MAGNETIC RESONANCE IN MEDICINE 21,238-246 ( 1992)
Spin-Echo M-Mode NMR Imaging TETSUYA MATSUDA,KOJISHIMIZU, * TSUNETARO SAKURAI,~ YAMASAKI, * YUTAKANAGANO,IKUTAROOKADA, SHINJIMIKI. AND CHUICHIKAWAI
KAZUNARI
Third Division, Department oflnternal Medicine, Faculty qf Medicine, Kyoto University, Kyoto, Japan; *Medical Systems Division, Shimadzu Corporution, Kyoto, Japan; and t Department of Biomedical Informatics, Kyoto University Hospital, Kyoto, Japan Received August 6, 1991; revised December 4, 199 1; accepted December 6, 1991 A nuclear magnetic resonance (NMR) imaging and display method for the observation of the continuous motion of objects is presented. By modifying a line scan technique, the spin-densitydistribution along a line is displayed in succession. Although spatial information is limited to only one dimension, the motion of the object is recorded at intervals of 55 ms by using a commercially available NMR imaging system. In a phantom study, this method yielded accurate velocity measurements along a single axis. When the method was applied to the human chest, an image analogous to that of M-mode echocardiography was obtained. This method, which can be called spin-echo M-mode NMR imaging, approaches the functional analysis of cardiac wall motion in regions where echocardiography is not possible. The effects of respiratory motion on the left ventricular wall were recorded in addition to its intrinsic contractile motion in an image obtained along a line parallel to the cranio-caudal axis of the body. The advantages of this method to assess cardiac wall motion in a patient with an arrhythmia were also demonstrated. o 1992Academic press. Inc. INTRODUCTION
The introduction of cine nuclear magnetic resonance (NMR) imaging added a functional analysis to morphological assessment in the diagnosis of cardiovascular disease. Several clinical applications of this method for functional analysis of the heart have been already described ( I , 2). However, in the cine NMR imaging method, images cannot be obtained directly after each single excitation but are created after acquiring all or a part of the K-space data sets. Therefore, continuous observation of moving objects cannot be realized with the conventional cine NMR imaging method, limiting its clinical usefulness in the assessment of cardiac function through consecutive cycles, especially in patients with arrhythmia. The echo-planar imaging technique succeeded in obtaining NMR images in real time ( 3 , 4 ) . The progress in hardware performance and the development of imaging techniques have improved image quality to meet clinical requirements. However, since this technique requires special equipment, it has not been applied to commercially available systems. With respect to the continuous observation of moving objects by means of NMR imaging, the line scan technique has the advantage of being able to obtain spin-density distribution along a line at the acquisition of each signal. Recently, Pearlman et al. ( 5 ) reported on an M-mode NMR imaging method using a two-dimensional spatially 0740-3 194/92 $5.00 Copyright 0 1992 by Academic Press, Inc. AU rights of reproductionin any form reserved
238
SPIN-ECHO M-MODE NMR IMAGING
239
selective pulse ( 6 , 7) and gradient echo acquisition. In contrast, another M-mode NMR imaging method is proposed here, one developed by modifying the simple spinecho, line scan technique reported by Maudsley ( 8 ) . The method can be called spinecho M-mode NMR imaging ( 9 ) .The purpose of this study was to develop this method for the continuous observation of motion with a commercially available NMR imaging system. Assessments of its accuracy in determining motion velocity are performed in a phantom study. Clinical application of this method is also presented. METHODS
The pulse sequence of the spin-echo M-mode NMR imaging method is shown in Fig. 1. A plane is selectively excited by an RF pulse with a flip angle of more than 90" and a gradient magnetic field G,. Then the gradient is switched into Gy, and the 180" RF pulse is irradiated to the plane perpendicular to that of selective excitation. Only the spins within the line of intersection of these two orthogonal planes are refocused and produce spin-echo signal at the echo time (TE). The signal contains the spin-density distribution along the sensitive line in the presence of read-out gradient G,. In conventional line scan techniques, the selected line is advanced laterally at each excitation, generating a two-dimensional image. However, in the present method, location of the selected line is fixed at a region during a series of excitations to allow the continuous observation of moving objects. The spin-density distribution along the selected line, given by the one-dimensional Fourier transform of each acquired signal, is displayed in succession. The resulting image is similar td that of M-mode echocardiography, in which images represent the motion of objects, although the spatial information is limited to only one dimension. Since a certain region is successively excited within a short repetition time, the spoiler gradients are included prior to the excitation pulse. An excitation fhp angle of more than 90" is used to prevent the saturation of spins, so that the spin magnetization is at a low angle after the 180" pulse. The NMR images in this study were obtained on a I .O-T, 1.O-m-bore system (Shimadzu Corporation, Kyoto, Japan), The image matrix was 128 along a selected line
TR
L i
'180-a'
TE
J
180"
spoiler read-out
Gz FIG. 1. Pulse sequence for spin-echo M-mode NMR imaging. By using orthogonal selective excitation and refocusing RF pulses, a line is determined as the intersection of these two planes. A spin-echo signal is sampled with a read-out gradient parallel to the selected line. The sequence repeats with a short TR, while the selected line is fixed in space. The spin-density distribution along the line, given with one-dimensional Fourier transform, is displayed in succession to produce an image analogous to that of M-mode echocardiography.
240
MATSUDA ET AL.
over a length of 36 cm, yielding a matrix of 2.8 mm in size. The cross-sectional area of the selected line was 20 X 20 mm2, and TE was 23 ms. TR and the flip angle were varied in each experiment.
Phantom Study Phantom experiments were conducted to determine whether this method could successfully produce images and correctly represent the motion velocity of objects. Four plastic bottles filled with a nickel chloride solution were placed on a plate. All the bottles were flat on the bottom, had vertical and horizontal cross sections, and were 2.0 cm wide. They were stood parallel to each other with 2.0-cm gaps between each (Fig. 2). The selected line for M-mode NMR imaging was positioned so as to cross the center of each bottle. The plate with accompanying bottles was advanced along the selected line in the magnet at a constant velocity. To examine whether the resulting image changes in response to the TR and motion velocity, M-mode NMR images were obtained with T R s of 55, 100,and 200 ms, while the bottles were stationary or while in motion at a constant velocity of 2.5 or 5.4 cm/s. The resolution of the velocity measurements of this method for a certain TR is determined by the spatial resolution of the image. The combinations of these T R s and velocities yield the motion &stance during each TR to cover a range of from approximately one-half of the pixel length to about four times the pixel length, i.e., from 1.4 to 10.8 mm/TR. In addition, these velocities are within the physiological velocity range of the left ventricular wall motion, which a prior study using M-mode echocardiography showed to be less than 10 cm/s for systole and 13 cm/s for diastole (10).These velocities were experimentally verified by averaging five measurements of distance traversed over a 10-s period. The flip angle was fixed at 150" in this phantom study. Since this method yields an image similar to that of M-mode echocardiography, a stationary object is displayed as a horizontal line and a moving one as an inclining line. The gradient of the line, therefore, represents the component of the velocity in the direction of the sensitive line. In this phantom study, four parallel bands representing the fluid in the bottles appear in the M-mode NMR image (Fig. 3 ) . The gradient of those bands corresponds to the velocity of the phantom.
FIG.2. Layout of phantom and sensitive line. Four rectangular plastic bottles filled with nickel chloride solution are placed on a plate. The sensitive line is positioned so as to cross the center of each bottle. The plate accompanying the bottles is advanced along the line at a constant velocity in the imaging system.
SPIN-ECHO M-MODE NMR IMAGING
24 1
FIG.3. M-mode NMR images of the phantom obtained at velocities of 0, 2.5, and 5.4 cm/s with TRs of 55, 100, and 200 ms. Four bands with high signal intensity in each image correspond to the fluid in the bottles. Note that the temporal scale (i.e., horizontal axis) of the image vanes with TR, so that the gradient of bands is different for each image obtained at the same velocity.
Human Study After the phantom study was completed, the method was applied to produce images of human heart motion in two healthy volunteers. To demonstrate the advantage of this method to record irregular heartbeats, two patients with atrial fibrillation also participated in this study. Informed consent was obtained from all the participants. In this human study, the direction of the sensitive line was set parallel to the craniocaudal axis of the subject, to enable the two typical motions in vivo, the cardiac contraction and respiration, to be observed simultaneously. The location of the sensitive line was set so as to cross the leA ventricle at its center, using a conventional sagittal image obtained with ECG gated spin-echo technique. After that, the M-mode NMR imaging was performed. In the first volunteer, the flip angles were changed from 90" to 170" at 10" increments to optimize the signal-to-noise ratio. An optimized flip angle of 150" was used for the other subjects. In the second volunteer, M-mode NMR images were obtained while he was breathing normally and while holding his breath at the end-inspiratory phase, in order to demonstrate the respiratory motion of both the diaphragm and heart. An ECG was taken during the M-mode NMR imaging in the patients with atrial fibrillation in order to compare the irregularity of heartbeats. With the exception of the second volunteer, a minimum TR of 55 ms to execute the M-mode sequence on our system was used, so that the resulting M-mode images would have the highest temporal resolution. To appreciate the respiratory motion at a lower frequency, a TR of 100 ms was used in the second volunteer.
242
MATSUDA ET AL. RESULTS
Phantom Study Figure 3 shows M-mode NMR images of the phantom, comparing images obtained at velocities of 0, 2.5, and 5.4 cm/s with T R s of 55, 100, and 200 ms. Four parallel bands displaying high signal intensity represent the fluid in the bottles of the phantom. The phantom moved parallel to the sensitive line, which corresponds to the vertical axis of the images. The horizontal axis indicates the time. The scale of this axis vanes with TR, since it determines the temporal resolution of each image. In the images obtained while the bottles were kept stationary, the bands lined up horizontally. Each horizontal band was 2.0 cm in width with 2.0-cm gaps between each other. This is the same as the actual size of the phantom shown in Fig. 2. On the other hand, the bands appear to tilt in the images obtained while the bottles were moving. The gradient became steep as the velocity of the phantom increased or as the TR was prolonged. Table 1 summarizes the velocity of the phantom determined by the M-mode MR images. Those measurements coincided with the actual velocities.
Human Study Figure 4 is an ECG gated sagittal image of the first volunteer indicating the typical location of the sensitive line for the human study. A comparison of the M-mode NMR images for various flip angles from 90" to 170" is shown in Fig. 5. In each image, the left ventricular wall motion is demonstrated as a waveform analogous to that seen in M-mode echocardiography. In addition to the left ventricular contraction, the respiratory motion of the heart induced by the diaphragm is superimposed on the waveform. Those patterns are delineated with clear contrast in the images obtained with a flip angle of 150" or 160". Figure 6 acquired from the second volunteer contrasts the M-mode NMR image obtained during normal breathing with that obtained while he was holding his breath at the end-inspiratory phase. The left ventricular free wall and the infenor wall appear above the diaphragm, since the sensitive line of these images is parallel to the craniocaudal axis. In Fig. 6a, the left ventricular motion induced by respiration is seen more comprehensively than in Fig. 5 , because a TR of 100 ms was appropriate for the observation of relatively low-frequency motion. Among the two frequency components
TABLE 1 Motion Velocity of Phantom Measured by Means of M-Mode NMR Imaging
Measured velocity from M-mode NMR image(cm/s) Actual velocity (cm/s) 0 2.50
5.36
TR
=
55
0 2.70 5.67
T R = 100
TR=200
0 2.52 5.33
0 2.57 5.41
SPIN-ECHO M-MODE NMR IMAGING
243
FIG.4. !hgittal image of the first volunteer. The vertical bar across the center of the left ventricle indicates the location of the sensitive line for M-mode images in Fig. 5. A location similar to this was used for other human subjects.
which form the left ventricular motion wave in Fig. 6a, the amplitude of the lowfrequency component is larger than that of the high-frequency component. This represents that in the cranio-caudal direction the left ventricular wall moves farther by respiration than by cardiac contraction, even during normal breathing. In contrast, all of the structures except for the left ventricular free wall and inferior wail appear to form straight lines in Fig. 6b. Thus, the left ventricular wall motion recorded in this image consists of the cardiac contraction without the effects of respiration.
FIG. 5. Comparison of M-mode NMR images obtained with flip angles from 90" to 170". Contractile motion of the left ventricular free wall and the inferior wall is demonstrated with clear contrast in the images obtained with flip angles of 150" and 160'
244
MATSUDA ET AL.
FIG.6. M-mode NMR images of the second volunteer showing respiratory motion. (a) Normal breathing. Low-frequency motion induced by respiration is superimposed on the higher-frequency motion of cardiac contraction. The large and small white arrows show the left ventricular free wall and inferior wall, respectively. The open arrow indicates the diaphragm. Note that the motion amplitude caused by respiration is larger than that by cardiac contraction, even during normal breathing. (b) Holding his breath at end-inspiratory phase. The cardiac contractile motion of left ventricle without the effects of respiration is demonstrated. FIG.7. M-mode NMR image of a patient with atrial fibrillation and simultaneous recording of ECG. The irregularity of the cardiac cycle shown in the image corresponds well with ECG.
The M-mode NMR image obtained from one of the patients with atrial fibrillation and a simultaneous ECG recording are shown in Fig. 7. The irregularity of cardiac cycle is accurately detected by the M-mode NMR image. DISCUSSION
From the signal sampled in the presence of a read-out gradient, the NMR imaging method is ready to display the spin-density distribution along an axis using a onedimensional Fourier transform. To exploit this property of NMR imaging, spin-echo M-mode NMR images were produced by modifying the line scan technique. The images obtained from the phantom and from the human subjects allowed continuous observation of moving objects using a standard NMR imaging system while limiting spatial information to only one dimension. In the phantom study, the velocity determined by using the M-mode NMR image demonstrated excellent agreement with the actual velocity, showing this method to be one suitable for measuring the component of the velocity along the direction of the sensitive line. The human study demonstrated the effectiveness of this method in the assessment of cardiac wall motion even in patients with arrythmia. In addition to showing cardiac contraction, the M-mode NMR image yielded a clear representation of cardiac motion induced by respiration. This result suggests that the assessment of respiration on cardiac function is one of the possible applications of this method, since echocardiography does not provide such a fine demonstration of respiratory motion. Since this M-mode NMR imaging method is based on the line scan technique, it does not require the two-dimensional signal sampling in the K-space. Thus, unlike a
SPIN-ECHO M-MODE NMR IMAGING
245
conventional NMR imaging method such as cine NMR imaging, this method allows continuous observation of motion in real time. An introduction of the cine NMR imaging technique ( I , 2) has added to the potential of NMR imaging in the diagnosis of cardiovascular disease. However, since conventional cine NMR imaging generates sequential images, each consisting of signals sampled over hundreds of heartbeats, limitations exist in the evaluation of definite motion such as beat-to-beat difference. It is particularly difficult to assess cardiac wall motion in patients with arrhythmia using the cine NMR imaging method. Another potential limitation of cine NMR imaging using the standard ECG gating technique is that it misses the rapid ventricular filling phase, since it does not acquire the signal immediately before the triggering R wave. An interesting result of this study is that even normal breathing induces a considerable amount of motion in the heart. Possible error may consequently occur when cardiac wall motion is analyzed using cine NMR images in which the respiratory motion is averaged out. The M-mode NMR imaging method is free from these problems. In addition, those images demonstrating the respiratory motion of the heart were made possible by applying this method to a parallel line to the long axis of the body. This result is not achieved by M-mode echocardiography, since its approaches are restricted to the acoustic window. The M-mode NMR imaging method presented here holds accessibility of NMR imaging to any location and direction. Another potential merit of this method arises from its capability of interactive display after each signal acquisition. The location of the line on which an image is to be made can be adjusted without delay in response to the preceding results. To obtain the accurate size of the left ventricle or its wall thickness, the measurement should be made in the appropriate direction. This method allows determination of an ideal direction for measurement by trial and error. The relationship between the M-mode NMR imaging method and a snapshot, twodimensional NMR imaging method is similar to the complementary relationship between M-mode and two-dimensional echocardiography. The echocardiography is currently one of the most common methods for noninvasive cardiac imaging. Two-dimensional echocardiography is useful for determining the shape of the object being examined, while M-mode echocardiography gives an excellent evaluation of the amount of motion in the axial direction ( 1 1 ) . Both complement each other to provide a complete echocardiographic examination. In the same way, the M-mode NMR imaging method can be used to do cardiac NMR imaging with a two-dimensional imaging method, such as the echo-planar technique. Two major limitations of this method at present are the temporal resolution and the signal-to-noise ratio. The highest repetition rate of approximately 18 times per second is yielded from the shortest TR of 55 ms to execute the pulse sequence of this method on the current system. This repetition rate is much less than that of M-mode echocardiography, the typical rate of which is over 1000 per second ( 1 1 ) . Although this method is insufficient to be employed for the analysis of high-frequency motion, the result of the human study showed that this method could present the cardiac contractile motion as well as the respiratory-induced motion. The signal-to-noiseratio of a line technique is essentially inferior to that of a planar technique. In addition, the short TR used in this method makes the situation worse. In the spin-echotechnique, a high flip angle was employed to prevent the spin magnetization from becoming
246
MATSUDA ET AL.
saturated after a 180” pulse. However, even this strategy failed to produce a sufficient signal when the cross-sectional area of the sensitive line was less than 20 X 20 mm2. A simple possible solution for both of these limitations is to employ a shorter TE. This would allow for the reduction of TR and yield an improved temporal resolution. A shorter TE is also effective in producing a higher signal. Since myocardium has a rather short T2 among human tissues, a shorter TE can result in a remarkable improvement in cardiac imaging. Another M-mode NMR imaging method reported by Pearlman d al. ( 5 ) employs a gradient echo signal acquisition. A two-dimensional spatially selective pulse (6, 7) allows gradient echo signal acquisition from a cylindric region, thus achieving a shorter TE and TR. This method produces white blood Mmode images, whereas the spin-echo M-mode NMR imaging method generates black blood images. Since cardiac valves produce little signal in either method, the gradient echo method is potentially suited to the assessment of valvular motion. However, the spin-echo M-mode NMR imaging method yields higher contrast when myocardium and blood pool are compared. In conclusion, the spin-echo M-mode NMR imaging method reported here allowed continuous observation of cardiac wall motion, although spatial information was limited to only one dimension. An M-mode image was obtained in the direction in which echocardiographyis not applied. Although the temporal resolution is poorer than that of M-mode echocardiography, this method demonstrated the continuous motion of the cardiac wall. The M-mode NMR imaging has the potential of complementing the two-dimensional high-speed imaging technique of NMR as well as the ultrasound examination. REFERENCES 1. C. B. HIGGINS, W. HOLT,P. PFLUGFELDER, AND U . SECHTEM, Mugn. Reson. Med. 6, 121 ( 1988). 2. J . A. UTZ, R. J. HERFKENS, J . A. HEINSIMER, T. BASHORE,R. CALIF, G. GLOVER,N. PELC,AND A. SHIMAKAWA, Amer. J. Roentgenol. 148,839 ( 1987). 3. P. MANSFIELD AND I. L. PYKETT, J. Magn. Reson. 29, 355 (1978). 4. M. DOYLE,B. CHAPMAN,R. TURNER, R. J. ORDILIGE,M.CAWLEY,R. COXON, P. GLOVER,R. E. COUPLARD,G . K. MORRIS, B. s.WORTHINGTON, AND P.MANSFIELD,k n e e l , 682 (&pt. 20, 1986). 5. J. D. PEARLMAN, C. J. HARDY,AND H. E. CLINE,Radiology 175, 369 ( 1990). 6. C. J. HARDY,P. A. BOTTOMLEY, M. ODONNELL,AND P.ROEMER,J. Mugn. Reson. 77,233 ( 1988). 7. J . PAULY,D. NISHIMURA, AND A. MACOVSKI, J. Mugn. Reson. 81,43 ( 1989). 8. A. A. MAUDSLEY,J. Mugn. Reson. 41, 112 (1980). 9. T. MATSUDA,K. SHIMIZU, T. SAKURAI, Y. NAGANO,K. YAMAZAKI, AND c. KAWAI,“Proceedings, 7th Annual Meeting of the Society of Magnetic Resonance in Medicine, 1988,” Vol. I , p. 202. 10. A. M. FOGELMAN,A. S. ABSASI,M.L. PEARCE,AND A. A. KATTUS, Circulalion 46,905 ( 1972). I I . H. FEIGENBAUM, “Echocardiography,” Lea & Febiger, Philadelphia, 1986.
Spin-Echo M-Mode NMR Imaging TETSUYA MATSUDA,KOJISHIMIZU, * TSUNETARO SAKURAI,~ YAMASAKI, * YUTAKANAGANO,IKUTAROOKADA, SHINJIMIKI. AND CHUICHIKAWAI
KAZUNARI
Third Division, Department oflnternal Medicine, Faculty qf Medicine, Kyoto University, Kyoto, Japan; *Medical Systems Division, Shimadzu Corporution, Kyoto, Japan; and t Department of Biomedical Informatics, Kyoto University Hospital, Kyoto, Japan Received August 6, 1991; revised December 4, 199 1; accepted December 6, 1991 A nuclear magnetic resonance (NMR) imaging and display method for the observation of the continuous motion of objects is presented. By modifying a line scan technique, the spin-densitydistribution along a line is displayed in succession. Although spatial information is limited to only one dimension, the motion of the object is recorded at intervals of 55 ms by using a commercially available NMR imaging system. In a phantom study, this method yielded accurate velocity measurements along a single axis. When the method was applied to the human chest, an image analogous to that of M-mode echocardiography was obtained. This method, which can be called spin-echo M-mode NMR imaging, approaches the functional analysis of cardiac wall motion in regions where echocardiography is not possible. The effects of respiratory motion on the left ventricular wall were recorded in addition to its intrinsic contractile motion in an image obtained along a line parallel to the cranio-caudal axis of the body. The advantages of this method to assess cardiac wall motion in a patient with an arrhythmia were also demonstrated. o 1992Academic press. Inc. INTRODUCTION
The introduction of cine nuclear magnetic resonance (NMR) imaging added a functional analysis to morphological assessment in the diagnosis of cardiovascular disease. Several clinical applications of this method for functional analysis of the heart have been already described ( I , 2). However, in the cine NMR imaging method, images cannot be obtained directly after each single excitation but are created after acquiring all or a part of the K-space data sets. Therefore, continuous observation of moving objects cannot be realized with the conventional cine NMR imaging method, limiting its clinical usefulness in the assessment of cardiac function through consecutive cycles, especially in patients with arrhythmia. The echo-planar imaging technique succeeded in obtaining NMR images in real time ( 3 , 4 ) . The progress in hardware performance and the development of imaging techniques have improved image quality to meet clinical requirements. However, since this technique requires special equipment, it has not been applied to commercially available systems. With respect to the continuous observation of moving objects by means of NMR imaging, the line scan technique has the advantage of being able to obtain spin-density distribution along a line at the acquisition of each signal. Recently, Pearlman et al. ( 5 ) reported on an M-mode NMR imaging method using a two-dimensional spatially 0740-3 194/92 $5.00 Copyright 0 1992 by Academic Press, Inc. AU rights of reproductionin any form reserved
238
SPIN-ECHO M-MODE NMR IMAGING
239
selective pulse ( 6 , 7) and gradient echo acquisition. In contrast, another M-mode NMR imaging method is proposed here, one developed by modifying the simple spinecho, line scan technique reported by Maudsley ( 8 ) . The method can be called spinecho M-mode NMR imaging ( 9 ) .The purpose of this study was to develop this method for the continuous observation of motion with a commercially available NMR imaging system. Assessments of its accuracy in determining motion velocity are performed in a phantom study. Clinical application of this method is also presented. METHODS
The pulse sequence of the spin-echo M-mode NMR imaging method is shown in Fig. 1. A plane is selectively excited by an RF pulse with a flip angle of more than 90" and a gradient magnetic field G,. Then the gradient is switched into Gy, and the 180" RF pulse is irradiated to the plane perpendicular to that of selective excitation. Only the spins within the line of intersection of these two orthogonal planes are refocused and produce spin-echo signal at the echo time (TE). The signal contains the spin-density distribution along the sensitive line in the presence of read-out gradient G,. In conventional line scan techniques, the selected line is advanced laterally at each excitation, generating a two-dimensional image. However, in the present method, location of the selected line is fixed at a region during a series of excitations to allow the continuous observation of moving objects. The spin-density distribution along the selected line, given by the one-dimensional Fourier transform of each acquired signal, is displayed in succession. The resulting image is similar td that of M-mode echocardiography, in which images represent the motion of objects, although the spatial information is limited to only one dimension. Since a certain region is successively excited within a short repetition time, the spoiler gradients are included prior to the excitation pulse. An excitation fhp angle of more than 90" is used to prevent the saturation of spins, so that the spin magnetization is at a low angle after the 180" pulse. The NMR images in this study were obtained on a I .O-T, 1.O-m-bore system (Shimadzu Corporation, Kyoto, Japan), The image matrix was 128 along a selected line
TR
L i
'180-a'
TE
J
180"
spoiler read-out
Gz FIG. 1. Pulse sequence for spin-echo M-mode NMR imaging. By using orthogonal selective excitation and refocusing RF pulses, a line is determined as the intersection of these two planes. A spin-echo signal is sampled with a read-out gradient parallel to the selected line. The sequence repeats with a short TR, while the selected line is fixed in space. The spin-density distribution along the line, given with one-dimensional Fourier transform, is displayed in succession to produce an image analogous to that of M-mode echocardiography.
240
MATSUDA ET AL.
over a length of 36 cm, yielding a matrix of 2.8 mm in size. The cross-sectional area of the selected line was 20 X 20 mm2, and TE was 23 ms. TR and the flip angle were varied in each experiment.
Phantom Study Phantom experiments were conducted to determine whether this method could successfully produce images and correctly represent the motion velocity of objects. Four plastic bottles filled with a nickel chloride solution were placed on a plate. All the bottles were flat on the bottom, had vertical and horizontal cross sections, and were 2.0 cm wide. They were stood parallel to each other with 2.0-cm gaps between each (Fig. 2). The selected line for M-mode NMR imaging was positioned so as to cross the center of each bottle. The plate with accompanying bottles was advanced along the selected line in the magnet at a constant velocity. To examine whether the resulting image changes in response to the TR and motion velocity, M-mode NMR images were obtained with T R s of 55, 100,and 200 ms, while the bottles were stationary or while in motion at a constant velocity of 2.5 or 5.4 cm/s. The resolution of the velocity measurements of this method for a certain TR is determined by the spatial resolution of the image. The combinations of these T R s and velocities yield the motion &stance during each TR to cover a range of from approximately one-half of the pixel length to about four times the pixel length, i.e., from 1.4 to 10.8 mm/TR. In addition, these velocities are within the physiological velocity range of the left ventricular wall motion, which a prior study using M-mode echocardiography showed to be less than 10 cm/s for systole and 13 cm/s for diastole (10).These velocities were experimentally verified by averaging five measurements of distance traversed over a 10-s period. The flip angle was fixed at 150" in this phantom study. Since this method yields an image similar to that of M-mode echocardiography, a stationary object is displayed as a horizontal line and a moving one as an inclining line. The gradient of the line, therefore, represents the component of the velocity in the direction of the sensitive line. In this phantom study, four parallel bands representing the fluid in the bottles appear in the M-mode NMR image (Fig. 3 ) . The gradient of those bands corresponds to the velocity of the phantom.
FIG.2. Layout of phantom and sensitive line. Four rectangular plastic bottles filled with nickel chloride solution are placed on a plate. The sensitive line is positioned so as to cross the center of each bottle. The plate accompanying the bottles is advanced along the line at a constant velocity in the imaging system.
SPIN-ECHO M-MODE NMR IMAGING
24 1
FIG.3. M-mode NMR images of the phantom obtained at velocities of 0, 2.5, and 5.4 cm/s with TRs of 55, 100, and 200 ms. Four bands with high signal intensity in each image correspond to the fluid in the bottles. Note that the temporal scale (i.e., horizontal axis) of the image vanes with TR, so that the gradient of bands is different for each image obtained at the same velocity.
Human Study After the phantom study was completed, the method was applied to produce images of human heart motion in two healthy volunteers. To demonstrate the advantage of this method to record irregular heartbeats, two patients with atrial fibrillation also participated in this study. Informed consent was obtained from all the participants. In this human study, the direction of the sensitive line was set parallel to the craniocaudal axis of the subject, to enable the two typical motions in vivo, the cardiac contraction and respiration, to be observed simultaneously. The location of the sensitive line was set so as to cross the leA ventricle at its center, using a conventional sagittal image obtained with ECG gated spin-echo technique. After that, the M-mode NMR imaging was performed. In the first volunteer, the flip angles were changed from 90" to 170" at 10" increments to optimize the signal-to-noise ratio. An optimized flip angle of 150" was used for the other subjects. In the second volunteer, M-mode NMR images were obtained while he was breathing normally and while holding his breath at the end-inspiratory phase, in order to demonstrate the respiratory motion of both the diaphragm and heart. An ECG was taken during the M-mode NMR imaging in the patients with atrial fibrillation in order to compare the irregularity of heartbeats. With the exception of the second volunteer, a minimum TR of 55 ms to execute the M-mode sequence on our system was used, so that the resulting M-mode images would have the highest temporal resolution. To appreciate the respiratory motion at a lower frequency, a TR of 100 ms was used in the second volunteer.
242
MATSUDA ET AL. RESULTS
Phantom Study Figure 3 shows M-mode NMR images of the phantom, comparing images obtained at velocities of 0, 2.5, and 5.4 cm/s with T R s of 55, 100, and 200 ms. Four parallel bands displaying high signal intensity represent the fluid in the bottles of the phantom. The phantom moved parallel to the sensitive line, which corresponds to the vertical axis of the images. The horizontal axis indicates the time. The scale of this axis vanes with TR, since it determines the temporal resolution of each image. In the images obtained while the bottles were kept stationary, the bands lined up horizontally. Each horizontal band was 2.0 cm in width with 2.0-cm gaps between each other. This is the same as the actual size of the phantom shown in Fig. 2. On the other hand, the bands appear to tilt in the images obtained while the bottles were moving. The gradient became steep as the velocity of the phantom increased or as the TR was prolonged. Table 1 summarizes the velocity of the phantom determined by the M-mode MR images. Those measurements coincided with the actual velocities.
Human Study Figure 4 is an ECG gated sagittal image of the first volunteer indicating the typical location of the sensitive line for the human study. A comparison of the M-mode NMR images for various flip angles from 90" to 170" is shown in Fig. 5. In each image, the left ventricular wall motion is demonstrated as a waveform analogous to that seen in M-mode echocardiography. In addition to the left ventricular contraction, the respiratory motion of the heart induced by the diaphragm is superimposed on the waveform. Those patterns are delineated with clear contrast in the images obtained with a flip angle of 150" or 160". Figure 6 acquired from the second volunteer contrasts the M-mode NMR image obtained during normal breathing with that obtained while he was holding his breath at the end-inspiratory phase. The left ventricular free wall and the infenor wall appear above the diaphragm, since the sensitive line of these images is parallel to the craniocaudal axis. In Fig. 6a, the left ventricular motion induced by respiration is seen more comprehensively than in Fig. 5 , because a TR of 100 ms was appropriate for the observation of relatively low-frequency motion. Among the two frequency components
TABLE 1 Motion Velocity of Phantom Measured by Means of M-Mode NMR Imaging
Measured velocity from M-mode NMR image(cm/s) Actual velocity (cm/s) 0 2.50
5.36
TR
=
55
0 2.70 5.67
T R = 100
TR=200
0 2.52 5.33
0 2.57 5.41
SPIN-ECHO M-MODE NMR IMAGING
243
FIG.4. !hgittal image of the first volunteer. The vertical bar across the center of the left ventricle indicates the location of the sensitive line for M-mode images in Fig. 5. A location similar to this was used for other human subjects.
which form the left ventricular motion wave in Fig. 6a, the amplitude of the lowfrequency component is larger than that of the high-frequency component. This represents that in the cranio-caudal direction the left ventricular wall moves farther by respiration than by cardiac contraction, even during normal breathing. In contrast, all of the structures except for the left ventricular free wall and inferior wail appear to form straight lines in Fig. 6b. Thus, the left ventricular wall motion recorded in this image consists of the cardiac contraction without the effects of respiration.
FIG. 5. Comparison of M-mode NMR images obtained with flip angles from 90" to 170". Contractile motion of the left ventricular free wall and the inferior wall is demonstrated with clear contrast in the images obtained with flip angles of 150" and 160'
244
MATSUDA ET AL.
FIG.6. M-mode NMR images of the second volunteer showing respiratory motion. (a) Normal breathing. Low-frequency motion induced by respiration is superimposed on the higher-frequency motion of cardiac contraction. The large and small white arrows show the left ventricular free wall and inferior wall, respectively. The open arrow indicates the diaphragm. Note that the motion amplitude caused by respiration is larger than that by cardiac contraction, even during normal breathing. (b) Holding his breath at end-inspiratory phase. The cardiac contractile motion of left ventricle without the effects of respiration is demonstrated. FIG.7. M-mode NMR image of a patient with atrial fibrillation and simultaneous recording of ECG. The irregularity of the cardiac cycle shown in the image corresponds well with ECG.
The M-mode NMR image obtained from one of the patients with atrial fibrillation and a simultaneous ECG recording are shown in Fig. 7. The irregularity of cardiac cycle is accurately detected by the M-mode NMR image. DISCUSSION
From the signal sampled in the presence of a read-out gradient, the NMR imaging method is ready to display the spin-density distribution along an axis using a onedimensional Fourier transform. To exploit this property of NMR imaging, spin-echo M-mode NMR images were produced by modifying the line scan technique. The images obtained from the phantom and from the human subjects allowed continuous observation of moving objects using a standard NMR imaging system while limiting spatial information to only one dimension. In the phantom study, the velocity determined by using the M-mode NMR image demonstrated excellent agreement with the actual velocity, showing this method to be one suitable for measuring the component of the velocity along the direction of the sensitive line. The human study demonstrated the effectiveness of this method in the assessment of cardiac wall motion even in patients with arrythmia. In addition to showing cardiac contraction, the M-mode NMR image yielded a clear representation of cardiac motion induced by respiration. This result suggests that the assessment of respiration on cardiac function is one of the possible applications of this method, since echocardiography does not provide such a fine demonstration of respiratory motion. Since this M-mode NMR imaging method is based on the line scan technique, it does not require the two-dimensional signal sampling in the K-space. Thus, unlike a
SPIN-ECHO M-MODE NMR IMAGING
245
conventional NMR imaging method such as cine NMR imaging, this method allows continuous observation of motion in real time. An introduction of the cine NMR imaging technique ( I , 2) has added to the potential of NMR imaging in the diagnosis of cardiovascular disease. However, since conventional cine NMR imaging generates sequential images, each consisting of signals sampled over hundreds of heartbeats, limitations exist in the evaluation of definite motion such as beat-to-beat difference. It is particularly difficult to assess cardiac wall motion in patients with arrhythmia using the cine NMR imaging method. Another potential limitation of cine NMR imaging using the standard ECG gating technique is that it misses the rapid ventricular filling phase, since it does not acquire the signal immediately before the triggering R wave. An interesting result of this study is that even normal breathing induces a considerable amount of motion in the heart. Possible error may consequently occur when cardiac wall motion is analyzed using cine NMR images in which the respiratory motion is averaged out. The M-mode NMR imaging method is free from these problems. In addition, those images demonstrating the respiratory motion of the heart were made possible by applying this method to a parallel line to the long axis of the body. This result is not achieved by M-mode echocardiography, since its approaches are restricted to the acoustic window. The M-mode NMR imaging method presented here holds accessibility of NMR imaging to any location and direction. Another potential merit of this method arises from its capability of interactive display after each signal acquisition. The location of the line on which an image is to be made can be adjusted without delay in response to the preceding results. To obtain the accurate size of the left ventricle or its wall thickness, the measurement should be made in the appropriate direction. This method allows determination of an ideal direction for measurement by trial and error. The relationship between the M-mode NMR imaging method and a snapshot, twodimensional NMR imaging method is similar to the complementary relationship between M-mode and two-dimensional echocardiography. The echocardiography is currently one of the most common methods for noninvasive cardiac imaging. Two-dimensional echocardiography is useful for determining the shape of the object being examined, while M-mode echocardiography gives an excellent evaluation of the amount of motion in the axial direction ( 1 1 ) . Both complement each other to provide a complete echocardiographic examination. In the same way, the M-mode NMR imaging method can be used to do cardiac NMR imaging with a two-dimensional imaging method, such as the echo-planar technique. Two major limitations of this method at present are the temporal resolution and the signal-to-noise ratio. The highest repetition rate of approximately 18 times per second is yielded from the shortest TR of 55 ms to execute the pulse sequence of this method on the current system. This repetition rate is much less than that of M-mode echocardiography, the typical rate of which is over 1000 per second ( 1 1 ) . Although this method is insufficient to be employed for the analysis of high-frequency motion, the result of the human study showed that this method could present the cardiac contractile motion as well as the respiratory-induced motion. The signal-to-noiseratio of a line technique is essentially inferior to that of a planar technique. In addition, the short TR used in this method makes the situation worse. In the spin-echotechnique, a high flip angle was employed to prevent the spin magnetization from becoming
246
MATSUDA ET AL.
saturated after a 180” pulse. However, even this strategy failed to produce a sufficient signal when the cross-sectional area of the sensitive line was less than 20 X 20 mm2. A simple possible solution for both of these limitations is to employ a shorter TE. This would allow for the reduction of TR and yield an improved temporal resolution. A shorter TE is also effective in producing a higher signal. Since myocardium has a rather short T2 among human tissues, a shorter TE can result in a remarkable improvement in cardiac imaging. Another M-mode NMR imaging method reported by Pearlman d al. ( 5 ) employs a gradient echo signal acquisition. A two-dimensional spatially selective pulse (6, 7) allows gradient echo signal acquisition from a cylindric region, thus achieving a shorter TE and TR. This method produces white blood Mmode images, whereas the spin-echo M-mode NMR imaging method generates black blood images. Since cardiac valves produce little signal in either method, the gradient echo method is potentially suited to the assessment of valvular motion. However, the spin-echo M-mode NMR imaging method yields higher contrast when myocardium and blood pool are compared. In conclusion, the spin-echo M-mode NMR imaging method reported here allowed continuous observation of cardiac wall motion, although spatial information was limited to only one dimension. An M-mode image was obtained in the direction in which echocardiographyis not applied. Although the temporal resolution is poorer than that of M-mode echocardiography, this method demonstrated the continuous motion of the cardiac wall. The M-mode NMR imaging has the potential of complementing the two-dimensional high-speed imaging technique of NMR as well as the ultrasound examination. REFERENCES 1. C. B. HIGGINS, W. HOLT,P. PFLUGFELDER, AND U . SECHTEM, Mugn. Reson. Med. 6, 121 ( 1988). 2. J . A. UTZ, R. J. HERFKENS, J . A. HEINSIMER, T. BASHORE,R. CALIF, G. GLOVER,N. PELC,AND A. SHIMAKAWA, Amer. J. Roentgenol. 148,839 ( 1987). 3. P. MANSFIELD AND I. L. PYKETT, J. Magn. Reson. 29, 355 (1978). 4. M. DOYLE,B. CHAPMAN,R. TURNER, R. J. ORDILIGE,M.CAWLEY,R. COXON, P. GLOVER,R. E. COUPLARD,G . K. MORRIS, B. s.WORTHINGTON, AND P.MANSFIELD,k n e e l , 682 (&pt. 20, 1986). 5. J. D. PEARLMAN, C. J. HARDY,AND H. E. CLINE,Radiology 175, 369 ( 1990). 6. C. J. HARDY,P. A. BOTTOMLEY, M. ODONNELL,AND P.ROEMER,J. Mugn. Reson. 77,233 ( 1988). 7. J . PAULY,D. NISHIMURA, AND A. MACOVSKI, J. Mugn. Reson. 81,43 ( 1989). 8. A. A. MAUDSLEY,J. Mugn. Reson. 41, 112 (1980). 9. T. MATSUDA,K. SHIMIZU, T. SAKURAI, Y. NAGANO,K. YAMAZAKI, AND c. KAWAI,“Proceedings, 7th Annual Meeting of the Society of Magnetic Resonance in Medicine, 1988,” Vol. I , p. 202. 10. A. M. FOGELMAN,A. S. ABSASI,M.L. PEARCE,AND A. A. KATTUS, Circulalion 46,905 ( 1972). I I . H. FEIGENBAUM, “Echocardiography,” Lea & Febiger, Philadelphia, 1986.
