Neuro
Neuroradiology (1992) 35:46-49
radiology 9 Springer-Verlag 1992
Magnetic resonance angiography compared to intra-arterial digital subtraction angiography in patients with subarachnoid haemorrhage A . G o u l i a m o s 1, E. Gotsis 2, L. V l a h o s 1, C. S a m a r a I, E. K a p s a l a k i 2, D . R o l o g i s 3, Z. Kapsalakis 2, and C. P a p a v a s i l i o u 1
Department of Radiology, University of Athens, 2 Diagnostic and Research Institute "Encephalos", 3D epartment of Neurosurgery, Athens General Hospital, Athens, Greece Received: 20 October 1991
Summary. In order to evaluate the sensitivity and specificity of magnetic resonance angiography ( M R A ) in spontaneous subarachnoid haemorrhage, 14 patients with recent h a e m o r r h a g e verified by CT or lumbar puncture were investigated with b o t h selective intra-arterial digital subtraction angiography ( I A - D S A ) and M R A by two independent teams, each having the same preangiographic information. The results were c o m p a r e d with each other and whenever possible (all positive cases except one) with those of surgical intervention. Seven patients were identified by M R A and I A - D S A as having a single aneurysm on the circle of Willis, 1 an aneurysm of the posterior inferior cerebellar artery 1 an aneurysm of the internal carotid artery (siphon) and 2 patients with two aneurysms on the circle of Willis. M R A and I A - D S A both failed to d e m o n strate aneurysms in 2 cases. T h r e e patients had negative results on b o t h methods and no surgical intervention was attempted. The aneurysms ranged f r o m 0.3 to 1.5 cm in size. In most cases there was agreement between M R A and D S A , leading us to believe that, if the p r o p e r protocols are followed, M R A is a powerful alternative to other established methods in the detection of intracranial aneurysms. A t this stage it will not replace I A - D S A prior to surgery, but the ability to obtain various projections using 3D M R A m a y improve surgical planning. K e y words: Aneurysm, cerebral - Cerebral angiography -
Magnetic resonance angiography
Magnetic resonance imaging (MRI) of subarachnoid h a e m o r r h a g e ( S A H ) has unique advantages: its sensitivity and specificity for the detection of non-acute h a e m o r rhage and blood b r e a k d o w n products and its potential for study of intracranial blood vessels for detection of aneurysms or arteriovenous malformations. With correct sequences at low field, M R I is m o r e sensitive than computed t o m o g r a p h y (CT) [1, 2]. While M R I is c o m m o n l y used for parenchymal changes routine application of flow effects for the study of the cerebral vasculature remains limited
[3-5].
The last few years have seen a rapid development of M R I techniques aiming at non-invasive imaging of intracranial and peripheral vasculature, usually based on 2D or 3D gradient echo pulse sequences with gradient m o m e n t refocussing ( G M R ) in the read-out and slice-select directions, or others including velocity encoding. These are referred to generically as magnetic resonance angiography (MRA). We have assessed the sensitivity and specificity of two techniques, a 3D time-of-flight (TOF) gradient echo sequence with first-order G M R and a 3D magnitude contrast pulse sequence (rephase-dephase), also with first-order G M R in the rephase data set. O u r results are therefore specific to these two methods. The aim of the present study was a comparison of M R A with intra-arterial digital subtraction angiography ( I A - D SA) and surgery in patients with spontaneous SAH.
Patients and m e t h o d s
Fourteen patients with spontaneous SAH were investigated with four vessels selective cerebral angiography on a digital subtraction unit, the results of which were not available to the MR operator or interpreter. There were 4 females and 10 males, 12~57 years old (mean 46). After routine MRI all patients underwent MRA. The two teams (DSA and MRA) had the same information prior to their respective examinations: the CT results and the history of the patient. Only when the results of both examinations were finalized were the findings compared and a surgical decision made. In most cases DSA preceded MRA by a few hours to i day, but in 5 cases MRA was carried out first, at the time of an MRI study. The MRA examinations were carried out on a 1.5 T superconducting MR imager, with very high homogeneity ( + 1.2 ppm in a 50 cm sphere). A circularly polarized head coil was used, and shimming was sometimes performed to correct inhomogeneities arising from paramagnetic fillings in teeth. The routine examinations included a series of axial Tl-weighted images [TR/TE 700/20 ms, 5 mm slice thickness, 256 x 256 matrix, 240 mm filed of view (FOV), and arterial presaturation], a series of 64 axial slices with a 3D rephase-dephase sequence, with partition thickness ranging from 1 to 1.5 mm, 200-230 mm FOV, 200 x 256 matrix and TRfrE 23/13 ms, 20 ~flip angle, and finally a series of 64 or 128 axial slices with a 3D TOF pulse sequence (FISP3D, TR/TE 25/7, 15 ~ flip angle, partition thickness 0.8-1.5 mm, 256x256
47
Fig. 1, a,b Digital subtraction angiography (DSA), left carotid arteriogram, anteroposterior and oblique projections, show normal vascular anatomy, without evidence of aneurysm, c 3D MR angiog-
raphy (MRA). Axial projection at a 25 ~rotation to the coronal plane also shows normal vascular anatomy
Fig.2. 3D MRA (case 1). Left anterior oblique projection with 30 ~ rotation allows precise localization of a 5 mm aneurysm on the anterior communicatingartery (arrow)
Fig.4. 3D MRA (case 7). Anteroposterior view obtained from a single 3D volume acquisition in the axial plane. The aneurysm is located at the trifurcation of the left middle cerebral artery (arrow)
Fig.3. 3D MRA (case 4). Left anterior oblique projection with 15 ~ rotation. The origin of the aneurysm from the left posterior communicatingartery (arrow) is well demonstrated
matrix). The MRA sequences included first-order GMR in the sliceselect and read-out directions. A series of 64 coronal 3D TOF slices was acquired in addition to or instead of the 3D TOF axial slices in some cases. In 2 patients examined early in the study, when magnitude contrast angiography was not available, only 3D TOF MR images were acquired with TR/TE 30/8 ms and other parameters as above. The 3D rephase-dephase sequence acquired two data sets (at the cost of doubling acquisition time), one with flow refocussing and one with flow defocussing. Substracting the second from the first resulted in a highly suppressed background, yieldingsuperb signal contrast of the vessels and discarding of high signal from static tissues simulating flow, such as subacute thrombus or haematoma. For the reconstruction of the angiograms a maximum intensity projection algorithm was used and run with an ultrafast image-processor (typical time for 12 projections with a 64-slice data set was 1 rain), allowing immediate visualization of the results of the first data set before planning the second or third. The speed of the image processor allowed the reconstruction of a large number of projections in a reasonable time so that a smooth hologram effect was simulated in cine mode. The cine mode was deemed valuable in resolving vessels which gave the impression of an aneurysm [6-8]. Correlation with IA-DSA performed within 1 week of the haemorrhage was available in all 14 patients. All studies included
anteroposterior, lateral and oblique projections, using a 0.3 mm focal spot and a caesium iodide image-intensifyingtube (Fig. 1 a, b). The MRA and spinecho studies were independentlyinterpreted by two neuroradiologists without knowing the history or the results of other imaging studies (Fig. 1 c). Aneurysm size was measured on IA-DSA and MRA studies in three dimensions (length, width, height). These three measurements were then averaged to give one number reflecting the anenrysm size. On IA-DSA allowance was made for magnification.
Results T h e results are s u m m a r i z e d in Table 1. T h e p r e s e n c e a n d site of the a n e u r y s m (or a n e u r y s m s in 2 cases) were correctly assessed b y M R A , in 12 of the 13 cases in which a n a n e u r y s m was c o n f i r m e d (92 %). T h e a n e u r y s m s were located as follows: 4 at the anterior c o m m u n i c a t i n g artery (Figs. 1, 5), 4 o n the m i d d l e c e r e b r a l artery (Fig.4), 3 at the p o s t e r i o r c o m m u n i c a t ing artery (Fig. 3), 1 at the origin of the p o s t e r i o r i n f e r i o r
48
Fig.5. a,b DSA (case 11). Right internal carotid artery, anteroposterior and lateral projections. There is suspicion of an aneurysm at the anterior communicating artery, e 3D MRA. Axial projection
allows precise identification of the 3 mm aneurysm at the origin of the anterior communicating artery (arrow)
cerebellar artery, and I on the internal carotid artery (siphon). The size of the aneurysm varied f r o m 0.3 to 1.5 cm and its neck (when present) was visualized by standard 3D T O F angiography. In case 9, one radiologist thought the D S A negative, but when the results of the M R A were revealed, all agreed that the D S A examination could be interpreted as positive. Because of the position of the aneurysm, and its pointing towards the sella turcica, it was not resolved well on DSA. Ten of the 11 DSA-positive patients were operated upon by two different neurosurgeons and the results were
confirmed surgically. The 11th has not yet been operated upon. Finally, 1 patient was operated upon on the results of M R A alone; an aneurysm on the middle cerebral artery was confirmed surgically (case 7). A n extensive haematom a was also present. Evidence of arterial spasm was seen in 4 cases and in 2 patients was identified by M R A . The aneurysms were located on the right in 9 cases and on the left in 4. In 2 cases m o r e than two aneurysms were diagnosed (cases 1 and 6). M R A did not show the second aneurysm in case 6. Other abnormalities seen included h a e m a t o m a s in 6 cases and infarcts in 2. The relative sizes of the aneurysms measured on M R A and I A - D S A are given in Table 1. The smaller aneurysms ( < 5 m m ) were accurately assessed using M R A . The relationship of the aneurysm to the parent vessel was easily visualized on M R A .
Table 1. Findings of intra-arterial digital subtraction angiography (IA-DSA) and magnetic resonance angiography (MRA)
Location of aneurysm Case 1 Case Case Case Case
2 3 4 5
Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Case 12 Case 13 Case 14
ACA (R)
~aCA (R) PCA (R) PCA (R) MCA (L) No aneurysm identified ACA (R) MCA (R) MCA (L) PICA (R) ICA (R) ACA (R) ACA (R) PCA (R) No aneurysm identified No aneurysm identified
Size (cm) MRA 0.5 0.3 0.3 0.5 0.5
0.5 Not visualized 0.5 1.0 1.5 0.5 0.3 0.5
IA DSA 0.8 0.5 0.5 0.5 0.7
0.3 0.2 1.0 1.5" 0.5 Not visualized 0.7
ACA, Anterior communicating artery; ICA, internal carotid artery; MCA, middle cerebral artery; PCA, posterior communicating artery; PICA, posterior inferior cerebellar artery "Initially one radiologist thought the DSA in this case negative
Discussion
All new methods must undergo scrutiny before they can be generally accepted in the medical establishment. In the case of M R A , there is a great variety of sequences to be tested and the final results can be site-specific as well, depending on magnet strength, eddy current compensation, homogeneity, etc., so that the pressure for scrutiny is even higher, if it is to be considered as an alternative to the wellestablished and tested D S A [9]. The highest resolution that can b e reached in 3D M R A varies, but it is not unreasonable to claim that less than i m m resolution in all three directions can be achieved with most 1.5 T imagers. O f course, such resolution does not m a t c h that of DSA, but the final question is: does one need higher resolution than that currently available? Obviously, most operable intracranial aneurysms are located on the circle of Willis, and are 2 m m in size or larger, within the detection limits of M R A , as this study and others have shown. Thus, one m a y be assured that aneurysms as small as 2 m m can be detected with 3 D T O F techniques, and if a h a e m a t o m a is present, making detection of the aneurysm difficult, magnitude contrast angiography or similar subtraction techniques can be used.
49 If time permits, one should use at least two different techniques for reassuring oneself that the findings are reproducible, as we have done in most cases. In the course of this study it became apparent that the highest possible resolution gave the most reliable results, so the last cases were examined under stricter conditions than the earlier ones. Our standard protocol now starts with a 3D T O F sequence (FISP 3D in the axial direction, T R / T E 21/7 ms, F O V 200 m m , 0.8 m m partition thickness, 256 x 256matrix). For comparison, a magnitude contrast sequence is used with a 1-1.25 m m partition thickness, 200 x 256 matrix and T R / T E 23/13 ms. Angiographic images are obtained in all possible directions so that all major vessels are well visualized. Display in cine m o d e is very helpful, and if doubt persists original slices are also examined. We found that if a lesion is not visible on the final angiogram no diagnosis can be m a d e from the original slices; thus, no effort has b e e n m a d e to rate detectability and specificity in terms of angiograms, angiograms + original, slices + cine mode, or any combination of the above with M R I , etc. We consider the reconstructed angiogram as the final conclusion, the findings of which can be confirmed by original slices, M R I etc., but not the reverse. The angiogram is of most use to the neurosurgeon. Conventional arteriography is considered the gold standard in the investigation of SAH. Additional studies which m a y help to identify asymptomatic aneurysms include CT, I V - D S A and M R A . CT and I V - D S A have limitations due to the spatial resolution and the need for contrast medium. M R A has potential value for visualization of intracranial vascular anatomy using techniques which allow increased contrast between vessels and surrounding tissues and appropriate manipulation of the images in the viewing console. The spatial resolution of 3D gradient-echo M R A is inferior to that of conventional angiography, but this is not a significant drawback, since aneurysms smaller than 3 m m in diameter rarely rupture [10, 11]. Limitations of M R A in the detection of intracranial aneurysms include: 1. Patient cooperation. 2. W h e n vasospasm is present, the 3D T O F technique tends to overestimate it. 3. Spatial resolution; it is expected that reliable ultra-high resolution (voxels with under 1 m m size) M R A of the circle of Willis willbe possible in less than 10 min.
The accuracy of M R A makes this m e t h o d ideal for screening patients presenting with headaches or other symptoms suggestive of possible intracranial aneurysm, before the development of SAH.
References 1. Jenkins A, Hadley D, Teasdale G, Condon B, Macpherson R Patterson J (1988) Magnetic resonance imaging of acute subarachnoid hemorrhage. J Neurosurg 68:431-736 2. Jenkins A, Paterson J, Hadley DM, et al. (1989) Use of in vitro magnetic resonance tissue studies to optimise pulse sequences in the imaging of intracranial hemorrhage. Magn Reson Imaging 7: 394-403 3. Ross JS, Masaryk T J, Modic MT, Ruggieri PM, Haacke EM, Selman WR (1990) Intracranial aneurysms: evaluation by MR angiography. AJNR 11:449-456 4. Masaryk TJ, Modic MT, Ross JS, Ruggieri PM, Laub GA, Lenz GW, Haacke EM, Selman WR, Wiznitzer M, Harik SI (1989) Intracranial circulation: preliminary clinical results with 3D (volume) MR angiography. Radiology 171:793-794 5. Gasparotti R, Orlandini A, Gualandi GF, Scipione V, Tansini A, Gnutti R Bonetti M, Lavezzi R Chiesa A, Galli G, Mearini M (1991) Uangiografia a risonanza magnetica nello studio del circolo cerebrale. I. Aneurismi cerebrali. Rev Neuroradiol 4: 153166 6. Spitzer CE, Blinder RA (1989) Vascular applications of MRI. Magn Reson Q 5:205-227 7. Weeden VJ, Rosen RR, Brady TJ (1987) Magnetic resonance angiography. Magnetic resonance annual. Raven Press, New York 8. Edelman RR, Mattle HR Atkinson D J, Hoogewould HM (1990) MR angiography. AJR 154:937-946 9. Pernicone JR, Siebert JE, Potchen E J, Pera A, Dumoulin CL, Souza SP (1990) Three-dimensional phase-contrast MR angiography in the head and neck: preliminary report. AJNR 11: 457466 10. Locksley HB (1969) Natural history of subarachnoid hemorrhage, intracranial aneurysms, and arteriovenous malformations. In: Sahs AL, Perret GE, Locksley HB, et al. (eds) Intracranial aneurysms and subarachnoid hemorrhage. A cooperative study. Lippincott, Philadelphia, pp 3%108 11. McCormick WF, Agosta-Rua GJ (1970) The size of intracranial saccular aneurysms. An autopsy study. J Neurosurg 33:422-427
Dr. A. Gouliamos Department of Radiology Areteion University Hospital 76, Vas. Sophias Ave. GR-115 28 Athens Greece
Neuroradiology (1992) 35:46-49
radiology 9 Springer-Verlag 1992
Magnetic resonance angiography compared to intra-arterial digital subtraction angiography in patients with subarachnoid haemorrhage A . G o u l i a m o s 1, E. Gotsis 2, L. V l a h o s 1, C. S a m a r a I, E. K a p s a l a k i 2, D . R o l o g i s 3, Z. Kapsalakis 2, and C. P a p a v a s i l i o u 1
Department of Radiology, University of Athens, 2 Diagnostic and Research Institute "Encephalos", 3D epartment of Neurosurgery, Athens General Hospital, Athens, Greece Received: 20 October 1991
Summary. In order to evaluate the sensitivity and specificity of magnetic resonance angiography ( M R A ) in spontaneous subarachnoid haemorrhage, 14 patients with recent h a e m o r r h a g e verified by CT or lumbar puncture were investigated with b o t h selective intra-arterial digital subtraction angiography ( I A - D S A ) and M R A by two independent teams, each having the same preangiographic information. The results were c o m p a r e d with each other and whenever possible (all positive cases except one) with those of surgical intervention. Seven patients were identified by M R A and I A - D S A as having a single aneurysm on the circle of Willis, 1 an aneurysm of the posterior inferior cerebellar artery 1 an aneurysm of the internal carotid artery (siphon) and 2 patients with two aneurysms on the circle of Willis. M R A and I A - D S A both failed to d e m o n strate aneurysms in 2 cases. T h r e e patients had negative results on b o t h methods and no surgical intervention was attempted. The aneurysms ranged f r o m 0.3 to 1.5 cm in size. In most cases there was agreement between M R A and D S A , leading us to believe that, if the p r o p e r protocols are followed, M R A is a powerful alternative to other established methods in the detection of intracranial aneurysms. A t this stage it will not replace I A - D S A prior to surgery, but the ability to obtain various projections using 3D M R A m a y improve surgical planning. K e y words: Aneurysm, cerebral - Cerebral angiography -
Magnetic resonance angiography
Magnetic resonance imaging (MRI) of subarachnoid h a e m o r r h a g e ( S A H ) has unique advantages: its sensitivity and specificity for the detection of non-acute h a e m o r rhage and blood b r e a k d o w n products and its potential for study of intracranial blood vessels for detection of aneurysms or arteriovenous malformations. With correct sequences at low field, M R I is m o r e sensitive than computed t o m o g r a p h y (CT) [1, 2]. While M R I is c o m m o n l y used for parenchymal changes routine application of flow effects for the study of the cerebral vasculature remains limited
[3-5].
The last few years have seen a rapid development of M R I techniques aiming at non-invasive imaging of intracranial and peripheral vasculature, usually based on 2D or 3D gradient echo pulse sequences with gradient m o m e n t refocussing ( G M R ) in the read-out and slice-select directions, or others including velocity encoding. These are referred to generically as magnetic resonance angiography (MRA). We have assessed the sensitivity and specificity of two techniques, a 3D time-of-flight (TOF) gradient echo sequence with first-order G M R and a 3D magnitude contrast pulse sequence (rephase-dephase), also with first-order G M R in the rephase data set. O u r results are therefore specific to these two methods. The aim of the present study was a comparison of M R A with intra-arterial digital subtraction angiography ( I A - D SA) and surgery in patients with spontaneous SAH.
Patients and m e t h o d s
Fourteen patients with spontaneous SAH were investigated with four vessels selective cerebral angiography on a digital subtraction unit, the results of which were not available to the MR operator or interpreter. There were 4 females and 10 males, 12~57 years old (mean 46). After routine MRI all patients underwent MRA. The two teams (DSA and MRA) had the same information prior to their respective examinations: the CT results and the history of the patient. Only when the results of both examinations were finalized were the findings compared and a surgical decision made. In most cases DSA preceded MRA by a few hours to i day, but in 5 cases MRA was carried out first, at the time of an MRI study. The MRA examinations were carried out on a 1.5 T superconducting MR imager, with very high homogeneity ( + 1.2 ppm in a 50 cm sphere). A circularly polarized head coil was used, and shimming was sometimes performed to correct inhomogeneities arising from paramagnetic fillings in teeth. The routine examinations included a series of axial Tl-weighted images [TR/TE 700/20 ms, 5 mm slice thickness, 256 x 256 matrix, 240 mm filed of view (FOV), and arterial presaturation], a series of 64 axial slices with a 3D rephase-dephase sequence, with partition thickness ranging from 1 to 1.5 mm, 200-230 mm FOV, 200 x 256 matrix and TRfrE 23/13 ms, 20 ~flip angle, and finally a series of 64 or 128 axial slices with a 3D TOF pulse sequence (FISP3D, TR/TE 25/7, 15 ~ flip angle, partition thickness 0.8-1.5 mm, 256x256
47
Fig. 1, a,b Digital subtraction angiography (DSA), left carotid arteriogram, anteroposterior and oblique projections, show normal vascular anatomy, without evidence of aneurysm, c 3D MR angiog-
raphy (MRA). Axial projection at a 25 ~rotation to the coronal plane also shows normal vascular anatomy
Fig.2. 3D MRA (case 1). Left anterior oblique projection with 30 ~ rotation allows precise localization of a 5 mm aneurysm on the anterior communicatingartery (arrow)
Fig.4. 3D MRA (case 7). Anteroposterior view obtained from a single 3D volume acquisition in the axial plane. The aneurysm is located at the trifurcation of the left middle cerebral artery (arrow)
Fig.3. 3D MRA (case 4). Left anterior oblique projection with 15 ~ rotation. The origin of the aneurysm from the left posterior communicatingartery (arrow) is well demonstrated
matrix). The MRA sequences included first-order GMR in the sliceselect and read-out directions. A series of 64 coronal 3D TOF slices was acquired in addition to or instead of the 3D TOF axial slices in some cases. In 2 patients examined early in the study, when magnitude contrast angiography was not available, only 3D TOF MR images were acquired with TR/TE 30/8 ms and other parameters as above. The 3D rephase-dephase sequence acquired two data sets (at the cost of doubling acquisition time), one with flow refocussing and one with flow defocussing. Substracting the second from the first resulted in a highly suppressed background, yieldingsuperb signal contrast of the vessels and discarding of high signal from static tissues simulating flow, such as subacute thrombus or haematoma. For the reconstruction of the angiograms a maximum intensity projection algorithm was used and run with an ultrafast image-processor (typical time for 12 projections with a 64-slice data set was 1 rain), allowing immediate visualization of the results of the first data set before planning the second or third. The speed of the image processor allowed the reconstruction of a large number of projections in a reasonable time so that a smooth hologram effect was simulated in cine mode. The cine mode was deemed valuable in resolving vessels which gave the impression of an aneurysm [6-8]. Correlation with IA-DSA performed within 1 week of the haemorrhage was available in all 14 patients. All studies included
anteroposterior, lateral and oblique projections, using a 0.3 mm focal spot and a caesium iodide image-intensifyingtube (Fig. 1 a, b). The MRA and spinecho studies were independentlyinterpreted by two neuroradiologists without knowing the history or the results of other imaging studies (Fig. 1 c). Aneurysm size was measured on IA-DSA and MRA studies in three dimensions (length, width, height). These three measurements were then averaged to give one number reflecting the anenrysm size. On IA-DSA allowance was made for magnification.
Results T h e results are s u m m a r i z e d in Table 1. T h e p r e s e n c e a n d site of the a n e u r y s m (or a n e u r y s m s in 2 cases) were correctly assessed b y M R A , in 12 of the 13 cases in which a n a n e u r y s m was c o n f i r m e d (92 %). T h e a n e u r y s m s were located as follows: 4 at the anterior c o m m u n i c a t i n g artery (Figs. 1, 5), 4 o n the m i d d l e c e r e b r a l artery (Fig.4), 3 at the p o s t e r i o r c o m m u n i c a t ing artery (Fig. 3), 1 at the origin of the p o s t e r i o r i n f e r i o r
48
Fig.5. a,b DSA (case 11). Right internal carotid artery, anteroposterior and lateral projections. There is suspicion of an aneurysm at the anterior communicating artery, e 3D MRA. Axial projection
allows precise identification of the 3 mm aneurysm at the origin of the anterior communicating artery (arrow)
cerebellar artery, and I on the internal carotid artery (siphon). The size of the aneurysm varied f r o m 0.3 to 1.5 cm and its neck (when present) was visualized by standard 3D T O F angiography. In case 9, one radiologist thought the D S A negative, but when the results of the M R A were revealed, all agreed that the D S A examination could be interpreted as positive. Because of the position of the aneurysm, and its pointing towards the sella turcica, it was not resolved well on DSA. Ten of the 11 DSA-positive patients were operated upon by two different neurosurgeons and the results were
confirmed surgically. The 11th has not yet been operated upon. Finally, 1 patient was operated upon on the results of M R A alone; an aneurysm on the middle cerebral artery was confirmed surgically (case 7). A n extensive haematom a was also present. Evidence of arterial spasm was seen in 4 cases and in 2 patients was identified by M R A . The aneurysms were located on the right in 9 cases and on the left in 4. In 2 cases m o r e than two aneurysms were diagnosed (cases 1 and 6). M R A did not show the second aneurysm in case 6. Other abnormalities seen included h a e m a t o m a s in 6 cases and infarcts in 2. The relative sizes of the aneurysms measured on M R A and I A - D S A are given in Table 1. The smaller aneurysms ( < 5 m m ) were accurately assessed using M R A . The relationship of the aneurysm to the parent vessel was easily visualized on M R A .
Table 1. Findings of intra-arterial digital subtraction angiography (IA-DSA) and magnetic resonance angiography (MRA)
Location of aneurysm Case 1 Case Case Case Case
2 3 4 5
Case 6 Case 7 Case 8 Case 9 Case 10 Case 11 Case 12 Case 13 Case 14
ACA (R)
~aCA (R) PCA (R) PCA (R) MCA (L) No aneurysm identified ACA (R) MCA (R) MCA (L) PICA (R) ICA (R) ACA (R) ACA (R) PCA (R) No aneurysm identified No aneurysm identified
Size (cm) MRA 0.5 0.3 0.3 0.5 0.5
0.5 Not visualized 0.5 1.0 1.5 0.5 0.3 0.5
IA DSA 0.8 0.5 0.5 0.5 0.7
0.3 0.2 1.0 1.5" 0.5 Not visualized 0.7
ACA, Anterior communicating artery; ICA, internal carotid artery; MCA, middle cerebral artery; PCA, posterior communicating artery; PICA, posterior inferior cerebellar artery "Initially one radiologist thought the DSA in this case negative
Discussion
All new methods must undergo scrutiny before they can be generally accepted in the medical establishment. In the case of M R A , there is a great variety of sequences to be tested and the final results can be site-specific as well, depending on magnet strength, eddy current compensation, homogeneity, etc., so that the pressure for scrutiny is even higher, if it is to be considered as an alternative to the wellestablished and tested D S A [9]. The highest resolution that can b e reached in 3D M R A varies, but it is not unreasonable to claim that less than i m m resolution in all three directions can be achieved with most 1.5 T imagers. O f course, such resolution does not m a t c h that of DSA, but the final question is: does one need higher resolution than that currently available? Obviously, most operable intracranial aneurysms are located on the circle of Willis, and are 2 m m in size or larger, within the detection limits of M R A , as this study and others have shown. Thus, one m a y be assured that aneurysms as small as 2 m m can be detected with 3 D T O F techniques, and if a h a e m a t o m a is present, making detection of the aneurysm difficult, magnitude contrast angiography or similar subtraction techniques can be used.
49 If time permits, one should use at least two different techniques for reassuring oneself that the findings are reproducible, as we have done in most cases. In the course of this study it became apparent that the highest possible resolution gave the most reliable results, so the last cases were examined under stricter conditions than the earlier ones. Our standard protocol now starts with a 3D T O F sequence (FISP 3D in the axial direction, T R / T E 21/7 ms, F O V 200 m m , 0.8 m m partition thickness, 256 x 256matrix). For comparison, a magnitude contrast sequence is used with a 1-1.25 m m partition thickness, 200 x 256 matrix and T R / T E 23/13 ms. Angiographic images are obtained in all possible directions so that all major vessels are well visualized. Display in cine m o d e is very helpful, and if doubt persists original slices are also examined. We found that if a lesion is not visible on the final angiogram no diagnosis can be m a d e from the original slices; thus, no effort has b e e n m a d e to rate detectability and specificity in terms of angiograms, angiograms + original, slices + cine mode, or any combination of the above with M R I , etc. We consider the reconstructed angiogram as the final conclusion, the findings of which can be confirmed by original slices, M R I etc., but not the reverse. The angiogram is of most use to the neurosurgeon. Conventional arteriography is considered the gold standard in the investigation of SAH. Additional studies which m a y help to identify asymptomatic aneurysms include CT, I V - D S A and M R A . CT and I V - D S A have limitations due to the spatial resolution and the need for contrast medium. M R A has potential value for visualization of intracranial vascular anatomy using techniques which allow increased contrast between vessels and surrounding tissues and appropriate manipulation of the images in the viewing console. The spatial resolution of 3D gradient-echo M R A is inferior to that of conventional angiography, but this is not a significant drawback, since aneurysms smaller than 3 m m in diameter rarely rupture [10, 11]. Limitations of M R A in the detection of intracranial aneurysms include: 1. Patient cooperation. 2. W h e n vasospasm is present, the 3D T O F technique tends to overestimate it. 3. Spatial resolution; it is expected that reliable ultra-high resolution (voxels with under 1 m m size) M R A of the circle of Willis willbe possible in less than 10 min.
The accuracy of M R A makes this m e t h o d ideal for screening patients presenting with headaches or other symptoms suggestive of possible intracranial aneurysm, before the development of SAH.
References 1. Jenkins A, Hadley D, Teasdale G, Condon B, Macpherson R Patterson J (1988) Magnetic resonance imaging of acute subarachnoid hemorrhage. J Neurosurg 68:431-736 2. Jenkins A, Paterson J, Hadley DM, et al. (1989) Use of in vitro magnetic resonance tissue studies to optimise pulse sequences in the imaging of intracranial hemorrhage. Magn Reson Imaging 7: 394-403 3. Ross JS, Masaryk T J, Modic MT, Ruggieri PM, Haacke EM, Selman WR (1990) Intracranial aneurysms: evaluation by MR angiography. AJNR 11:449-456 4. Masaryk TJ, Modic MT, Ross JS, Ruggieri PM, Laub GA, Lenz GW, Haacke EM, Selman WR, Wiznitzer M, Harik SI (1989) Intracranial circulation: preliminary clinical results with 3D (volume) MR angiography. Radiology 171:793-794 5. Gasparotti R, Orlandini A, Gualandi GF, Scipione V, Tansini A, Gnutti R Bonetti M, Lavezzi R Chiesa A, Galli G, Mearini M (1991) Uangiografia a risonanza magnetica nello studio del circolo cerebrale. I. Aneurismi cerebrali. Rev Neuroradiol 4: 153166 6. Spitzer CE, Blinder RA (1989) Vascular applications of MRI. Magn Reson Q 5:205-227 7. Weeden VJ, Rosen RR, Brady TJ (1987) Magnetic resonance angiography. Magnetic resonance annual. Raven Press, New York 8. Edelman RR, Mattle HR Atkinson D J, Hoogewould HM (1990) MR angiography. AJR 154:937-946 9. Pernicone JR, Siebert JE, Potchen E J, Pera A, Dumoulin CL, Souza SP (1990) Three-dimensional phase-contrast MR angiography in the head and neck: preliminary report. AJNR 11: 457466 10. Locksley HB (1969) Natural history of subarachnoid hemorrhage, intracranial aneurysms, and arteriovenous malformations. In: Sahs AL, Perret GE, Locksley HB, et al. (eds) Intracranial aneurysms and subarachnoid hemorrhage. A cooperative study. Lippincott, Philadelphia, pp 3%108 11. McCormick WF, Agosta-Rua GJ (1970) The size of intracranial saccular aneurysms. An autopsy study. J Neurosurg 33:422-427
Dr. A. Gouliamos Department of Radiology Areteion University Hospital 76, Vas. Sophias Ave. GR-115 28 Athens Greece
