Cardiovasc Int.:rvenl Radiol
( 199~)
15:32-42
Cardio\!iscuJar
and InterventionaI
Radiology
c: Springer-Verlag New York
Inc. 1991
Single-Shot Magnetic Resonance Imaging: Applications to Angiography Adrian P. Crawley, Mark S. Cohen, E. Kent Yucel, Brigitte Ponce leI. and Thomas J . Brady Massachusens General Hospital NMR Center. Charlestown . Massachuseus. USA
Abstract_ Recently developed technologies that allow the collection of magnetic re sonance imaging (MRI) in as little as 26 msec have been explored in their application to angiography. Advantage s are demonst rated in scan time reduction. insensitivity to patient motion (especially in abdominal applications), flow quantification, and temporal resolution. We demonstrate that because suc h single-shot techniques are inherently resistant to flow dephasing during acquisition that allow for sustained high signal intensities to be ach ieved when images must be combined through the cardiac cycle. Such high temporal resolution scans may be utilized for the collection of time-resolved angiograms. With these techniques we demonst rate the collection of complete MR angiograms in the course of reasonable 10-25 sec breath holds . The relati ve simplici ty of the technique. coupled with its overall short acquisi tion time. allows us (0 inco rporate a ngiography inla other imaging pro{Qcols without adding significant time burdens. Result s to dat e are promising for further impro vement s in spatia l re sol ution. without extension of sca n time . Key words: MR angiography-Instascan-Echo planar imaging
In recent years it has become practical , using "single-shot"' approaches with s pecialized hardware. to acquire complete magnetic resonance (MR) image s in as liule as 26 msec. The real-time lnstascan technique [II, for example, allows the collection of complete MR movies during the course of a single heartbeal. Using this method, we have obtained movies. Mark S. Cohen . Ph.D .. Massachu· seIlS General Hospital NMR Center. Building 149. Thirteenth Street. Charlestown. MA 02129, USA
Addrl'ss rl'prim f/'quI'SlS to :
free of motion artifacts, with contrast simi lar to cine [2\, but without the need for prospective or retrospective gat ing. The method has also been used to track the passage ofa bolus of cont rast agent through the brain, enabling relative cerebral blood volume maps to be generated [3. 4]. We have recently explored some of the unique capabilities offered by single-shot scann ing in magnetic resonance angiography (MRA). Broadl y speaking , the single-shot method offers three advantages: reduction of examination times, enabling complete angiograms to be obtained within a single breath hold: relative resistance to flow- and motioninduced signal loss: and enhanced temporal resolution. such that signal variations within the card iac cycle may be studied readily. The overall goals of this researc h have been to circumvent the several problems of conve ntional MRA method s. as outlined below, and to deve lop techniques enabling us to acquire high qualit y abdominal angiograms in the space of a few breath holds. In this article we will review briefly the Instascan method, which we use for si ngle-shot imaging, and we will describe several of the ways that we have collected angiographic data using single-shot technology . We will consider applications of the si ngle-shot technique to the problems of quantitation, and we will discuss the potential limitations and advantages of these techniques as compared to other available MRA methods. Single-Shot Scanning The MR signal is it self transient. decaying at the rate T~ Which, for biological tissues, is typically less than 100 msec [5]. Spatial encoding in MR takes some time, however. and must be completed before the signal has decayed significantly. Ultimately. the rate at which the signal can be encoded s patially is deter-
A. P. Crawley scanner retrofitted with a commercially available Advanced NMR Systems' highspeed imaging package. The Advanced NMR retrofit incorporates hardware modifications to several ke y components of the Signa. includ ing the gradient s. RF. and data acqui si tion subsystem . Figure I shows the tim ing diagram fo r a typical gradient echo In stas-
can sequence. modified for flow compensation in the slice selection and phase-encod ing directions . Th e MIPs were performed with the Mercury MC3200 coprocessor. which is used for ··real -l ime " image processing in standard in stasca n imaging: each MIP was formed in a fraction of a sec . To demonstrate the Instascan time-of-flight angiography method. Figure 2 shows a coronal angiogram in a normal volunteer. The study was completed in four breath hold s. each averagingjust under 20 sec. Six images were acquired with an echo lim e (TE) of 10 msec and a TR of 150 msec at each of 53 axial slice location s (a to tal of 318 images in 80 sec of imaging). The in-plane reso lution for each slice was 3 mm and the slice thickne ss was 6 mm. yielding a total su perio r-to-infe rior volu me coverage of 3 1.8 cm. Flow compensation (sec below) was not used. At this TR. we found that a 90° flip angle produced good static signal suppression. Four "co ndition ing· · pul ses preceded the acquisition at each location to bring the signal into equilibrium. Figure 3 demo nstrates thi s approach to equilibrium for a single sl ice plane: successive images show improved contrast between blood and statio na ry ti ssue . Because the images are acqu ired in a time sherI compared with the ordinary R-R interval. timeresolved cine angiograms can be produced. Adding the dimension of time-re solution requires sorting the individual images from each slice locat ion according to their latency from the card iac R wave, then calculating separate projection images at each time point. We elected to use a retrospective, rather than a pros pective, cardiac ordering because of its ease of setup. its relative resistance to errors in triggering. and the higher sampling densit y per breat h hold (trig> gered methods wou ld necessaril y introduce pau ses into the imaging expe riment). Figu re 4 shows four frames from a time-resolved angiographic movie of the abdomen acquired in 48 sec of imaging . As flow ve locity drops to zero. a result of cardiac pul sa tion . the signal in the aorta and vena cava is seen to decrease . Th e po rtal veins are see n clearly between the aorta and vena cava and show non pulsatile flow as expected . Suc h time-re solved studie s arc most conve nientl y viewed in cine (movie) format. We have refined the basic angiography method for imaging of the portal ve in . where comparativel y slow flow s are present. and where the flow rate is not a strong function of the cardiac cycle . Twelve axial contiguous slices with 5 mm thi ckness are typically acquired. with in-plane resolution of 3 x 1.5 mm. A TR of 100 msec and a 90° flip angle produce good angiograp hic contrast. Because the flow is nonpulsatile. only four time points are required at each level. Figure 5B shows the axial M1P for a patient who was being evaluated for a liver transplant.
,' .P. Cr. ,wiram formed from eight images ill ea,h of)'J ~Ji,e, . E,L,h of the four imalles in lhb figure rep rese nlS an .mgiogr.lm at a ~uc,essi vely laler ph>l~e in the cOL rd iac cyc te . OLS ind icOLted above ea,h inlOLge . The OLonOL. inferior vena ,ava. and hepalic vei ns show obviou ~ c hanl>e~ in signal intensity during vascul;lr pub;uion. whereas the portOLI vein Irunk shw~ li ul e puls'J! ilily. Slice thicknes s - 6 mm. TR >= I~~ msec. TE '" IJ m ~e"
the readout gradient polarity produces only a small sensitivity with an even echo rephasing between successive lines of data. Model ing of these effects on the signal [39) suggests that the dominant phase shifts and signal losses will occur as a resu lt of the slice selection and phase-encoding gradients, obviating the need for velocity compensation of the read out grad ient. Contrary to what might be expected from conventional MRI. we found that at a TE of 15 msec and with a partial k-space acqui sition, flow compensat ion produced negligible improvement in the angiograms. At least two mechan isms are important in loss of signal from How dephasing: de phasing, and signal cancellation with in a tissue voxe l, and signal cancellation from phase shifts which differ from one line to the next in the MR raw data set. It is notable that the line-to-line ph ase variations are fundamentally minimized with single -shot approaches because a limited time window ex ists during which such vari ations can develop. In light of these data, we preli minarily conclude that in travoxe l dephasi ng represen ts a relatively small component of the motion-related signal losses. at least in normal volunteers. a conclusion supported by other data , as discussed below. Echo Time and Conjllg{l(e Synthesis OUf imager is capable of acquiring two raw data li nes per msec . A 64 x 128 image thu s requires a total
encoding period of 32 msec . If the echo is constrained to fall at the center of k-s pace . a ful l 16 msec must elapse prior to the ec ho ce nter. Th is . coupled with the requirement s fo r slice selection and appropriate pre-encoding. resu lt s in a relatively long minimum TE. for angiograph ic imaging. of 23 msec. Modificat ion s to the method [81 have reduced the ec ho time by tak ing advantage of the conjugate sy nthesi s technique [40). Here. it is poss ible to acqui re just a few raw data lines prior to TE. then to use complex conjugation to calc ulate the missing data lines. Therefore. the conj ugate sy nthesis method ca n be expected to have certai n li mitations in the presence of mot ion . The technique depends upon the ex istence of complex conjugate symmetry expected in a data set for an image with no phase difference s across it [41]. but this condi tio n is necessarily violated in moving sampl es. Acquisition of slight ly more than half of the raw data set enables proper phase correction of low spatial frequency variations in phase. but cannot correct for phase variations in small featu re s. and is thus ineffective in correcting the phase shifts from intravascu lar Hows. Our hypothesis the n. is that some loss in signal or resolution will occur when partial acquisitions are used. With the conjugate sy nthesis. or " partial-k: ' method we were able to achieve minimum echo time s of 10 msec without moment nUlling . and 14 msec with gradient moment compensation. Compari sons were made of acquisitions covering 100% of k·space at a TE of 31 msec (Fig. 6A) and 62.59C
A.P. Cmwle y et "I. ' Single·Shol MRJ
37
ety of time points with respect to cardiac activity and to sum the resulting image s. Sampling of the Cardiac Cycle
Fig. 5. A Axial (top) and coronal (bottom) high speed MRAs of th e abdominal vasculature in a normal volunteer. TR "" 100 msec. TE "" 10 msec. 3 mm slice. four repetit io ns at each slice level. A toml of) I slices were acquired in a single breath hold. B Instascan axial MIP through 12 sl ices obtained in a single breath hold and show ing extensive peritoneal varices and a portal vein occlusion in a patien t evaluated for live r transplant. TR = 100 msec. Flip angle = 90". TE .. 18 ms ec (with tirst order flo w compensation ). Slice thickness 5 mm. in·plane resolution 3 mrn x 3 mm.
of k-space (Fig. 68) at a TE of 10 msec. Note that the partial-k method shows some loss of signal as compared with the full data set. even at much shorte r echo times. The short TE scan. however, does appear to perform better in the larger vessels. Pulsatile Signal Variations
As demonstrated in the time-resolved angiograms shown above, the time-of-flight signal intensity varies considerably throughout the cardiac cycle as a consequence of variations in blood flow velocity. When single static images representing the vascular system are desired, it is necessary to sample a vari-
The signal variations through the cardiac cycle differ for the arterial and venous systems. ranging from the highly pulsatile flow in the descending aorta to the near continuous flow in the portal venous sys· tem. Different strategies shou ld therefore be used to collect data from the various portions of the vascular system. To represent accurately the blood signal in the veins generally requires sampli ng of only a sma ll number of time points; peak signals in the arteries, on the other hand. are achieved only during a small portion of the cardiac cycle , where flow velocities are maximal. Our initial objective in developing an ln slascan angiography sequence was to design an abdominal exam that required, at most, three breath hold s. Although conventional short TR MRA can be used . with each slice location acqu ired during an 8-sec breath hold, the scan requires considerable patient cooperation. Breath-holding after expiration ensures that the position is reasonably reproducible . but slice misregistration is always a potential prob· lem. By using the lnstascan method, we are able to image mUltiple time frames for each slice , with a scan time of approximate ly I sec/sli ce to completely cover the cardiac cycle. Therefore. it is always possible to cove r a reasonable imaging volume within an acceptable breath hold of 10-25 sees. On the other hand. when suc h complete coverage of the cardiac cycle is not required, it is possible to increase the volume coverage appropriate ly while st ill maintaining the overall short acquisition times . Summation: Magnitude Averaging, Complex A veraging , and MIP
The conve ntional MR image is formed as the twodimensional Fourier transform of data co llected line by line during repeated, separate acquisitions. Under such circumstances, phase shifts in the data due to sample variations during the collection of the im· age result in some degree of signal cancellation and loss. In many physiological examination s, such phase shifts arise largely from motion during the exam, and are particularly significant in regions of turbulent fluid flow . As discussed above. the singleshot technique is resistant to such line-by-line variations, as the entire raw data set is acquired over a very short period . When multiple images are acquired through the cardiac cycle, however. such
"
,\ . P. Cr:mie )'':1 "I : SIO.\!Ic-Sh.ll Mltl
t"ig. 6 . Comp; lf i'on uf t hc traJe ·"rr bc["~cn
TE "oJ
),;-~P"Cc
cuva"!;!,, . A
TE • IU m,c..:. (,:,~'fl k·~pace a..:qui,j,iun 18 ) TE - .1 1 m,,:c 11K!'"/ . I.>,p,,~·e "C4ul",;un . The full It· 'pace "ellui,iliun pr,)lj lJce, belh: r sm,,11 \'0:,,0:1 cun 'picuily. t-ut Ihc ,hurler T E ,(;,,0 I!!i,-c\ beHt'r ',gnal in Ihc I"rgcr 11(,,,:1, , Cumpln l ei "nJ m:'tlnilu Je :.I" cr.'g.:' jD) uflhe I! repeilliun, fur each ,licc prooJu(t' " re,uit inferior to lhc M IP of Ihc ,
( 199~)
15:32-42
Cardio\!iscuJar
and InterventionaI
Radiology
c: Springer-Verlag New York
Inc. 1991
Single-Shot Magnetic Resonance Imaging: Applications to Angiography Adrian P. Crawley, Mark S. Cohen, E. Kent Yucel, Brigitte Ponce leI. and Thomas J . Brady Massachusens General Hospital NMR Center. Charlestown . Massachuseus. USA
Abstract_ Recently developed technologies that allow the collection of magnetic re sonance imaging (MRI) in as little as 26 msec have been explored in their application to angiography. Advantage s are demonst rated in scan time reduction. insensitivity to patient motion (especially in abdominal applications), flow quantification, and temporal resolution. We demonstrate that because suc h single-shot techniques are inherently resistant to flow dephasing during acquisition that allow for sustained high signal intensities to be ach ieved when images must be combined through the cardiac cycle. Such high temporal resolution scans may be utilized for the collection of time-resolved angiograms. With these techniques we demonst rate the collection of complete MR angiograms in the course of reasonable 10-25 sec breath holds . The relati ve simplici ty of the technique. coupled with its overall short acquisi tion time. allows us (0 inco rporate a ngiography inla other imaging pro{Qcols without adding significant time burdens. Result s to dat e are promising for further impro vement s in spatia l re sol ution. without extension of sca n time . Key words: MR angiography-Instascan-Echo planar imaging
In recent years it has become practical , using "single-shot"' approaches with s pecialized hardware. to acquire complete magnetic resonance (MR) image s in as liule as 26 msec. The real-time lnstascan technique [II, for example, allows the collection of complete MR movies during the course of a single heartbeal. Using this method, we have obtained movies. Mark S. Cohen . Ph.D .. Massachu· seIlS General Hospital NMR Center. Building 149. Thirteenth Street. Charlestown. MA 02129, USA
Addrl'ss rl'prim f/'quI'SlS to :
free of motion artifacts, with contrast simi lar to cine [2\, but without the need for prospective or retrospective gat ing. The method has also been used to track the passage ofa bolus of cont rast agent through the brain, enabling relative cerebral blood volume maps to be generated [3. 4]. We have recently explored some of the unique capabilities offered by single-shot scann ing in magnetic resonance angiography (MRA). Broadl y speaking , the single-shot method offers three advantages: reduction of examination times, enabling complete angiograms to be obtained within a single breath hold: relative resistance to flow- and motioninduced signal loss: and enhanced temporal resolution. such that signal variations within the card iac cycle may be studied readily. The overall goals of this researc h have been to circumvent the several problems of conve ntional MRA method s. as outlined below, and to deve lop techniques enabling us to acquire high qualit y abdominal angiograms in the space of a few breath holds. In this article we will review briefly the Instascan method, which we use for si ngle-shot imaging, and we will describe several of the ways that we have collected angiographic data using single-shot technology . We will consider applications of the si ngle-shot technique to the problems of quantitation, and we will discuss the potential limitations and advantages of these techniques as compared to other available MRA methods. Single-Shot Scanning The MR signal is it self transient. decaying at the rate T~ Which, for biological tissues, is typically less than 100 msec [5]. Spatial encoding in MR takes some time, however. and must be completed before the signal has decayed significantly. Ultimately. the rate at which the signal can be encoded s patially is deter-
A. P. Crawley scanner retrofitted with a commercially available Advanced NMR Systems' highspeed imaging package. The Advanced NMR retrofit incorporates hardware modifications to several ke y components of the Signa. includ ing the gradient s. RF. and data acqui si tion subsystem . Figure I shows the tim ing diagram fo r a typical gradient echo In stas-
can sequence. modified for flow compensation in the slice selection and phase-encod ing directions . Th e MIPs were performed with the Mercury MC3200 coprocessor. which is used for ··real -l ime " image processing in standard in stasca n imaging: each MIP was formed in a fraction of a sec . To demonstrate the Instascan time-of-flight angiography method. Figure 2 shows a coronal angiogram in a normal volunteer. The study was completed in four breath hold s. each averagingjust under 20 sec. Six images were acquired with an echo lim e (TE) of 10 msec and a TR of 150 msec at each of 53 axial slice location s (a to tal of 318 images in 80 sec of imaging). The in-plane reso lution for each slice was 3 mm and the slice thickne ss was 6 mm. yielding a total su perio r-to-infe rior volu me coverage of 3 1.8 cm. Flow compensation (sec below) was not used. At this TR. we found that a 90° flip angle produced good static signal suppression. Four "co ndition ing· · pul ses preceded the acquisition at each location to bring the signal into equilibrium. Figure 3 demo nstrates thi s approach to equilibrium for a single sl ice plane: successive images show improved contrast between blood and statio na ry ti ssue . Because the images are acqu ired in a time sherI compared with the ordinary R-R interval. timeresolved cine angiograms can be produced. Adding the dimension of time-re solution requires sorting the individual images from each slice locat ion according to their latency from the card iac R wave, then calculating separate projection images at each time point. We elected to use a retrospective, rather than a pros pective, cardiac ordering because of its ease of setup. its relative resistance to errors in triggering. and the higher sampling densit y per breat h hold (trig> gered methods wou ld necessaril y introduce pau ses into the imaging expe riment). Figu re 4 shows four frames from a time-resolved angiographic movie of the abdomen acquired in 48 sec of imaging . As flow ve locity drops to zero. a result of cardiac pul sa tion . the signal in the aorta and vena cava is seen to decrease . Th e po rtal veins are see n clearly between the aorta and vena cava and show non pulsatile flow as expected . Suc h time-re solved studie s arc most conve nientl y viewed in cine (movie) format. We have refined the basic angiography method for imaging of the portal ve in . where comparativel y slow flow s are present. and where the flow rate is not a strong function of the cardiac cycle . Twelve axial contiguous slices with 5 mm thi ckness are typically acquired. with in-plane resolution of 3 x 1.5 mm. A TR of 100 msec and a 90° flip angle produce good angiograp hic contrast. Because the flow is nonpulsatile. only four time points are required at each level. Figure 5B shows the axial M1P for a patient who was being evaluated for a liver transplant.
,' .P. Cr. ,wiram formed from eight images ill ea,h of)'J ~Ji,e, . E,L,h of the four imalles in lhb figure rep rese nlS an .mgiogr.lm at a ~uc,essi vely laler ph>l~e in the cOL rd iac cyc te . OLS ind icOLted above ea,h inlOLge . The OLonOL. inferior vena ,ava. and hepalic vei ns show obviou ~ c hanl>e~ in signal intensity during vascul;lr pub;uion. whereas the portOLI vein Irunk shw~ li ul e puls'J! ilily. Slice thicknes s - 6 mm. TR >= I~~ msec. TE '" IJ m ~e"
the readout gradient polarity produces only a small sensitivity with an even echo rephasing between successive lines of data. Model ing of these effects on the signal [39) suggests that the dominant phase shifts and signal losses will occur as a resu lt of the slice selection and phase-encoding gradients, obviating the need for velocity compensation of the read out grad ient. Contrary to what might be expected from conventional MRI. we found that at a TE of 15 msec and with a partial k-space acqui sition, flow compensat ion produced negligible improvement in the angiograms. At least two mechan isms are important in loss of signal from How dephasing: de phasing, and signal cancellation with in a tissue voxe l, and signal cancellation from phase shifts which differ from one line to the next in the MR raw data set. It is notable that the line-to-line ph ase variations are fundamentally minimized with single -shot approaches because a limited time window ex ists during which such vari ations can develop. In light of these data, we preli minarily conclude that in travoxe l dephasi ng represen ts a relatively small component of the motion-related signal losses. at least in normal volunteers. a conclusion supported by other data , as discussed below. Echo Time and Conjllg{l(e Synthesis OUf imager is capable of acquiring two raw data li nes per msec . A 64 x 128 image thu s requires a total
encoding period of 32 msec . If the echo is constrained to fall at the center of k-s pace . a ful l 16 msec must elapse prior to the ec ho ce nter. Th is . coupled with the requirement s fo r slice selection and appropriate pre-encoding. resu lt s in a relatively long minimum TE. for angiograph ic imaging. of 23 msec. Modificat ion s to the method [81 have reduced the ec ho time by tak ing advantage of the conjugate sy nthesi s technique [40). Here. it is poss ible to acqui re just a few raw data lines prior to TE. then to use complex conjugation to calc ulate the missing data lines. Therefore. the conj ugate sy nthesis method ca n be expected to have certai n li mitations in the presence of mot ion . The technique depends upon the ex istence of complex conjugate symmetry expected in a data set for an image with no phase difference s across it [41]. but this condi tio n is necessarily violated in moving sampl es. Acquisition of slight ly more than half of the raw data set enables proper phase correction of low spatial frequency variations in phase. but cannot correct for phase variations in small featu re s. and is thus ineffective in correcting the phase shifts from intravascu lar Hows. Our hypothesis the n. is that some loss in signal or resolution will occur when partial acquisitions are used. With the conjugate sy nthesis. or " partial-k: ' method we were able to achieve minimum echo time s of 10 msec without moment nUlling . and 14 msec with gradient moment compensation. Compari sons were made of acquisitions covering 100% of k·space at a TE of 31 msec (Fig. 6A) and 62.59C
A.P. Cmwle y et "I. ' Single·Shol MRJ
37
ety of time points with respect to cardiac activity and to sum the resulting image s. Sampling of the Cardiac Cycle
Fig. 5. A Axial (top) and coronal (bottom) high speed MRAs of th e abdominal vasculature in a normal volunteer. TR "" 100 msec. TE "" 10 msec. 3 mm slice. four repetit io ns at each slice level. A toml of) I slices were acquired in a single breath hold. B Instascan axial MIP through 12 sl ices obtained in a single breath hold and show ing extensive peritoneal varices and a portal vein occlusion in a patien t evaluated for live r transplant. TR = 100 msec. Flip angle = 90". TE .. 18 ms ec (with tirst order flo w compensation ). Slice thickness 5 mm. in·plane resolution 3 mrn x 3 mm.
of k-space (Fig. 68) at a TE of 10 msec. Note that the partial-k method shows some loss of signal as compared with the full data set. even at much shorte r echo times. The short TE scan. however, does appear to perform better in the larger vessels. Pulsatile Signal Variations
As demonstrated in the time-resolved angiograms shown above, the time-of-flight signal intensity varies considerably throughout the cardiac cycle as a consequence of variations in blood flow velocity. When single static images representing the vascular system are desired, it is necessary to sample a vari-
The signal variations through the cardiac cycle differ for the arterial and venous systems. ranging from the highly pulsatile flow in the descending aorta to the near continuous flow in the portal venous sys· tem. Different strategies shou ld therefore be used to collect data from the various portions of the vascular system. To represent accurately the blood signal in the veins generally requires sampli ng of only a sma ll number of time points; peak signals in the arteries, on the other hand. are achieved only during a small portion of the cardiac cycle , where flow velocities are maximal. Our initial objective in developing an ln slascan angiography sequence was to design an abdominal exam that required, at most, three breath hold s. Although conventional short TR MRA can be used . with each slice location acqu ired during an 8-sec breath hold, the scan requires considerable patient cooperation. Breath-holding after expiration ensures that the position is reasonably reproducible . but slice misregistration is always a potential prob· lem. By using the lnstascan method, we are able to image mUltiple time frames for each slice , with a scan time of approximate ly I sec/sli ce to completely cover the cardiac cycle. Therefore. it is always possible to cove r a reasonable imaging volume within an acceptable breath hold of 10-25 sees. On the other hand. when suc h complete coverage of the cardiac cycle is not required, it is possible to increase the volume coverage appropriate ly while st ill maintaining the overall short acquisition times . Summation: Magnitude Averaging, Complex A veraging , and MIP
The conve ntional MR image is formed as the twodimensional Fourier transform of data co llected line by line during repeated, separate acquisitions. Under such circumstances, phase shifts in the data due to sample variations during the collection of the im· age result in some degree of signal cancellation and loss. In many physiological examination s, such phase shifts arise largely from motion during the exam, and are particularly significant in regions of turbulent fluid flow . As discussed above. the singleshot technique is resistant to such line-by-line variations, as the entire raw data set is acquired over a very short period . When multiple images are acquired through the cardiac cycle, however. such
"
,\ . P. Cr:mie )'':1 "I : SIO.\!Ic-Sh.ll Mltl
t"ig. 6 . Comp; lf i'on uf t hc traJe ·"rr bc["~cn
TE "oJ
),;-~P"Cc
cuva"!;!,, . A
TE • IU m,c..:. (,:,~'fl k·~pace a..:qui,j,iun 18 ) TE - .1 1 m,,:c 11K!'"/ . I.>,p,,~·e "C4ul",;un . The full It· 'pace "ellui,iliun pr,)lj lJce, belh: r sm,,11 \'0:,,0:1 cun 'picuily. t-ut Ihc ,hurler T E ,(;,,0 I!!i,-c\ beHt'r ',gnal in Ihc I"rgcr 11(,,,:1, , Cumpln l ei "nJ m:'tlnilu Je :.I" cr.'g.:' jD) uflhe I! repeilliun, fur each ,licc prooJu(t' " re,uit inferior to lhc M IP of Ihc ,
