British Journal of Radiology (2003) 76, 788-791
© 2003 British Institute of Radiology
doi: 10.1259/bjr/28705135
Contrast-to-noise ratio of multiple slice spin lock technique: prospects for liver imaging
J Halavaara, MD
1
S Lukkarinen, MSc
2
R Sepponen, PhD
2
A Markkola, MD
3 and
J Tanttu, PhD
4
1 Jorvi Hospital, Department of Radiology, Turuntie 150, 02740 Espoo, 2 Helsinki University of Technology, Faculty of Electrical Engineering, Otakaari 5 A, 02150 Espoo, 3 Helsinki University Central Hospital, Department of Radiology, Haartmaninkatu 4, 00290 Helsinki and 4 Philips Medical Imaging, Helsinki, Finland
 |
Abstract
|
|---|
Spin lock (SL) MRI technique has been demonstrated to provide similar lesion/liver contrast to conventional MR technique. Multiple slice SL technique allows a large number of slices to be collected within a given repetition time due to the short echo time. In addition, the short echo time reduces movement and susceptibility artefacts. In the present study, the potential of the multiple slice SL technique in liver imaging was evaluated by using tissue nuclear magnetic resonance (NMR) information and tissue NMR parameters obtained at 0.1 T. 10 healthy volunteers were imaged at 0.1 T for the measurement of tissue T1
, T1, and T2 relaxation times. Tissue radiofrequency-attenuation information was obtained from the literature, and included in the contrast-to-noise ratio (CNR) calculations. Our results demonstrated that by increasing the number of slices the acquired liver-to-spleen CNR decreases with all locking field durations (locking time, TL). However, with small TLs, the difference is small which is important for liver MRI where a wide coverage, i.e. large number of slices, is important. Long locking pulse durations are more favourable than short TLs if large flip angles are used. With an optimal combination of a moderate amount of slices, reasonably large flip angle, and TL of the order of 20 ms, high CNR is achieved in SL MRI.
|
Introduction
|
|---|
The feasibility of the multiple slice spin lock (SL) technique to generate magnetization transfer (MT) like contrast with a good lesion discrimination capability in liver imaging has been demonstrated [1]. When low locking field strengths (B1L) are utilized, the multiple slice SL imaging technique produces images with contrast similar to that of T2 weighted sequences, and the contrast is generated in the whole imaging volume during short locking periods (locking time, TL). The locking pulses are applied between sequential excitations of slices. Images with accumulated T1-contrast may be obtained by using a short echo time (TE) [2]. The inherent advantages of the multiple slice SL technique are a large number of slices (N) that may be obtained during a given repetition time (TR), small motion and susceptibility artefacts due to short TE, and the possibility to utilize gradient echo (GE) technique. These advantages have been demonstrated in head and neck [3] and high field [4] spin lock MRI.
The contrast provided by conventional spin echo (SE) technique is more sensitive to the variation of the intensity of the excitatory radiofrequency (RF) field than that provided by the GE technique. Due to the high electrical conductivity of liver tissue, the penetration depth of the RF field is rather small and decreases as the frequency increases [5, 6]. An optimal design of the multiple slice SL sequence for clinical imaging assumes knowledge of the nuclear magnetic resonance (NMR) and electrical properties of target tissues. The purpose of the present investigation was to evaluate the potential of the multiple slice SL technique in liver imaging by performing computer simulation based on signal equations with determined tissue NMR parameters obtained at 0.1 T. On the basis of the results, optimal imaging parameters for the multiple slice SL technique in liver MRI may be proposed.
|
Methods and materials
|
|---|
Measurements of tissue parameters
10 healthy volunteers were imaged at 0.1 T (Merit, Picker Nordstar Inc., Helsinki, Finland) by using a standard solenoidal body coil. There were 2 females and 8 males with a mean age of 41 years (range, 23–66 years). From each volunteer, T1, T2 and T1 relaxation times from selected abdominal tissues were determined. After a coronal localizer, a transaxial multiple slice inversion recovery (IR) sequence with a TR of 1500 ms and a TE of 40 ms, and inversion times (TI) of 50 ms, 200 ms, and 387 ms was employed for the T1-relaxation time measurements. With two averages, the acquisition time (TA) was 12 min 48 s. Then, an optimal slice section including liver, spleen, paravertebral muscle, and subcutaneous fat tissue was selected for the determination of T1 and T2 relaxation times from single slice acquisitions. The T1 relaxation times were measured with the SL sequence, TR=1500 ms, TE=40 ms and four durations of locking pulses (TL=10 ms, 50 ms, 100 ms, and 150 ms, respectively) with B1L=40 µT and TA of 12 min 48 s. The T2 relaxation times were obtained by using a SE sequence with TR of 1500 ms, TEs of 40 ms and 90 ms and TA of 6 min 40 s. The matrix size (MA) of 256 x 192 pixels, field-of-view (FOV) of 553 x 415 mm, and a slice thickness of 10 mm were used for all measurements. The standard region-of-interest (ROI) method was used for the signal intensity measurements.
The T1 was calculated by measuring tissue signal intensity as a function of TL. In order to obtain the T1 values, the signal intensity values (S) were fitted to the Equation 1
with the least-squares method.
where S0 is the tissue signal intensity corresponding to the equilibrium magnetization. The T1 and T2 values were obtained by fitting the corresponding measured intensities to the known signal equations of IR and SE sequences [7].
Contrast simulations
Liver/spleen contrast-to-noise (CNR) simulations were based on tissue relaxation data. The tissue NMR information thus obtained was used to evaluate the performance of the multiple slice SL technique with different imaging parameters. The variables were the number of acquired slices (N), the strength of the locking pulse, the length of the locking pulse, and flip angle (
). The simulations were done by using a standard PC and MATLAB numeric computation software (version 5.2, The Mathworks Inc, Natick, MA).
The signal intensity for multiple slice SL sequence was calculated by using Equation 2 [7].
where S0 is the equilibrium magnetization, is the flip angle, N is the number of slices, and TD was obtained from the Equation 3.
where TR=1500 ms, TE=15 ms and N=10.
For the SE sequence, the signal intensity was obtained from the Equation 4 (7):
where TRref(SE)=1500 ms, TEref(SE)=120 ms, the flip angle=90°, and TD(SE)=TR-TE/2.
The effect of RF attenuation on SE signal intensity due to the change of was included via the approximate correction function, 
. The attenuation of signals emitted by the excited spins is equal for all sequences, and therefore only the attenuation effect via has been included.
The liver/spleen CNR was calculated by using Equation 5.
where S2 and S1 are the signal intensities for spleen and liver, respectively, and N is the noise. The liver/spleen contrast was chosen because it has previously been demonstrated that in liver tumour imaging a sequence with good liver/spleen contrast properties may also be effective in liver tumour MRI.
|
Results
|
|---|
The obtained relaxation data are summarized in Table 1 (see also Figure 1). The liver and muscle demonstrated the shortest T1 values at B1L=40 mT, 82.1±19.2 ms and 106.7±27.4 ms, respectively. The T1 for the spleen was considerably longer, 121.5±31.4 ms. However, the subcutaneous fat tissue showed the longest T1 relaxation with T1 of 153.0±15.6 ms. This was equal to the T1 value (152.5±11.9 ms) for fat tissue.
View larger version (149K):
[in this window]
[in a new window]
|
Figure 1. A section of liver obtained with the SL technique with different locking pulse durations (TL). (a) TL=0 ms, (b) TL=10 ms, (c) TL=20 ms and (d) TL=40 ms. Note the strongly decreasing signal from normal liver parenchyma as the TL increases.
|
|
The simulated CNR curves for the liver-spleen tissue pair are presented in Figures 2 and 3. As demonstrated in Figure 2, with different TL values the CNR demonstrated strong dependence on the amount of acquired slices. The highest CNR was achieved with long TLs and small amount of slices. The differences in the CNR values with various amounts of slices were small when the length of the locking pulse was kept short. In Figure 3, the CNR of the multiple slice SL sequence is shown as a function of and TL with varying from 0 to 90° and TL from 0 to 40 ms in steps of 10 ms. Maximal liver-spleen contrast-to-noise is achieved with the TL value of 20 ms and with about 60°.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 2. Graph demonstrating the contrast-to-noise ratio (CNR) dependence on the number of slices (N) with varying locking field lengths (TL). The fewer the number of slices, the higher the acquired CNR. However, with short locking pulse durations, the differencies in CNR are small.
|
|
|
Discussion
|
|---|
Previously, the multiple slice SL MRI technique has shown potential in liver tumour imaging [8], and in the differentiation between benign and malignant head and neck tumours [3]. It has been demonstrated that relaxation mechanisms based on protein/water interaction are very efficient at low magnetic field strengths [9]. The SL technique allows the exploration of spin relaxation mechanisms at very low magnetic field strengths (of the order of microteslas), while maintaining the high signal-to-noise provided by the polarizing main magnetic field [10, 11]. With low specific absorption rates (SAR) inherent in multiple slice SL imaging, the technique is applicable also at high field systems as suggested by the results reported by Engelhardt and Johnson [4]. They proposed the SL technique to be used at high field strengths in order to avoid susceptibility and motion artefacts. In their study, the SL technique showed also potential in MR histology.
Several reports have demonstrated that the on-resonance SL technique with low B1L strength provides image contrast properties similar to T2 weighted MR sequences. In the head and neck region, Markkola et al [3] measured T1 values at B1L of 35 µT for protein-rich tissues close to the respective T2 relaxation times. Recently, Virta et al [12] demonstrated in samples with variable protein content that T2 and T1 values were similar with zero B1L, and that T1 increased with increasing B1L. Our present results demonstrated considerably longer T1 values at 40 µT for all evaluated tissues when compared with the respective T2 values. The liver T1 was substantially longer than the liver T2 value. This may have several explanations. Normal liver parenchyma contains a variable amount of fat (in norml livers less than 5%). As shown with our results and supported by the findings by Markkola et al [3] and Virta et al [12], fat tissue has practically as long T1 relaxation time as T1 at 0.1 T. Thus, liver fat content may lenghten the T1 value while having smaller effects on the T2 relaxation time. Also, normal liver parenchyma contains a considerable amount of iron, with 1.6 mg of iron per gram of liver dry weight as the maximum allowed liver iron load [13]. Magnetic substances such as iron and copper present in liver parenchyma may play a role in the difference between the measured T1 and T2 values for the liver due to the smaller sensitivity of T1 on the susceptibility effects when compared with T2.
The observed difference between T2 and T1 values of paravertebral muscle may be explained by the relatively high fat content. A similar finding has been detected by Virta et al [14] in muscles with myositis with high fat content. If the fat content is low as in the muscles in the head and neck region, the T1 and T2 values are close to one another at B1L 35 µT [15].
The RF-attenuation in tissues causes variation in the flip angle [5, 6]. Because most of the liver lies deep in the abdominal cavity, the RF-attenuation may be assumed significant. By using the SL technique, the dependence of the CNR on the flip-angle is small when large angles are employed (Figure 3
). Conventional T2 weighted SE sequence demonstrates sin(3) dependence. Therefore, there is more variation in the flip-angle reducing the CNR of the SE technique proportionally more than the CNR of the SL technique. This is a problem especially at high field strengths, where there is even more variation in the flip-angles due to the increased RF-attenuation of higher frequencies. The contrast in conventional spin echo MR images is not dependent on the amount of acquired slices. Our results, however, demonstrate that when multiple slice SL method is used, the CNR is highly dependent on the slice number (Figure 2). This is due to the non-selective nature, and the cumulative effect of the SL-pulses. Therefore, in multiple slice SL technique image contrast may be manipulated by choosing different amounts of slices.
Previously, we have demonstrated that the multiple slice SL technique generates good tissue contrast with liver tumour MRI [1, 8]. As shown by our present results, the multiple slice SL technique provides good tissue contrast properties also for normal tissues. The technique allows a large number of slices to be collected in a given TR, and shows reduced sensitivity to susceptibility and motion artefacts and excitatory B1 field inhomogeneities. Therefore, we conclude that with the choice of optimal imaging parameters the multiple slice SL technique is an attractive additional imaging sequence for abdominal MRI.
Received for publication December 23, 2002.
Revision received June 11, 2003.
Accepted for publication July 9, 2003.
|
References
|
|---|
- Halavaara JT, Sepponen RE, Lamminen AE, Vehmas T, Bondestam S. Spin lock and magnetization transfer MR imaging of focal liver lesions. Mag Res Imaging 1998;16:359–64.[CrossRef][Medline]
- Sepponen RE, Korhola OK, Lamminen AE, Tanttu JI. Multiple slice SL imaging—an effective method for T1 imaging. In: Book of Abstracts: Twelfth Annual Meeting of the Society of Magnetic Resonance in Medicine. New York, USA: SMRM, 1993:1218.
- Markkola AT, Aronen HJ, Paavonen T, Hopsu E, Sipilä LM, Tanttu JI, et al. T1 dispersion imaging of head and neck tumors: a comparison to spin lock and magnetization transfer techniques. J Magn Reson Imaging 1997;7:873–9.[Medline]
- Engelhardt RT, Johnson GA. T1 relaxation and its application to MR histology. Magn Res Med 1996;35:781–6.[CrossRef][Medline]
- Röschmann P. Radiofrequency penetration and absorption in the human body: Limitations to high-field whole-body nuclear magnetic resonance imaging. Med Phys 1987;14:922–31.[Medline]
- Sled JG, Pike GB. Standing-wave and RF penetration artifacts caused by elliptic geometry: an electrodynamic analysis of MRI. IEEE Trans Med Imag 1998;17:653–62.
- Tanttu J. Magnetic resonance imaging methods for improvement of tissue contrast especially applicaple at low magnetic fields. In: Thesis, Helsinki University of Technology Applied Electronics Laboratory, Series B: Research Reports B5, 1999.
- Halavaara JT, Lamminen AE, Bondestam S, Sepponen RE, Tanttu JI. Differentiation of hepatic hemangiomas and metastases with multiple slice spin lock MR imaging. Br J Radiol 1995;81:395–403.
- Bottomley PA, Foster TH, Argersinger RE, Pfeifer LM. Review of normal tissue hydrogen relaxation times and relaxation mechanisms from 1-100Hz: dependence on tissue type, NMR frequency, temperature, species, excision, and age. Med Phys 1984;11:425–48.[CrossRef][Medline]
- Sepponen RE. Rotating frame and magnetization transfer. In: Stark DD, Bradley WG Jr, editors. Magnetic resonance imaging, 2nd edn. St. Louis: Mosby-Year Boon, Inc. 1992:204–18.
- Dixon WT, Oshinski JN, Trudeau JD, Arnold BC, Pettigrew RI. Myocardial suppression in vivo by spin locking with composite pulses. Magn Res Med 1996;36:90–4.[Medline]
- Virta A, Komu M, Kormano M. T1 of protein solutions at very low fields: dependence on molecular weight, concentration, and structures. Magn Res Med 1997;37:53–7.[Medline]
- Zeisler R, Harrison SH, Wise SA. Trace elements in human liver using quality control in the complete analytical process. Biol Trace Ele Res 1984;6:31–49.
- Virta A, Komu M, Lundbom N, Jääskeläinen S, Kalimo H, Airio A, et al. Low field T1 imaging of myositis. Magn Reson Imaging 1998;16:385–91.[CrossRef][Medline]
- Markkola AM, Aronen HJ, Ramadan UA, Halavaara JT, Tanttu JI, Sepponen RE. Determination of T1 values for head and neck tissues at 0.1T: a comparison to T1 and T2 relaxation times. Magn Reson Imaging 1998;16:377–83.[CrossRef][Medline]
Top
Abstract
Introduction
