSE) CCF ( FVCS, CSE) Magnetic Resonance Ang Maximum Intensity Projection Images CCF** CCF CCF Normal Normal Normal Normal CCF jography Source Images CCF** CCF CCF CCF CCF CCF CCF CCF Barrow CCF type* D D B D D D D D * Based on digital subtraction conventional catheter angiography. ** In this case, all MRI and MRA images had wrongly suggested a Type A CCF. FVCS, flow void in cavernous sinus; SOVE, superior ophthalmic vein engorgement; CSE, cavernous sinus enlargement; CVE, cortical vein engorgement. 286 © 2004 Lippincott Williams & Wilkins Source Images ofMRA in Depicting Indirect CCF JNeuro- Ophthalmol, Vol. 24, No. 4, 2004 FIG. 1. Patient 8. Concordance of findings between fast spin- echo magnetic resonance ( MRI), three-dimensional time- of- flight magnetic resonance angiography ( 3D TOF MRA) maximal intensity projection ( MIP) and source images in the diagnosis of indirect carotid- cavernous fistula ( CCF). Axial T1 ( A) and axial T2 ( B) MRI at the level of pituitary gland demonstrate flow voids at the medial and posterior aspects of the right cavernous sinus ( large arrow). Engorgement of bilateral inferior petrosal sinuses is evident on T1 ( small arrows). Flow artifacts resulting from the internal carotid artery are shown on T1 ( arrowheads). C. MIP image discloses flow- related enhancement in both cavernous sinuses ( arrows). D. The details of abnormal drainage vessels are demonstrated more clearly on MRA source images. In concert with engorgement of bilateral cavernous and inferior petrosal sinuses ( large arrows), the imaging of an intercavernous sinus ( small arrows), which is slightly corrupted by flow artifacts, is suggestive of a CCF. FIG. 2. Patient 1. Type D CCF misdiagnosed by MRI and MRA as type A CCF. A. Coronal T1 MRI shows a round, well- defined flow void in the posterior aspect of the right cavernous sinus ( arrowheads), which could represent an intracavernous aneurysm. B. MRA source image shows that the signal intensity of the aneu-rysm- like lesion is inhomogeneous ( large arrow), presumably caused by turbulent flow artifact ( arrow). There is engorgement of the right inferior petrosal sinus ( arrowhead), making the diagnosis of CCF complicating an aneurysmal rupture likely. C. MRA source image at the level of the midbrain discloses engorged cortical draining veins ( large and small arrows), further suggesting the diagnosis of CCF. D. Conventional angiography, right ICA injection, shows the cavernous stain of a CCF. There is no evidence of associated aneurysm. E. Conventional angiography, right external carotid artery injection, shows that the fistula also receives blood from the meningeal branches of the internal maxillary artery. The main drainage vessel is a cortical vein ( arrowheads). This is therefore a type D CCF. 287 JNeuro- Ophthalmol, Vol. 24, No. 4, 2004 Tsai et al FIG. 3. Patient 4. CCF missed by MRI and MRA MIP but diagnosed by MRA source images. A. Axial T2 image shows crescentic flow- void at the medial aspect of the left cavernous sinus { arrow). The lesion was initially overlooked. B. The MIP image is unable to depict the lesion. C. The MRA source image discloses flow-related enhancement in the medial and posterior aspects of the left cavernous sinus. D, E. Conventional angiography ( D, left internal carotid; E, left external carotid artery injection) shows that the CCF not only receives blood from the meningo-hypophyseal trunk, but from the meningeal branches of the internal maxillary artery ( E, arrow). other images in different planes. On a few occasions, the pulsation of the cerebrospinal fluid resulted in flow voids in the prepontine cistern mimicking enlarged abnormal vessels ( Fig. 3). Air in the sphenoid sinus, especially when prominent, led to a diagnostic pitfall because of susceptibility artifacts or partial volume effects at air/ CS boundaries ( Fig. 3). The cortex of the petrous tip, which lies close to the posterior aspect of CS, also appears as low signal intensity on MRI and was also an obstacle to accurate diagnosis of CCF ( Fig. 4). In our experience, reviewing the MRA source images eliminated most of these diagnostic pitfalls. The most popular and frequently used MRA technique is TOF, which is based on suppressing the signals from background static tissues and selectively imaging the FIG. 4. Patient 5. CCF missed by MRI and MRA MIP but diagnosed by MRA source images. A. T2 MRI shows a flow void at the posterior aspect of the left cavernous sinus { arrow). This finding was misinterpreted as partial volume averaging of the petrous tip. B. The MRA MIP image does not show a clear abnormality. C. The MRA source image clearly depicts flow-related enhancement in the left cavernous sinus { arrow). D. The source image at another level shows flow- related enhancement of the cerebrospinal fluid { small arrows). Careful inspection discloses that this is separate from flow- related enhancement of an enlarged inferior petrosal sinus { large arrow), a manifestation of a CCF. 288 © 2004 Lippincott Williams & Wilkins Source Images ofMRA in Depicting Indirect CCF JNeuro- Ophthalmol, Vol. 24, No. 4, 2004 inflowing spins ( blood), a phenomenon called " flow-related enhancement" ( 2). The original data ( the source images), on which blood appears as high signal, can be acquired as a series of overlapping thin sections and reconstructed in a familiar 3D " angiographic" format. Among the techniques used for image reconstruction, the MIP method is the one most frequently used because its implementation is relatively simple and it conveys the den-sitometric information of the source images without needing to change any parameters ( 9). In addition, MIP images are easier to interpret than are the source images because the viewer can inspect the angiography- like images in a 3D fashion from several different angles of projection to avoid vessel overlap. Most clinicians think that MIP images are representative of MRA. This is not true. Information seen on the source images may be lost in the reconstructed MIP images. Thus, reliance on these images alone may lead to misdiagnosis, as occurred in 50% of our cases. Vascular distortion resulting from MIP reconstruction has been well described by Anderson et al ( 10). For MIP to have a high probability of success, the intensity of the vessel should be on average at least two standard deviations above background intensity on the source images ( 10,11). Therefore, small or slow-flowing vessels may be poorly visible or even missed on MIP images ( 9- 11). By contrast, MRA source images were able to depict all eight indirect CCFs in our series, although it is well known that blood signal may be lost even on the source images in regions of slow, turbulent, pulsatile, oblique, or in- plane flow ( 12). MRA is a collection of all related methods for obtaining angiographic data, not merely the reconstructed 3D angiographic display, but also the original source images. To minimize the diagnostic pitfalls and to achieve the most appropriate diagnosis in CCF, one must have a thorough understanding of the technical limitations of MRA and a careful reviewing of all the images, including MIP and source images. Despite its ability to diagnose CCF, MRA still cannot compete with DSA in demonstrating the details of feeding and draining vessels. In our series, a case of type D fistula was misinterpreted by MRI and MRA as a type A fistula associated with rupture of an intracavernous aneurysm ( Patient 1). An enlarged CS with turbulent flow might explain this shortcoming of MRA ( Fig. 2). REFERENCES 1. Dion JE, Gates PC, Fox AJ, et al. Clinical events following neu-roangiography: a prospective study. Stroke 1987; 18: 997- 1004. 2. Gullberg GT, Wehrli FW, Shimakawa A. MR vascular imaging with a fast gradient refocusing pulse sequence and reformatted images from trans- axial sections. Radiology 1987; 165: 241- 6. 3. Ouanounou S, Tomsick TA, Heitsman C, et al. Cavernous sinus and inferior petrosal sinus flow signal on three- dimensional time- of-flight MR angiography. Am JNeuro Radiol 1999; 20: 1476- 81. 4. Ikawa F, Uozumi T, Kiya K, et al. Diagnosis of carotid- cavernous fistulas with magnetic resonance angiography- demonstrating the draining veins utilizing 3- D time- of- flight and 3- D phase- contrast techniques. NeurosurgRev 1996; 19: 7- 12. 5. Barrow DL, Spector RH, Braun IF, et al. Classification and treatment of spontaneous carotid- cavernous sinus fistulas. J Neurosurg 1985; 62: 248- 56. 6. Uchino A, Hasuo K, Matsumoto S, et al. MRI of dural carotid-cavernous fistulas. Comparisons with postcontrast CT. Clin Imaging 1992; 16: 263- 8. 7. Hirabuki N, Miura T, Mitomo M, et al. MR imaging of dural arteriovenous malformations with ocular signs. Neuroradiology 1988; 30: 390- 4. 8. Chen JC, Tsuruda JS, Halbach VV. Suspected dural arteriovenous fistula: results with screening MR angiography in seven patients. Radiology 1992; 183: 265- 71. 9. Sato Y, Shiraga N, Nakajima S, et al. Local maximum intensity projection ( LMIP): a new rendering method for vascular visualization. JComput Assist Tomogr 1998; 22: 912- 7. 10. Anderson CM, Saloner D, Tsuruda JS, et al. Artifacts in maximum-intensity- projection display of MR angiograms. AJR Am J Roentgenol 1990; 154: 623- 9. 11. Tsuruda J, Saloner D, Norman D. Artifacts associated with MR neu-roangiography. Am J Neuro Radiol 1992; 13: 1411- 22. 12. Pipe JG. Limits of time- of- flight magnetic resonance angiography. Top Magn Reson Imaging 2001; 12: 163- 74. 289 |
