CimJioval>c Inlervenl RlIdiol (1992) 15:23-31
Cardio\ascular
and IntenuJtional
Radiology
e
Springer.Vo:rlag No:w York Inc .
1 99~
Magnetic Resonance Coronary Artery Imaging Cy nthia B. Pasc haL l., E. Mark Haacke,'-" Lee P. Adler. ' and Daniel A. Finelli I I Depanmenl
'Physics.
of Radiology. U niversily Hospilals of Cleveland. Cleveland. Ohio: and Depurlmo:nts of ' Biomedical Engineering and Weslern Reserve Unive rsity. Cleveland. Ohio. USA
Ca~e
Abstract . Coronary artery imaging with magnetic resonance imaging carries the potential for a noninvasive, essen lially ri sk-free scree ning test for those at risk for coronary heart di sease. Many physiologic and anatomic c hallenges incl uding cardiac and respiralory motion and Ihe small size, tortuosity, and variable flow charactc ristics ofthc coronary arteries hamper effort s to achicve thi s goal. Thi s article reviews the cfforts o f several research groups to su rmount t hese d ifficulties through the use of20 and 3D tec hniques; s pin echo, gradient echo, and ultrafast sequences; satu ration pulses: and cont rast agents. Promising resu lt s have been and continue to be reported although no obv iou s o ptimal solution has yet been determi ned. Key words: Coronary arteries-Coronary angiograph y- Magneti c resonance imaging- Magnetic reso· nance angiography
Magnetic rcsonance angiography (MRA) has been used 10 image the ce rebral vasculature with submil· limeter resolution . The coro nary arteries are comparable in size 10 many of the vessels clearly seen in MRA of the head , yet the lechnique is not yet routinely used clinically in their evaluation. The diffic ult y with MRA of the coronaries lies in the presence of cardiac and respi ratory motion which. if left unaccounled for, significant ly reduces resolulion. For thi s reason, efforts in magnet ic resonance imaging of the coronary arteries have been likened to Ihe quest for the hol y grail. Indeed. the claim that magnetic resonance imaging (MRI) of the coronaries "may in fac t prove to be a technical imposs ibilit y" has been made III. Howeve r. the A ddrt'H rt'pr;nt r;'4!1t.~I.f 10: Cynt hia B. Paschal. Magn O:l ic Reso·
nance Imuging. University Hospitals of Cleveland. 2074 Abington Rood . Cleveland. OH 44106. USA
cl ini cal ml.ltivalion is so strong that a number of researcher groups contin ue to pursue the goal of coronary MRA . In fact, there are a variety of obstacles to be dealt wit h. Among them are cardiac and res piralory motion. the small size and to rtuos ity of the coronary arteries, variations in fl ow rate, and com peting signal from neighboring blood pools. To dale. most effort s to image the coronaries wit h MR have been based on time-of-flight (TO F) methods although some velocit y phase mapping has been done (2]. TOF meth ods are less se nsit ive to motion than phase contrast tec hniq ues: this is an important co nsideralio n in cardiac imaging. Saturatio n techniques can al so be used to tag blood or remove unwanted signal. Both twoand three-dimensional (20 a nd 3D) method s have bee n employed with ne ither method yet demon strating s upe ri ority. In addition to basic MRA technique s. 20/3 0 postprocess ing methods are being uti lized to prese nt the acqu ired data in an optimal format. The purpose of this review is to high light past and present effort s which may lead to MR coronary angiography becoming a clinical reality .
Motivation Coronary heart disease affl icts about 5 mi ll ion individuals in the U.S. and is Ihe leading cause of death today. killing over 500,000 annuall y (31. No nin vasive or minimall y invasive diagnost ic procedures such as electrocardiograms or SPECT thall ium perfusion stud ies can revea l regio nal electrical conduction defects and myocardi al perfusion defects, re spectivel y. which are frequent ly associated with coronary heart disease. However. accurate localization of coro nary artery stenosis or occlusion requ ires a coronary a ngiogram. Convent ional X-ray coronary angiography is an
in vasive procedure and invo lves the risks of damage to vessel lining. dislodgement of embo li . cardiac arrhythmias. myocardial infarctions. all ergic reactio ns to the cont rast medium. and co ntrast nephrotoxicity. These ri sks can be fatal and. in fact. a monality rate of 0.1-0.3 % has bee n reported for coronary angiography [4). A convent ional coronary angiogram provides high quality projection images with exce llent temporal and spatial resoluti on. yet the associated risks. monetary costs. an d required hospitalization preclude its use as a sc reening test in populations at risk for coronary heart disease. Magneti c resonance· angiogra ph y I S free of these risks and offers the potential for a safe. effective 3D sc reening test. Physiologic and Anatomic Considerations Th e physiologic and analOmic characte ri stics of the co ronary arteries present a number of challenges to MRA. The coronaries are small in size. tortuous. and have multiphas ic pul sat ile How. These factors alo ne would not prohibit successful MRA but car· diac and respiratory motion greatl y increase the d if· ficu lty of the task. Also. signal from blood in the atria and ventric les can obscu re signal from the coronary arteries in the images. At their largest caliber. the right and left main coronary arteries (RCA. LM ) measure roughly 4-5 mm in diamete r. The LM bifurcates withi n 2 cm of its origin into the left anle rio r descending (LAD) and ci rcumflex (eX) whic h are about 2-4 mm in diameter. If a 50% stenos is of a 2-mm diameter artery is to be visualized. then the criterion of two pixe ls per st ructu re mand ates that resolution be at least 0.5 mm. Thio:; re solut ion with a sufficient signa lto-noise ratio (SN R) is difficult to achieve with body coil MRI of adult humans . The tortuosit y of the co ronary arteries pre· cludes selection of a single imagi ng plane perpendicular to the arteries which could capital ize on through plane Ho w. Coronary blood suppl y comes from retrograde flow down the aortic root wi th the bu lk of LM flow su pplied during early di astole and so mewhat more than half the RCA How supplied during systole. The flow through the coronaries va ri es greatly o ve r the cardiac cycle. ranging from o 10 100 ml / min [5). with peak linear velocities of about 30 cmlsec for the LM and 22 cm/sec for the RCA [6) occurring in early diastole. In the LM . the peak flow rate in early diastole is more than twice the peak rate during systole whereas in the RCA. the peak flow rate in diastole is eq ual to or slightly less than the maximum rate during systole [5J . In both vesse ls. there is marked decele ratio n
of flow at the start and end of systole. As cardiac motion is more extreme during systole. bette r image quality is obtained duri ng diastole which is o ptimal for imaging How in the LM. CX. and LAD but is less optima l for the RCA. Synchronizing data acqu isition to the e lect rocardiogram wit h prospective [7) or retrospective [7.8) methods eliminates much of the blurring and ghosting due to cardiac motion. Ho wever. such synchron izat ion cannot eliminate all errors as the heart does not return to the same position from one cycle to the next. Retrospective methods based on heart position exist [9. 10J but lack the accuracy to eliminate all blurring and ghosting e rrors [ I I). Thus. though synchronizatio n reduces much of the blurri ng and ghosting. due to card iac motion. some will remain. Respiratory motion also introduces blurring and ghosting in images of the thorax. During respiration . the heart moves in the cran io-caudal and anteriorposterior direct ions. Unfortunately. the period of the respiratory cyc le is so long (4-6 sec) that for conventiona l 20 o r 3D imaging methods. sy nchronizing or gat ing for respiration wou ld result in im· practically long scan times. Other methods {O deal with respiratory motion range from the si mple utilization of a rest ra ining be lt around the thorax to mathematical corrections using monitor echoes [12. Steagall and Haacke, unpub li shed datal. The latte r method has been shown to significa ntly reduce blurring and ghosting from the edge of the abdomen and heart (Steagall and Haacke) . Coronary Imaging Strategies As research in coronary MRA has grown. a wide variety of technique s have been tested. The technique s can be categorized by the met hods used to control contrast. obta in desired temporal and spatial re solution . and process and di splay the final images. Contrast in co ro nary MRA has been cont rolled by use of spin echo. grad ien t echo FlSP and FLASH, and sat uration schemes incl uding inversion recovery and STEAM (STimulated Echo Acquisit ion Mode) [ 14] seq uences as well as othe r saturation pulses. Temporal and spatial re soluti on issues have been addre ssed with 2D and 3D methods. cine. ultrafast. and echo planar imaging (EPI) methods. Postprocessi ng and di splay techniques publ ished in MRI li terat ure have included seeded region growing. thresholding. subject ive editing. multip lanar reconstructions (MPR ). maximum intensity projections (MIP) . and surface re nderi ngs. In the following sections. technique. co ntrast. and resolution issues will be addressed.
C.B . Pasch.1I el a!. : MR Coronary Arlery Im,,!;inl?-
Fig. I. Spin echo (A) amI @:rmlienl echo i B) Imnsverse ima@:e~ through the heart demon strating a patent saphenous vein graflto the RCA htraight arrow) seen as a void on spin echo and bright on @:radientecho. A patent graCt from the !eft internal m"mmary arl ery !O the LAD is al~o ~ee n (curved arrow I
Cardio\ascular
and IntenuJtional
Radiology
e
Springer.Vo:rlag No:w York Inc .
1 99~
Magnetic Resonance Coronary Artery Imaging Cy nthia B. Pasc haL l., E. Mark Haacke,'-" Lee P. Adler. ' and Daniel A. Finelli I I Depanmenl
'Physics.
of Radiology. U niversily Hospilals of Cleveland. Cleveland. Ohio: and Depurlmo:nts of ' Biomedical Engineering and Weslern Reserve Unive rsity. Cleveland. Ohio. USA
Ca~e
Abstract . Coronary artery imaging with magnetic resonance imaging carries the potential for a noninvasive, essen lially ri sk-free scree ning test for those at risk for coronary heart di sease. Many physiologic and anatomic c hallenges incl uding cardiac and respiralory motion and Ihe small size, tortuosity, and variable flow charactc ristics ofthc coronary arteries hamper effort s to achicve thi s goal. Thi s article reviews the cfforts o f several research groups to su rmount t hese d ifficulties through the use of20 and 3D tec hniques; s pin echo, gradient echo, and ultrafast sequences; satu ration pulses: and cont rast agents. Promising resu lt s have been and continue to be reported although no obv iou s o ptimal solution has yet been determi ned. Key words: Coronary arteries-Coronary angiograph y- Magneti c resonance imaging- Magnetic reso· nance angiography
Magnetic rcsonance angiography (MRA) has been used 10 image the ce rebral vasculature with submil· limeter resolution . The coro nary arteries are comparable in size 10 many of the vessels clearly seen in MRA of the head , yet the lechnique is not yet routinely used clinically in their evaluation. The diffic ult y with MRA of the coronaries lies in the presence of cardiac and respi ratory motion which. if left unaccounled for, significant ly reduces resolulion. For thi s reason, efforts in magnet ic resonance imaging of the coronary arteries have been likened to Ihe quest for the hol y grail. Indeed. the claim that magnetic resonance imaging (MRI) of the coronaries "may in fac t prove to be a technical imposs ibilit y" has been made III. Howeve r. the A ddrt'H rt'pr;nt r;'4!1t.~I.f 10: Cynt hia B. Paschal. Magn O:l ic Reso·
nance Imuging. University Hospitals of Cleveland. 2074 Abington Rood . Cleveland. OH 44106. USA
cl ini cal ml.ltivalion is so strong that a number of researcher groups contin ue to pursue the goal of coronary MRA . In fact, there are a variety of obstacles to be dealt wit h. Among them are cardiac and res piralory motion. the small size and to rtuos ity of the coronary arteries, variations in fl ow rate, and com peting signal from neighboring blood pools. To dale. most effort s to image the coronaries wit h MR have been based on time-of-flight (TO F) methods although some velocit y phase mapping has been done (2]. TOF meth ods are less se nsit ive to motion than phase contrast tec hniq ues: this is an important co nsideralio n in cardiac imaging. Saturatio n techniques can al so be used to tag blood or remove unwanted signal. Both twoand three-dimensional (20 a nd 3D) method s have bee n employed with ne ither method yet demon strating s upe ri ority. In addition to basic MRA technique s. 20/3 0 postprocess ing methods are being uti lized to prese nt the acqu ired data in an optimal format. The purpose of this review is to high light past and present effort s which may lead to MR coronary angiography becoming a clinical reality .
Motivation Coronary heart disease affl icts about 5 mi ll ion individuals in the U.S. and is Ihe leading cause of death today. killing over 500,000 annuall y (31. No nin vasive or minimall y invasive diagnost ic procedures such as electrocardiograms or SPECT thall ium perfusion stud ies can revea l regio nal electrical conduction defects and myocardi al perfusion defects, re spectivel y. which are frequent ly associated with coronary heart disease. However. accurate localization of coro nary artery stenosis or occlusion requ ires a coronary a ngiogram. Convent ional X-ray coronary angiography is an
in vasive procedure and invo lves the risks of damage to vessel lining. dislodgement of embo li . cardiac arrhythmias. myocardial infarctions. all ergic reactio ns to the cont rast medium. and co ntrast nephrotoxicity. These ri sks can be fatal and. in fact. a monality rate of 0.1-0.3 % has bee n reported for coronary angiography [4). A convent ional coronary angiogram provides high quality projection images with exce llent temporal and spatial resoluti on. yet the associated risks. monetary costs. an d required hospitalization preclude its use as a sc reening test in populations at risk for coronary heart disease. Magneti c resonance· angiogra ph y I S free of these risks and offers the potential for a safe. effective 3D sc reening test. Physiologic and Anatomic Considerations Th e physiologic and analOmic characte ri stics of the co ronary arteries present a number of challenges to MRA. The coronaries are small in size. tortuous. and have multiphas ic pul sat ile How. These factors alo ne would not prohibit successful MRA but car· diac and respiratory motion greatl y increase the d if· ficu lty of the task. Also. signal from blood in the atria and ventric les can obscu re signal from the coronary arteries in the images. At their largest caliber. the right and left main coronary arteries (RCA. LM ) measure roughly 4-5 mm in diamete r. The LM bifurcates withi n 2 cm of its origin into the left anle rio r descending (LAD) and ci rcumflex (eX) whic h are about 2-4 mm in diameter. If a 50% stenos is of a 2-mm diameter artery is to be visualized. then the criterion of two pixe ls per st ructu re mand ates that resolution be at least 0.5 mm. Thio:; re solut ion with a sufficient signa lto-noise ratio (SN R) is difficult to achieve with body coil MRI of adult humans . The tortuosit y of the co ronary arteries pre· cludes selection of a single imagi ng plane perpendicular to the arteries which could capital ize on through plane Ho w. Coronary blood suppl y comes from retrograde flow down the aortic root wi th the bu lk of LM flow su pplied during early di astole and so mewhat more than half the RCA How supplied during systole. The flow through the coronaries va ri es greatly o ve r the cardiac cycle. ranging from o 10 100 ml / min [5). with peak linear velocities of about 30 cmlsec for the LM and 22 cm/sec for the RCA [6) occurring in early diastole. In the LM . the peak flow rate in early diastole is more than twice the peak rate during systole whereas in the RCA. the peak flow rate in diastole is eq ual to or slightly less than the maximum rate during systole [5J . In both vesse ls. there is marked decele ratio n
of flow at the start and end of systole. As cardiac motion is more extreme during systole. bette r image quality is obtained duri ng diastole which is o ptimal for imaging How in the LM. CX. and LAD but is less optima l for the RCA. Synchronizing data acqu isition to the e lect rocardiogram wit h prospective [7) or retrospective [7.8) methods eliminates much of the blurring and ghosting due to cardiac motion. Ho wever. such synchron izat ion cannot eliminate all errors as the heart does not return to the same position from one cycle to the next. Retrospective methods based on heart position exist [9. 10J but lack the accuracy to eliminate all blurring and ghosting e rrors [ I I). Thus. though synchronizatio n reduces much of the blurri ng and ghosting. due to card iac motion. some will remain. Respiratory motion also introduces blurring and ghosting in images of the thorax. During respiration . the heart moves in the cran io-caudal and anteriorposterior direct ions. Unfortunately. the period of the respiratory cyc le is so long (4-6 sec) that for conventiona l 20 o r 3D imaging methods. sy nchronizing or gat ing for respiration wou ld result in im· practically long scan times. Other methods {O deal with respiratory motion range from the si mple utilization of a rest ra ining be lt around the thorax to mathematical corrections using monitor echoes [12. Steagall and Haacke, unpub li shed datal. The latte r method has been shown to significa ntly reduce blurring and ghosting from the edge of the abdomen and heart (Steagall and Haacke) . Coronary Imaging Strategies As research in coronary MRA has grown. a wide variety of technique s have been tested. The technique s can be categorized by the met hods used to control contrast. obta in desired temporal and spatial re solution . and process and di splay the final images. Contrast in co ro nary MRA has been cont rolled by use of spin echo. grad ien t echo FlSP and FLASH, and sat uration schemes incl uding inversion recovery and STEAM (STimulated Echo Acquisit ion Mode) [ 14] seq uences as well as othe r saturation pulses. Temporal and spatial re soluti on issues have been addre ssed with 2D and 3D methods. cine. ultrafast. and echo planar imaging (EPI) methods. Postprocessi ng and di splay techniques publ ished in MRI li terat ure have included seeded region growing. thresholding. subject ive editing. multip lanar reconstructions (MPR ). maximum intensity projections (MIP) . and surface re nderi ngs. In the following sections. technique. co ntrast. and resolution issues will be addressed.
C.B . Pasch.1I el a!. : MR Coronary Arlery Im,,!;inl?-
Fig. I. Spin echo (A) amI @:rmlienl echo i B) Imnsverse ima@:e~ through the heart demon strating a patent saphenous vein graflto the RCA htraight arrow) seen as a void on spin echo and bright on @:radientecho. A patent graCt from the !eft internal m"mmary arl ery !O the LAD is al~o ~ee n (curved arrow I
