3D-SNAPSHOT FLASH NMR imaging of the human heart

Magneric Resonance Imaging. Vol. g. pp. 377.379, 1994 Printed in the USA. All rights reserved.

CopyrIght

0730-725X190 $3.00 + .oO 0 1990 Pergamon Pros plc

l Original Contribution

3D-SNAPSHOT FLASH NMR IMAGING OF THE HUMAN HEART D. HENRICH, BRUKER Medizintechnik,

A.

HAASE,*

AND D. MATTHAEI~

Karlsruhe, *University of Wiirzburg, j-university

of Gottingen,

FRG

SNAPSHOT-FLASH

is a recently developed, ultrafast imaging technique, based on conventional FLASH imaging. The application of this new variant to 3D imaging allows the acquisition of a 128 x 128 x 32 data set in 12.5 seconds without triggering, or for cardiac imaging with gating within 32 heartbeats. Compared to standard 3D-FLASH this is 128 times faster, because triggering is only required when the 3D phase-encoding gradient is incremented. The method depicts for the first time fast three-dimensional views of the human heart without motional artifacts. The images are spin-density weighted. Using suitable prepulses any desired T,- or T2- contrast may be achieved. The generation of 3D movies is possible without an increase of the total scan time. Keywords:

3D imaging; Rapid imaging; Heart.

msec, resulting in a total scan time of 200 msec for

INTRODUCTION

a 128 x 64 data matrix. Without cardiac triggering, 5 frames per second can be imaged. By replacing the slice gradient with a second phase-encoding gradient,

of the human heart and other moving organs is a difficult task for MRI. To freeze all movements it is necessary to use very fast imaging techniques such as Echo-Planar-Imaging (EPI) or SNAPSHOT-FLASHImaging (SFI). 2,3 In EPI it is possible to obtain a 128 x 64 image with a single RF-excitation in a time between 32 msec and 100 msec, whereas in SF1 using multiple low angle excitation it now takes about 170 to 200 msec for one image with an identical matrix size. The advantage of SF1 is the insensitivity of the image contrast to the SFI-pulse sequence. With variable prepulses a wide range of relaxation contrast can be achieved in the image. Moreover SF1 is much more easily adapted to routine clinical MR-equipment, while in EPI the hardware requirements are much more stringent. Imaging

G-SLICE

i! (’

G-READ

v

METHODS

\

ECHO

G-PHASE

SF1 has been implemented on a 2-Tesla BRUKER whole-body MR system. A low inductance gradient coil can be switched to provide a gradient strength of 5 mT/m within 300 1~s. A schematic diagram of the 2DFT-sequence is shown in Fig. 1. As in standard FLASH a low-angle excitation pulse is used. The echo time is only 1.5 msec, while the repetition time is 3

w Fig. 1. SNAPSHOT-FLASH. The timing of the 2DFTsequence (TE = 1.5 msec, TR = 3 msec).

11/10/89; ACCEPTED 2/18/90. Address correspondence to Dr. D. Henrich, BRUKER

Medizintechnik, FRG.

RECEIVED

377

Wikinger-Str . 13, D-7500 Karlsruhe

21,

378

Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990

Fig. 2. Coronal views from the first ECG-triggered 32 heartbeats.

3D data set of the human heart (matrix size 128 x 128 x 32) obtained

within

3D-SNAPSHOT FLASH NMR

imaging of the human heart l

the pulse sequence is converted into a 3D imaging sequence. It is also possible to acquire an anisotropic 3D data set in nonorthogonal orientations, i.e., an arbitrary orientation of the 3D volume may be chosen. The spatial resolution of one voxel corresponds to 2.5 x 2.5 x 10 mm3. In all of the experiments described here we used a whole-body resonator coil for RF-transmission and a Helmholtz surface coil for signal detection. A pin diode switch device was used for active decoupling of the two RF-coils. For the excitation of the sample a lo-degree flip angle cx was applied, using a Gaussian-shaped RF-pulse of 0.5 msec duration. The signal intensity S can be described by the following expression (Ref. 4) if the residual transverse magnetization has been spoiled: s = kp(1 - exp(-TR/T1)) sin(a) exp(-TR/T,) 1 - COS((Y)exp( -TR/T’) (1) The factor k denotes an instrumental constant and p is the spin density. The variable phase-encoding gradients spoil the transverse magnetization. Due to the experimental conditions employed, short TR, TE and low CY,the signal intensity S is mainly dependent on the spin density and not on the relaxation times Tr and T2. The resulting images are strongly spin-density weighted. By performing a conventional NMR experiment prior to the image acquisition any desired contrast can be achieved. RESULTS

AND DISCUSSION

An untriggered 3D data set with an image matrix size of 128 x 128 x 32 (or 128 x 64 x 64) can be acquired in only 12.5 sec. This type of experiment was applied mainly for imaging the human head or the abdomen during breath-holding. For imaging the human heart we used the 3D-SF1 method with cardiac gating. The pulse sequence starts after a predefined delay following the R-wave of the ECG-signal. Because the scan time for a single image with a matrix size of 128 x 128 is only 400 msec, cardiac triggering is only necessary during the incrementation of the second phase-encoding gradient that provides the third spatial dimension. Therefore, the total scan time for a complete 3D data set is equa1 to the duration of 32 heartbeats, or about half a minute. In conventional 3D-FLASH imaging of the heart, triggering must occur after the first phase-encoding interval in the second spatial direction because it takes at least about 1.5 to 2 set to scan one image plane. Therefore, a 3D-

D. HENRICHET AL.

379

dataset with a matrix size of 128 x 128 x 32 would require 4096 heartbeats to be sampled - corresponding to a scan time of about 68 min (!) in comparison to only the half minute required with 3D-SFI. Therefore, 3D imaging of the human heart may well establish clinical relevancy. Figure 2 shows 15 coronal slices cut from the first 3D data set of the human heart which could be obtained up to now. Because of the spin density weighting of the images, no delineation between the heart muscle and the blood is visible. As mentioned above, in order to see such a delineation, prepulses must be used. Due to the very short echo time of 1.5 msec, no artifacts arising from susceptibility differences within the body could be observed. As a consequence of this fact, the blood vessels of the lung are for the first time clearly visible without any additional manipulation. This may help for example in the diagnosis of pulmonary artery embolism. Furthermore, no flow or motional artifacts can be discerned in the images. CONCLUSION

Using SFI, 3D imaging of the human heart with a matrix size of 128 x 128 x 32 can be performed in the reasonably short time of 32 heartbeats. Therefore this method is now suitable for clinical cardiac studies, allowing the investigation of pathologies such as infarction and valvular diseases. In combination with prepulses, relaxation enhancement of the image contrast can be achieved. The generation of 3D heartmovies is also possible. When using a 128 x 64 x 64 image matrix, up to five 3D datasets each with a different delay within the heart cycle can be obtained in approximately one min. An improvement in temporal and spatial resolution is necessary and possible to realize. REFERENCES

Haase, A.; Frahm, J.; Matthaei, D.; Merboldt, K.D.; Hanicke, W. FLASH imaging. Rapid NMR imaging using low flip angle pulses. J. Magn. Reson. 67:258-266; 1986. Haase, A. SNAPSHOT FLASH MRI. Applications to T,-, T,-, and chemical shift imaging. J. Magn. Reson. 13:77-89; 1990. Haase, A.; Matthaei, D.; Henrich, D.; Norris, D.; Leibfritz, D. Cardiac NMR imaging using SNAPSHOT FLASH NMR. SMRM Abstracts, p. 56; 1989. Waugh, J.S. Sensitivity in Fourier-transform NMR spectroscopy of slowly relaxing systems. J. Mol. Spectrosc. 35:298-305;

1970.