Resonance
Imaging, Vol. 10. Pp. l-6, FTinted in the USA. All rights reserved.
Magnetic
1992 Copyright
0730-725X/92 SS.00 + .@I 0 1992 PerSamon F’ms plc
0 Original Contribution
FAST INVERSION RECOVERY Tl CONTRAST AND CHEMICAL SHIFT CONTRAST IN HIGH-RESOLUTION SNAPSHOT FLASH MR IMAGES DIETER MATTHAEI, * AXEL HAASE,~DIETMAR HENRICH,* AND ECKHART DUHMKE* *Klinik Kir Strahlentherapie, 3400 Gottingen, Robert-Koch-Strasse40, tUniversitlt Wurzburg, 8700 Wiirzburg, PhysikalischesInstitut, Am Hubland, and SBRUKBRMedizintechnik,7500 Karlsruhe, Germany Fast MR imaging attracts the interest of both clinicians and physicists because new diagnostic information arises with reduced artifacts due to short investigational times. With the acceleration of the Snapshot FLASH MR sequence, the measurement of high-resolution images with 256 x 256 matrix is reported, together with contrasting prepulses that are applied to attain contrast in combination with higher in-plane resolution. Measuring times are in the range of a second. For whole-body imaging, a TR = 5.2 msec and a TE = 2.6 msec could be attained measuring omit 256 x 256 matrix images. Artifact-free images demonstrating T, contrast and contrast from chemical shift are performed on moving organs (heart, intestine) in different experiments. These applications can easily be performed in a couple of minutes for clinical use. Especially in the lung, short TE and high resolution result in a new imaging quality of pulmonary and mediastinal vessels. Keywords: Magnetic resonance (MR) physics; Magnetic resonance (MR) pulse sequences; Magnetic resonance (MR) contrast enhancement; Magnetic resonance (MR) fast imaging; Magnetic resonance (MR) cardiac imaging; Magnetic resonance (MR) pulmonary imaging; Magnetic resonance (MR) abdominal imaging.
INTRODUCTION
More than 10 years ago, Peter Mansfield’ proposed the ECHO PLANAR sequence, which allowed MR imaging investigations in fractions of a second with reduced in-plane resolution. Recently, higher spatial resolution for ECHO PLANAR imaging is reported* which obviously makes the sequence gradually longer. The FLASH MR sequence was proposed in 19853, that uses a partial flip angle for excitation and a gradient echo for the detection of the signal. This idea began with somewhat slow images and is now being developed to become faster and more highly resolved. At the beginning, this sequence allowed the measurement of images in a few seconds on conventional imaging hardware. Dynamic delineation of vascular and parenchyma1 contrast after Gd-DTPA4 could be shown with this sequence, as well as three-dimensional MR imaging in some minutes for the first time. With reduced spatial resolution and fast gradient RECEIVED 2128191; ACCEDED6128191. Address reprint requests to Dieter Matthaei, Klinik fiir
switchings, a variation of this sequence, dubbed Snapshot FLASH, was proposed last year for the fast imaging of dynamic subsecond processes5 . Rapid cardiac and vascular clinical applications were introduced6 employing FLASH sequences. Until now, the further acceleration of the FLASH MR sequence has allowed measurement with a TR in the range of 5 msec for 256 x 256 matrices, thus enabling high-resolution images in about a second. Comparable in-plane resolutions for head imaging, however, were demonstrated in around a second.7 With high-resolution FLASH MR images, the delineation of many anatomical details is feasible, which has already been shown for slower versions of the method. It was, however, questionable, with former timings and sensitivity of the electronics, whether these images would be able to display contrast due to T, relaxation and due to chemical shift selective imaging, which is performed by presaturation of the unwanted population (fat or water). Strahlentherapie, Gemany.
3400 Gbttingen, Robert-Koch-Strasse
40,
2
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
Fig. 1. Two Snapshot FLASH MR sequences are demonstrated with preparing pulses before them. (a) A 180” pulse is placed before the imaging sequence with different delays. Here for the images in Fig. 3 a lOO-, 200-, and 300-msec delay were used. (b) For the saturation of one line in the proton spectrum, a selective 90” pulse was placed before the imaging sequence. A lo-msec spoiler delay was used additionally to dephase the saturated spins.
spins in the whole object. As shown in Fig. l(a), a delay of 100,200, and 300 msec TI between the pulse and the sequence was chosen here to demonstrate that the evolution of the contrast becomes visible. Additional experiments were done to demonstrate the capability of the Snapshot FLASH MR sequence of delineating contrast in the images due to selective saturation of one proton species. Here selective 90°C pulses were used before the sequence to saturate either the water or the fat protons as shown in Fig. l(b). To dephase the saturated spins, an additional gradient of 10 msec duration was applied before imaging.
RESULTS High-resolution Snapshot FLASH MR images with a T, and chemical shift contrast were done with coronal
MATERIALS AND METHODS Here high-resolution Snapshot FLASH MR wholebody images are reported, which on a commercial imaging machine, with only a reduced gradient diameter (60 cm), are measured with a TR of 5.2 msec. Thus a 256 X 256 matrix is measured in about 1.4 sec. Most images were measured with single excitation. For these experiments, a 2-T whole-body MR imaging machine (BRUKER Medizintechnik, Karlsruhe) was used in combination with gradients of about 12 mT/m. The diameter of the gradient system was 60 cm to reduce the artifacts and allow lower gradient switching times. The investigations were done on healthy volunteers after careful physical investigations. Informed consent was obtained before each imaging procedure. The above images were obtained in the abdomen and thorax to show the measurement of the beating heart without triggering and the intraabdominal organs without artifacts form peristalsis. On the one hand, the demonstration of high-resolution Snapshot FLASH MR images in the human trunk here was planned to demonstrate the in-plane resolution without artifacts. On the other hand it was intended to show whether pulses before the imaging experimentaffecting the spins in the imaging plane- would keep their contrasting effect with the imaging speed attained here. Two contrasting experiments were thus done to evaluate the capability of the high-resolution Snapshot FLASH MR images of presenting inversion recovery-Z’, contrast and chemical shift contrast for fat and water protons. To demonstrate IR contrast in Snapshot FLASH MR images, an inversion pulse (180”) was placed in front of the imaging sequence, thus creating inversion of the
slice orientation in the human abdomen and thorax. A TR = 5.2 msec for 256 x 256 matrix Snapshot FLASH MR images was necessary to observe the reported effects on the lower images without special calculations or multiple averaging of the images. In the thorax, the images show an exact delineation of the organs. This can be seen in Fig. 2, where 12 coronal images are demonstrated, These images represent a dynamic multislice measurement without contrasting prepulses. Here the heart obviously is shown in the late diastole, which is the longest period of the heart cycle. On the images (256 x 256 matrix) there is no contrast visible. In comparison to slower-resolution FLASH MR images, there are no artifacts from movement or flow in these images. Additionally the pulmonary vessels are depicted further to the periphery of the lung. The observation that the augmentation of the flip angle (higher than 7”) effectively produces attenuated bright vessels is not demonstrated here. Saturation contrast by higher flip angles, however, was hardly achieved and overall S/N was lost by this procedure. To delineate the contrast that can be achieved by prepulses, only typical examples for inversion recovery’ and selective saturation are shown.‘,” In the thorax with a 180” non-selective rf pulse preceding the sequence as shown in Fig. 3, the blood becomes dark due to effective T, relaxation of about 700 msec after the inversion pulse. This is the time from the pulse to the midprojection of the imaging sequence. In Fig. 3 a series of different TI delays is shown (100, 200, and 400 msec). The continuous rise of the blood signal is shown from one image to the next. It should be noted that the subcutaneous fat is not as bright on these images as on selectively water-saturated images demonstrated later.
Fast inversion T, contrast and chemical shift contrast 0 D.
MATTHAEIET AL.
Fig. 2. Coronal Snapshot FLASH MR images of the human thorax. Measuring time 1.4 set, matrix 256 X 256, slice thickness 1.5 cm, flip angle, TE = 2.6 msec, TR = 5.2 msec. Twelve MR images have been measured in a set demonstrating the concise vcntrodosal anatomy. The absence of the movement artifacts can be demonstrated as well as the intratboracic vessels without blurring due to susceptibility effects.
In Fig. 4, the effect of a saturation CHESS-pulse (CHESS = chemical shift selective) is demonstrated on images from the abdomen and lower thorax. Here on the right upper image it can be seen that, after a frequency selective pulse on the protons of water, the image intensity is mostly dominated by the fat-containing structures. The water-containing structures, however, are dark and even black such as the blood that has a T, in the range of 1 sec. In this figure a comparison with the inversion recovery image of this region at the lower left of Fig. 4 is shown. In this image a comparable contrast can be seen; however, fat intesity is lower. The reported experiments demonstrate that fat saturation is not effective due to the short T, relaxation. Thus, as shown at the lower right of Fig. 4, the water image must be generated from the spin-density-weighted image without prepulses by subtraction of the fatweighted image from this data set. This image has most information from water-containing structures. Here the stomach wall, the liver, and the spleen are demarcated in the abdomen, whereas in the thorax and abdomen the great vessels are contrasted by their contents. This is water-containing, long r, blood. The regions that are fat-containing here give dark rims to water-containing structures.
DISCUSSION
The typical Snapshot FLASH MR images have a low contrast and a high signal-to-noise ratio because the signal is read out at the very beginning of the FID. With flip angles between 7 and 12”, an equilibrated magnetization shows images with only faint contrast of inflowing magnetization. Thus, as long as the whole image measurement is shorter than physical and biological processes such as T,, T2, or peristaltic motion, these processes can be shown by the sequence. It was this observation with 64 X 128 matrix images that lead us to call the sequence Snapshot FLASH MRI.” Physical and physiological processes could first be snapped on “neutral” FLASH MR images. The studies reported here were done to provide examples for fast high-resolution Snapshot FLASH MR images and answer the question of whether a useful contrast can be obtained. Thus the most important message from the results reported here is the possibility to generate high-resolution (256 x 256) Snapshot FLASH MR images that contain contrast from preceding pulses. The TR for 256 x 256 MR images of only 5.2 msec is a prerequisite for obtaining these images with this sequence. Slower FLASH
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
Fig. 3. cJoronal MR images of the human upper abdomen and thorax. Here especially the left ventricle is shown with a measuring time of II.4 set and a matrix of 256 x 256 pixels. Two averages have been measured for a better S/N ratio. The upper left image demoonsnrates the Snaoshot FLASH MR image without nreoaration pulses. The upper right image and the bottom row have been measured after an l&P preparing inversion &se.
sequences will show motional artifacts and are not useful for obtaining relaxation- or saturation-contrast, because the entire measuring time exceeds the process under investigations (e.g., motion, I’,, T, and so on). Here the images without preparatory pulses and images after an inversion 180” pulse and after a saturation 90” pulse are shown with a matrix of 256 X 256 measured in about 1.3 sec. Technically, Snapshot FLASH MR images as shown here are still hampered by artifacts from hardware limitations, especially with regard to the gradient systems. Thus, here some disturbed structures at the borders of the image should be ignored. In the center of the images, however, the image is of good quality. Additionally, the incomplete saturation of water-containing structures in the saturated images are due to remaining inhomogeneities of the Bi and Be field. FLASH MR imaging has been disturbed by motion artifacts and artifacts due to susceptibility changes (T2,&. Now high-resolution images (256 X 256 mea-
sured matrix) can be performed in about a second with the Snapshot FLASH MR sequence showing almost no susceptibility artifacts. This can most clearly be seen in the pulmonary region of the images in Fig. 2, where the intrathoracic vessels are visible quite far in the periphery. Here the direct neighborhood of air and tissue creates susceptibility steps, which cause the vessels to appear cut off at the pulmonary hili on images with long TE. The S/N ratio in the images reported here is high, and with hardware improvement, an overall sufficient appearance of the “one-shot” images for clinical use will be possible. Thus it is shown that with reduction in TR there is high signal, few artifacts, and reproducible contrast. For many clinical applications in the future, the one-shot images with an even reduction in measuring time will be a useful on-line imaging device. Thus the results in these first images (measured matrix of 256 x 256 in the range of a second) demonstrate that continuous measurement of the images, 3D MR
Fast inversion T, contrast and chemical shift contrast 0 D. MA~THAEI FI AL.
Fig. 4. Coronal MR images of the human upper abdomen and thorax. Here especially the aorta and vena cava in the thorax are shown with measuring time of 1.4 set and a matrix of 256 x 256 pixels. In the abdomen the liver, the spleen, and the stomach are depicted in detail. Two averages have been measured for a better S/N ratio. The upper left image demonstrates the scene without preparation pulses. On the upper right image the influence of a preparing 90’ water-selective CHESS pulse can be seen. The lower left image shows the appearance of the image after a 180” preparation pulse. The lower right image is the subtraction of the upper right from the upper left image. This image with restrictions discussed in the text can be called a water image.
imaging, and even further acceleration is possible. From earlier studies done with lower resolution, it can be stated that reaching this reduction of measuring times below 500 msec (two images per second), the dynamic performance of contrast becomes possible as well as dynamic motion and perfusion studies of inner organs in high resolution films. Tl contrast with a preceding inversion pulse is effective, and there seems to be a useful effect for on-line clinical tissue differentiation. Here it must be further investigated whether quantitative contrast is achievable by this method. The CHESS (chemical chift selective) contrast can only be achieved for water suppression as demonstrated in Fig. 4. This is due to the long T1 of most water-containing structures. Here a 90” pulse on the water component of the proton spectrum results in the saturation and dephasing of the water protons. The liver and parts
of the kidneys are obviously not totally saturated during the midprojections of the image due to Tl relaxation values in the range of 500-700 msec. Thus, other than in saturation before each pulse as described earlier,’ the procedure reported here does not result in a complete saturation observable during the total imaging time. An attenuation of the unsaturated species and partial saturation recovery of the saturated spins is depicted concomitantly. Thus, fat protons that are not affected from the pulse are the dominant parts of the resulting images, whereas saturated spins are suppressed and depicted with contrast with respect to their T1 relaxation value. As stated above, the same is impossible with fat saturation. This is because the complete relaxation of the fat happens before the central projections of the image, and thus a pure water image cannot be obtained by saturation as long as the measuring time is in the order of 1 sec. As shown in Fig. 4, however, the water image
6
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
can easily be obtained by subtracting the fat-attenuated image from the proton image. This procedure is not realtime imaging. At this stage of the development, it seems to be of great importance that the T, contrast and even chemical shift imaging stay available even after the “quite” long sequence employed here. With the results from these experiments even the 256 x 256 matrix 1.3-set FLASH MR images can now be called “Snapshot” FLASH because they freeze motion and depict processes such as the T, relaxation. It seems possible to further accelerate the technique to about 50% reduced measuring times. This would mean 700 msec for a measured 256 x 256 matrix. Until now, some fast MR imaging methods have been available. However, they are infrequently used because some drawbacks of the method still exist. The actual speed of ECHO PLANAR Ml? imaging and the spatial and contrast resolution of images reported here give hope for clinical practice that-with foreseeable developments of the techniques-interactive multiparameter clinical MR imaging will be available soon. REFERENCES 1. Mansfield, P.; Pykett, I.L. Biological and medical imaging by NMR. J. Magn. Reson. 29:355-373; 1978.
2. Rzedzian, R.; Pykett, I.L. Routine acquisition of MR cardiac movies using the instant scan technique. Radiology 161(P):338; 1986. (Abstract.) 3. Haase, A.; Frahm, J.; Matthaei, D.; H%nicke, W.; Merboldt, K.D. FLASH imaging. Rapid NMR imaging using low flip angle pulses. J. Magn. Reson. 67:258-263; 1986. 4. Matthaei, D.; Frahm, J.; Haase, A.; H&nicke, W. Regional physiologic functions depicted by sequences of rapid magnetic resonance images. Lancet 2:893; 1985. 5. Haase, A. SNAPSHOT FLASH MRI. Application to T,-, T,- and chemical shift imaging. Magn. Reson. Med. 13: 77-89; 1990. 6. Matthaei, D.; Haase, A.; Hemich, D.; Dtihmke, E. Cardiac and vascular imaging with an MR Snapshot technique. Radiology 177527-532; 1990. 7. Klose, U.; Nagele, T.; Grodd, W.; Peterson, D. Variation of contrast between different brain tissues with an MR Snapshot technique. Radiology 176:578-581; 1990. 8. Bydder, G.M.; Young, I.R. MRI: Clinical use of the inversion recovery sequence. J. Comput. Assist. Tomogr. 9(6): 1020-1032; 1985. 9. Haase, A.; Frahm, J.; Hanicke, W.; Matthaei, D. ‘HNMR chemical shift selective (CHESS) imaging. Phys. Med. Biol. 30:341-344, 1985. 10. Bydder, G.M.; Young, I.R. Clinical use of the partial saturation and saturation recovery sequence in MR imaging. J. Comput. Assist. Tomogr. 9:1020-1032; 1985. 11. Haase, A.; Matthaei, G.; Bartkowski, R.; Dtihmke, E.; Leibfritz, D. Inversion recovery SNAPSHOT FLASH MRI, fast dynamic T, contrast. J. Comput. Assist. Tomogr. 13:10361040; 1989.
Imaging, Vol. 10. Pp. l-6, FTinted in the USA. All rights reserved.
Magnetic
1992 Copyright
0730-725X/92 SS.00 + .@I 0 1992 PerSamon F’ms plc
0 Original Contribution
FAST INVERSION RECOVERY Tl CONTRAST AND CHEMICAL SHIFT CONTRAST IN HIGH-RESOLUTION SNAPSHOT FLASH MR IMAGES DIETER MATTHAEI, * AXEL HAASE,~DIETMAR HENRICH,* AND ECKHART DUHMKE* *Klinik Kir Strahlentherapie, 3400 Gottingen, Robert-Koch-Strasse40, tUniversitlt Wurzburg, 8700 Wiirzburg, PhysikalischesInstitut, Am Hubland, and SBRUKBRMedizintechnik,7500 Karlsruhe, Germany Fast MR imaging attracts the interest of both clinicians and physicists because new diagnostic information arises with reduced artifacts due to short investigational times. With the acceleration of the Snapshot FLASH MR sequence, the measurement of high-resolution images with 256 x 256 matrix is reported, together with contrasting prepulses that are applied to attain contrast in combination with higher in-plane resolution. Measuring times are in the range of a second. For whole-body imaging, a TR = 5.2 msec and a TE = 2.6 msec could be attained measuring omit 256 x 256 matrix images. Artifact-free images demonstrating T, contrast and contrast from chemical shift are performed on moving organs (heart, intestine) in different experiments. These applications can easily be performed in a couple of minutes for clinical use. Especially in the lung, short TE and high resolution result in a new imaging quality of pulmonary and mediastinal vessels. Keywords: Magnetic resonance (MR) physics; Magnetic resonance (MR) pulse sequences; Magnetic resonance (MR) contrast enhancement; Magnetic resonance (MR) fast imaging; Magnetic resonance (MR) cardiac imaging; Magnetic resonance (MR) pulmonary imaging; Magnetic resonance (MR) abdominal imaging.
INTRODUCTION
More than 10 years ago, Peter Mansfield’ proposed the ECHO PLANAR sequence, which allowed MR imaging investigations in fractions of a second with reduced in-plane resolution. Recently, higher spatial resolution for ECHO PLANAR imaging is reported* which obviously makes the sequence gradually longer. The FLASH MR sequence was proposed in 19853, that uses a partial flip angle for excitation and a gradient echo for the detection of the signal. This idea began with somewhat slow images and is now being developed to become faster and more highly resolved. At the beginning, this sequence allowed the measurement of images in a few seconds on conventional imaging hardware. Dynamic delineation of vascular and parenchyma1 contrast after Gd-DTPA4 could be shown with this sequence, as well as three-dimensional MR imaging in some minutes for the first time. With reduced spatial resolution and fast gradient RECEIVED 2128191; ACCEDED6128191. Address reprint requests to Dieter Matthaei, Klinik fiir
switchings, a variation of this sequence, dubbed Snapshot FLASH, was proposed last year for the fast imaging of dynamic subsecond processes5 . Rapid cardiac and vascular clinical applications were introduced6 employing FLASH sequences. Until now, the further acceleration of the FLASH MR sequence has allowed measurement with a TR in the range of 5 msec for 256 x 256 matrices, thus enabling high-resolution images in about a second. Comparable in-plane resolutions for head imaging, however, were demonstrated in around a second.7 With high-resolution FLASH MR images, the delineation of many anatomical details is feasible, which has already been shown for slower versions of the method. It was, however, questionable, with former timings and sensitivity of the electronics, whether these images would be able to display contrast due to T, relaxation and due to chemical shift selective imaging, which is performed by presaturation of the unwanted population (fat or water). Strahlentherapie, Gemany.
3400 Gbttingen, Robert-Koch-Strasse
40,
2
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
Fig. 1. Two Snapshot FLASH MR sequences are demonstrated with preparing pulses before them. (a) A 180” pulse is placed before the imaging sequence with different delays. Here for the images in Fig. 3 a lOO-, 200-, and 300-msec delay were used. (b) For the saturation of one line in the proton spectrum, a selective 90” pulse was placed before the imaging sequence. A lo-msec spoiler delay was used additionally to dephase the saturated spins.
spins in the whole object. As shown in Fig. l(a), a delay of 100,200, and 300 msec TI between the pulse and the sequence was chosen here to demonstrate that the evolution of the contrast becomes visible. Additional experiments were done to demonstrate the capability of the Snapshot FLASH MR sequence of delineating contrast in the images due to selective saturation of one proton species. Here selective 90°C pulses were used before the sequence to saturate either the water or the fat protons as shown in Fig. l(b). To dephase the saturated spins, an additional gradient of 10 msec duration was applied before imaging.
RESULTS High-resolution Snapshot FLASH MR images with a T, and chemical shift contrast were done with coronal
MATERIALS AND METHODS Here high-resolution Snapshot FLASH MR wholebody images are reported, which on a commercial imaging machine, with only a reduced gradient diameter (60 cm), are measured with a TR of 5.2 msec. Thus a 256 X 256 matrix is measured in about 1.4 sec. Most images were measured with single excitation. For these experiments, a 2-T whole-body MR imaging machine (BRUKER Medizintechnik, Karlsruhe) was used in combination with gradients of about 12 mT/m. The diameter of the gradient system was 60 cm to reduce the artifacts and allow lower gradient switching times. The investigations were done on healthy volunteers after careful physical investigations. Informed consent was obtained before each imaging procedure. The above images were obtained in the abdomen and thorax to show the measurement of the beating heart without triggering and the intraabdominal organs without artifacts form peristalsis. On the one hand, the demonstration of high-resolution Snapshot FLASH MR images in the human trunk here was planned to demonstrate the in-plane resolution without artifacts. On the other hand it was intended to show whether pulses before the imaging experimentaffecting the spins in the imaging plane- would keep their contrasting effect with the imaging speed attained here. Two contrasting experiments were thus done to evaluate the capability of the high-resolution Snapshot FLASH MR images of presenting inversion recovery-Z’, contrast and chemical shift contrast for fat and water protons. To demonstrate IR contrast in Snapshot FLASH MR images, an inversion pulse (180”) was placed in front of the imaging sequence, thus creating inversion of the
slice orientation in the human abdomen and thorax. A TR = 5.2 msec for 256 x 256 matrix Snapshot FLASH MR images was necessary to observe the reported effects on the lower images without special calculations or multiple averaging of the images. In the thorax, the images show an exact delineation of the organs. This can be seen in Fig. 2, where 12 coronal images are demonstrated, These images represent a dynamic multislice measurement without contrasting prepulses. Here the heart obviously is shown in the late diastole, which is the longest period of the heart cycle. On the images (256 x 256 matrix) there is no contrast visible. In comparison to slower-resolution FLASH MR images, there are no artifacts from movement or flow in these images. Additionally the pulmonary vessels are depicted further to the periphery of the lung. The observation that the augmentation of the flip angle (higher than 7”) effectively produces attenuated bright vessels is not demonstrated here. Saturation contrast by higher flip angles, however, was hardly achieved and overall S/N was lost by this procedure. To delineate the contrast that can be achieved by prepulses, only typical examples for inversion recovery’ and selective saturation are shown.‘,” In the thorax with a 180” non-selective rf pulse preceding the sequence as shown in Fig. 3, the blood becomes dark due to effective T, relaxation of about 700 msec after the inversion pulse. This is the time from the pulse to the midprojection of the imaging sequence. In Fig. 3 a series of different TI delays is shown (100, 200, and 400 msec). The continuous rise of the blood signal is shown from one image to the next. It should be noted that the subcutaneous fat is not as bright on these images as on selectively water-saturated images demonstrated later.
Fast inversion T, contrast and chemical shift contrast 0 D.
MATTHAEIET AL.
Fig. 2. Coronal Snapshot FLASH MR images of the human thorax. Measuring time 1.4 set, matrix 256 X 256, slice thickness 1.5 cm, flip angle, TE = 2.6 msec, TR = 5.2 msec. Twelve MR images have been measured in a set demonstrating the concise vcntrodosal anatomy. The absence of the movement artifacts can be demonstrated as well as the intratboracic vessels without blurring due to susceptibility effects.
In Fig. 4, the effect of a saturation CHESS-pulse (CHESS = chemical shift selective) is demonstrated on images from the abdomen and lower thorax. Here on the right upper image it can be seen that, after a frequency selective pulse on the protons of water, the image intensity is mostly dominated by the fat-containing structures. The water-containing structures, however, are dark and even black such as the blood that has a T, in the range of 1 sec. In this figure a comparison with the inversion recovery image of this region at the lower left of Fig. 4 is shown. In this image a comparable contrast can be seen; however, fat intesity is lower. The reported experiments demonstrate that fat saturation is not effective due to the short T, relaxation. Thus, as shown at the lower right of Fig. 4, the water image must be generated from the spin-density-weighted image without prepulses by subtraction of the fatweighted image from this data set. This image has most information from water-containing structures. Here the stomach wall, the liver, and the spleen are demarcated in the abdomen, whereas in the thorax and abdomen the great vessels are contrasted by their contents. This is water-containing, long r, blood. The regions that are fat-containing here give dark rims to water-containing structures.
DISCUSSION
The typical Snapshot FLASH MR images have a low contrast and a high signal-to-noise ratio because the signal is read out at the very beginning of the FID. With flip angles between 7 and 12”, an equilibrated magnetization shows images with only faint contrast of inflowing magnetization. Thus, as long as the whole image measurement is shorter than physical and biological processes such as T,, T2, or peristaltic motion, these processes can be shown by the sequence. It was this observation with 64 X 128 matrix images that lead us to call the sequence Snapshot FLASH MRI.” Physical and physiological processes could first be snapped on “neutral” FLASH MR images. The studies reported here were done to provide examples for fast high-resolution Snapshot FLASH MR images and answer the question of whether a useful contrast can be obtained. Thus the most important message from the results reported here is the possibility to generate high-resolution (256 x 256) Snapshot FLASH MR images that contain contrast from preceding pulses. The TR for 256 x 256 MR images of only 5.2 msec is a prerequisite for obtaining these images with this sequence. Slower FLASH
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
Fig. 3. cJoronal MR images of the human upper abdomen and thorax. Here especially the left ventricle is shown with a measuring time of II.4 set and a matrix of 256 x 256 pixels. Two averages have been measured for a better S/N ratio. The upper left image demoonsnrates the Snaoshot FLASH MR image without nreoaration pulses. The upper right image and the bottom row have been measured after an l&P preparing inversion &se.
sequences will show motional artifacts and are not useful for obtaining relaxation- or saturation-contrast, because the entire measuring time exceeds the process under investigations (e.g., motion, I’,, T, and so on). Here the images without preparatory pulses and images after an inversion 180” pulse and after a saturation 90” pulse are shown with a matrix of 256 X 256 measured in about 1.3 sec. Technically, Snapshot FLASH MR images as shown here are still hampered by artifacts from hardware limitations, especially with regard to the gradient systems. Thus, here some disturbed structures at the borders of the image should be ignored. In the center of the images, however, the image is of good quality. Additionally, the incomplete saturation of water-containing structures in the saturated images are due to remaining inhomogeneities of the Bi and Be field. FLASH MR imaging has been disturbed by motion artifacts and artifacts due to susceptibility changes (T2,&. Now high-resolution images (256 X 256 mea-
sured matrix) can be performed in about a second with the Snapshot FLASH MR sequence showing almost no susceptibility artifacts. This can most clearly be seen in the pulmonary region of the images in Fig. 2, where the intrathoracic vessels are visible quite far in the periphery. Here the direct neighborhood of air and tissue creates susceptibility steps, which cause the vessels to appear cut off at the pulmonary hili on images with long TE. The S/N ratio in the images reported here is high, and with hardware improvement, an overall sufficient appearance of the “one-shot” images for clinical use will be possible. Thus it is shown that with reduction in TR there is high signal, few artifacts, and reproducible contrast. For many clinical applications in the future, the one-shot images with an even reduction in measuring time will be a useful on-line imaging device. Thus the results in these first images (measured matrix of 256 x 256 in the range of a second) demonstrate that continuous measurement of the images, 3D MR
Fast inversion T, contrast and chemical shift contrast 0 D. MA~THAEI FI AL.
Fig. 4. Coronal MR images of the human upper abdomen and thorax. Here especially the aorta and vena cava in the thorax are shown with measuring time of 1.4 set and a matrix of 256 x 256 pixels. In the abdomen the liver, the spleen, and the stomach are depicted in detail. Two averages have been measured for a better S/N ratio. The upper left image demonstrates the scene without preparation pulses. On the upper right image the influence of a preparing 90’ water-selective CHESS pulse can be seen. The lower left image shows the appearance of the image after a 180” preparation pulse. The lower right image is the subtraction of the upper right from the upper left image. This image with restrictions discussed in the text can be called a water image.
imaging, and even further acceleration is possible. From earlier studies done with lower resolution, it can be stated that reaching this reduction of measuring times below 500 msec (two images per second), the dynamic performance of contrast becomes possible as well as dynamic motion and perfusion studies of inner organs in high resolution films. Tl contrast with a preceding inversion pulse is effective, and there seems to be a useful effect for on-line clinical tissue differentiation. Here it must be further investigated whether quantitative contrast is achievable by this method. The CHESS (chemical chift selective) contrast can only be achieved for water suppression as demonstrated in Fig. 4. This is due to the long T1 of most water-containing structures. Here a 90” pulse on the water component of the proton spectrum results in the saturation and dephasing of the water protons. The liver and parts
of the kidneys are obviously not totally saturated during the midprojections of the image due to Tl relaxation values in the range of 500-700 msec. Thus, other than in saturation before each pulse as described earlier,’ the procedure reported here does not result in a complete saturation observable during the total imaging time. An attenuation of the unsaturated species and partial saturation recovery of the saturated spins is depicted concomitantly. Thus, fat protons that are not affected from the pulse are the dominant parts of the resulting images, whereas saturated spins are suppressed and depicted with contrast with respect to their T1 relaxation value. As stated above, the same is impossible with fat saturation. This is because the complete relaxation of the fat happens before the central projections of the image, and thus a pure water image cannot be obtained by saturation as long as the measuring time is in the order of 1 sec. As shown in Fig. 4, however, the water image
6
Magnetic Resonance Imaging 0 Volume 10, Number 1, 1992
can easily be obtained by subtracting the fat-attenuated image from the proton image. This procedure is not realtime imaging. At this stage of the development, it seems to be of great importance that the T, contrast and even chemical shift imaging stay available even after the “quite” long sequence employed here. With the results from these experiments even the 256 x 256 matrix 1.3-set FLASH MR images can now be called “Snapshot” FLASH because they freeze motion and depict processes such as the T, relaxation. It seems possible to further accelerate the technique to about 50% reduced measuring times. This would mean 700 msec for a measured 256 x 256 matrix. Until now, some fast MR imaging methods have been available. However, they are infrequently used because some drawbacks of the method still exist. The actual speed of ECHO PLANAR Ml? imaging and the spatial and contrast resolution of images reported here give hope for clinical practice that-with foreseeable developments of the techniques-interactive multiparameter clinical MR imaging will be available soon. REFERENCES 1. Mansfield, P.; Pykett, I.L. Biological and medical imaging by NMR. J. Magn. Reson. 29:355-373; 1978.
2. Rzedzian, R.; Pykett, I.L. Routine acquisition of MR cardiac movies using the instant scan technique. Radiology 161(P):338; 1986. (Abstract.) 3. Haase, A.; Frahm, J.; Matthaei, D.; H%nicke, W.; Merboldt, K.D. FLASH imaging. Rapid NMR imaging using low flip angle pulses. J. Magn. Reson. 67:258-263; 1986. 4. Matthaei, D.; Frahm, J.; Haase, A.; H&nicke, W. Regional physiologic functions depicted by sequences of rapid magnetic resonance images. Lancet 2:893; 1985. 5. Haase, A. SNAPSHOT FLASH MRI. Application to T,-, T,- and chemical shift imaging. Magn. Reson. Med. 13: 77-89; 1990. 6. Matthaei, D.; Haase, A.; Hemich, D.; Dtihmke, E. Cardiac and vascular imaging with an MR Snapshot technique. Radiology 177527-532; 1990. 7. Klose, U.; Nagele, T.; Grodd, W.; Peterson, D. Variation of contrast between different brain tissues with an MR Snapshot technique. Radiology 176:578-581; 1990. 8. Bydder, G.M.; Young, I.R. MRI: Clinical use of the inversion recovery sequence. J. Comput. Assist. Tomogr. 9(6): 1020-1032; 1985. 9. Haase, A.; Frahm, J.; Hanicke, W.; Matthaei, D. ‘HNMR chemical shift selective (CHESS) imaging. Phys. Med. Biol. 30:341-344, 1985. 10. Bydder, G.M.; Young, I.R. Clinical use of the partial saturation and saturation recovery sequence in MR imaging. J. Comput. Assist. Tomogr. 9:1020-1032; 1985. 11. Haase, A.; Matthaei, G.; Bartkowski, R.; Dtihmke, E.; Leibfritz, D. Inversion recovery SNAPSHOT FLASH MRI, fast dynamic T, contrast. J. Comput. Assist. Tomogr. 13:10361040; 1989.
