British Journal of Radiology (2005) 78, 299-302
© 2005 British Institute of Radiology
doi: 10.1259/bjr/23825228
A comparison of fast MRI of hips with and without parallel imaging using SENSE
M Ryan, FFR RCSI1,
P Cunningham, FFR RCSI1,
C Cantwell, AFRCSI1,
D Brennan, FRCR1 and
S Eustace, FRCR2
1 Cappagh National Orthopaedic Hospital, Finglas, Dublin 11, Ireland and 2 Mater Misericordiae Hospital, Eccles Street, Dublin 7, Ireland
Correspondence: Dr Stephen Eustace, Department of Radiology, Cappagh National Orthopaedic Hospital, Finglas, Dublin 11, Ireland
 |
Abstract
|
|---|
This study was undertaken to qualitatively and quantitatively compare fast MRI of hips with and without parallel imaging using SENSE (sensitivity encoding). 27 patients underwent MRI of the hips with coronal T1 turbo spin echo (TSE) (repetition time (TR) 500 ms, effective echo time (TEeff) 15 ms, Turbo Factor 4), coronal IR-TSE (TR 2000 ms, TEeff 30 ms, inversion time (TI) 160 ms, Turbo Factor 20) and axial T2 TSE (TR 3000 ms, TEeff 80 ms, Turbo Factor 20) weighted images acquired with and without SENSE with a reduction factor of 2. Conventional imaging was performed in 8 min and 36 s. Images acquired with SENSE were acquired in 5 min and 31 s without a discernible reduction in image quality or a significant quantitative reduction in image signal to noise ratio, contrast to noise ratio or edge enhancement.
|
Introduction
|
|---|
Previous authors have described the use of parallel imaging techniques, SENSE (Sensitivity Encoding MRI) and SMASH (Simultaneous Acquisition of Spatial Harmonics) to accelerate MR image acquisition. In this role, parallel imaging techniques have been employed to facilitate breath hold techniques in the abdomen, cine imaging of the heart and bolus tracking at MR angiography [1–4]. To date its application to facilitate rapid imaging of the extremities has been employed in a single study evaluating the role of SMASH to allow fast imaging of the shoulder [5]. To date, no publication has qualitatively and quantitatively assessed the use of SENSE in allowing fast imaging of the extremities [6, 7].
This study was undertaken to compare the quality and diagnostic accuracy of images of the hips acquired by fast MRI sequences with and without SENSE (parallel imaging).
|
Subjects and methods
|
|---|
Patients
27 consecutive patients referred for MRI of the hips were included in the study.
Imaging
All patients underwent imaging on a 1.5 T Philips Intera system (Philips Medical Systems, Best, The Netherlands) using a Synergy body coil and a 40 cm field of view (FOV). Each patient was imaged with coronal T1 weighted turbo spin echo (TSE) (repetition time (TR) 500 ms, effective echo time (TEeff) 15 ms, Turbo Factor 4), coronal inversion recovery (IR)-TSE (TR 2000 ms, TEeff 30 ms, inversion time (TI) 160 ms, Turbo Factor 20) and axial T2 weighted TSE (TR 3000 ms, TEeff 80 ms, Turbo Factor 20) sequences with and without SENSE employed with a reduction factor of two. The same contrast parameters and number of excitations were used for both sets of sequences. A 256 x 256 matrix was used to generate both sets of images. Slice number and thickness (3 mm) was identical in both image sets.
Image interpretation
Qualitative
In each case images were read independently blinded to the imaging technique, with discrepancies when present resolved by consensus. In each case, evaluation was made of the acetabular labrum, articular cartilage and bone marrow of the femoral head. Image quality was scored as poor=1, satisfactory=2, good=3 in each case.
Quantitative
Subsequently each image was evaluated on an EasyVision workstation (Philips Medical Systems) using region of interest mapping to determine image signal to noise ratio, contrast to noise ratio (the difference between signal to noise ratios at two sites), and edge enhancement of bone marrow, fat and muscle (based on the rate of change in signal intensity through a straight line on acquired images).
In each case regions of interest were selected by two reviewers who used side by side monitor evaluation of conventional images and SENSE images to ensure that identical regions of interest were selected on both sets of images.
|
Results
|
|---|
Acquisition time
Coronal IR TSE, coronal T1 weighted TSE and axial T2 weighted TSE images of the hips were acquired in 8 min 36 s in total. When SENSE was added with a reduction factor of 2, the total imaging time was reduced to 5 min 31 s.
Side by side image interpretation
Recorded image quality of conventional images and of those with SENSE was identical in 27 of 27 patients.
Findings included eight normal studies, nine patients had osteoarthritis of the hips, three patients had femoral head avascular necrosis, six patients had acetabular dysplasia, and one patient had a fracture of the femoral neck. No perceived difference in image quality of images acquired with and without sense was documented using a 3 point scoring system.
Assessment of marrow was recorded as good (3) by both readers in all cases. Assessment of the acetabular labrum was recorded as good (3) in all cases with the exception of two patients with osteoarthritis and two patients with acetabular dysplasia where the labrum evaluation was considered satisfactory (2) on both sets of images by both readers. Assessment of acetabular cartilage was considered good (3) in nine patients (eight normals, one patient post fracture), and satisfactory (2) in remaining cases by both readers.
Workstation image evaluation
Quantitative evaluation of images was on the basis of review of T1 weighted images with and without SENSE. Quantitative assessment of coronal IR TSE and axial TSE T2 weighted images was not undertaken as the effect of SENSE on image quality and acquisition time is the same for all sequences if all other parameters (matrix, slice thickness and field of view) remain unchanged.
In 27 patients, the recorded mean signal to noise of fat was 1080/88 without sense compared with a mean of 984/81 when imaged with SENSE (p<0.5).
In 27 patients, the recorded mean signal to noise of muscle was 366/88 without sense, compared with 290/81 when imaged with SENSE (p<0.5).
The contrast to noise yielded by comparing the mean signal of fat and muscle relative to the mean signal of noise was 740/88 without sense, compared with 632/81 when imaged with SENSE (p<0.5).
No significant difference in edge enhancement was documented on the images with and without SENSE.
View larger version (118K):
[in this window]
[in a new window]
|
Figure 1. 50-year-old male with right hip pain. (a) Coronal turbo spin echo (TSE) T1 weighted image without sensitivity encoding (SENSE), (repetition time (TR) 500 ms, effective echo time (TEeff) 15 ms, turbo factor (TF) 4, low high filling of K space) showing pattern of signal change through a line drawn through both hips (left prosthesis) with associated graph depicting signal intensity change per millimetre. (b) Coronal TSE T1 weighted image with SENSE, (TR 500 ms, TEeff 15 ms, TF 4, low high filling of K space) showing almost identical pattern of signal change through a line drawn through both hips (left prosthesis) with associated graph depicting almost identical signal intensity change per millimetre.
|
|
View larger version (128K):
[in this window]
[in a new window]
|
Figure 2. 32-year-old female with bilateral hip pain. (a) Coronal turbo spin echo (TSE) T1 weighted image without sensitivity encoding (SENSE), (repetition time (TR) 500 ms, effective echo time (TEeff) 15 ms, turbo factor (TF) 4, low high filling of K space) showing 250 pixel regions of interest yielding noise (14), muscle signal (15), and fat signal (16) with associated histogram indicating range and mean signal intensity in fat (1277). (b) Coronal TSE T1 weighted image with SENSE, (TR 500 ms, TEeff 15 ms, TF 4, low high filling of K space) showing 250 pixel regions of interest yielding noise (18), muscle signal (19), and fat signal (20) with associated histogram indicating range and mean signal intensity in fat (1266).
|
|
|
Discussion
|
|---|
The last decade has seen the widespread introduction of fast imaging sequences supported by the tandem development of flexible localizing gradients. Such developments have allowed dramatic reduction in acquisition times paving the way for innumerable new clinical applications previously vulnerable to motion, including diffusion imaging in the brain, gated cardiac techniques and abdominal applications such as magnetic resonance cholangiopanceatography (MRCP) [6]. Although not primarily benefiting from motion reduction, fast techniques have allowed dramatic increases in patient throughput in the extremities without a reduction in image quality. Recognizing that there is a limit to potential reduction in acquisition times afforded by fast sequences and optimized gradients alone, this report reviews the potential of parallel imaging techniques to produce further acquisition time reductions without significant impact on the diagnostic quality of the acquired images. The paper specifically reviews the impact of image reduction times on contrast, edge enhancement, image signal and diagnostic effectiveness of acquired images.
Parallel imaging techniques employ phased array surface coils and allow reconstruction algorithms to unwrap or unalias useful information and signal acquired in neighbouring coils outside the field of view of individual coil elements [6, 7]. In such a way fewer phase encode steps are required to fill image K space within each coil element, as potential deficiencies are filled by unwrapped information to generate a high spatial resolution image. Intuitively this process should result in both a reduction in image acquisition time and also a reduction in image signal. However, the relative reduction in image signal versus acquisition time and relative impact of SENSE acquired images on diagnostic accuracy when employed to image the extremities has not been previously assessed.
In this report, we employed optimized turbo spin echo sequences to allow imaging of both hips with a total acquisition time of 8 min 36 s compared with 5 min 30 s when the same parameters were employed with SENSE. Although one might predict that the application of SENSE with a reduction factor of 2 would result in a direct halving of acquisition time, in practice, the implementation of SENSE requires the acquisition of a sensitivity map prior to image acquisition at a cost of 50 s. Further reductions in acquisition time are likely to be generated by the development of more complex phased array surface coils with a further increase in the number of individual coil elements. The necessity to acquire a sensitivity map to localize signal prior to imaging also increases the possibility of misregistration if there is motion between mapping and signal acquisition. Since the SENSE technique involves a reduction in phase encoding steps, which is compensated by the addition of unwrapped information, one would anticipate a significant reduction in image signal. This reduction is determined as a factor of the reduction factor. Therefore a reduction factor of two as employed in this study should theoretically lead to a drop off in signal to noise of 1.4 and so on. Such an expected reduction in image signal was not encountered in this study. Such a result may reflect the reduction in scan acquisition time. The increased acquisition speed using SENSE results in less time for proton dephasing resulting in preservation of image signal. The SNR also depends on slice setup and individual coil design. The under sampling of K space by increasing the distance of sampling positions while maintaining the maximum K values will allow spatial resolution to be maintained. In this study no significant reduction in image signal, contrast or edge enhancement was recorded. Owing to the entirely independent physics of the encoding effect, SENSE does not interfere with any of the numerous known contrast mechanisms or acquisition schemes. The SENSE approach can therefore be applied to practically all known MRI techniques [6].
In summary, the results of this study suggest that parallel imaging techniques particularly SENSE can be successfully employed to image the extremities. The use of SENSE allows a reduction in acquisition times without a significant reduction in subjective image quality or objective workstation assessment of image signal, contrast or edge enhancement.
Received for publication February 12, 2004.
Revision received July 6, 2004.
Accepted for publication November 9, 2004.
|
References
|
|---|
- Sodickson DK, Griswold MA, Jacob PM. SMASH imaging. Magn Reson Imaging Clin N Am 1999;7:237–54.[Medline]
- Sodickson DK, McKenzie CA, Ohlinger MA, et al. Recent advances in image reconstruction, coil sensitivity, calibration, and coil array design for SMASH and generalized parallel MRI. MAGMA 2002;13:158–63.
- Sodickson DK, McKenzie CA. A generalized approach to parallel magnetic resonance imaging. Med Phys 2001;28:1629–43.[CrossRef][Medline]
- Bydder M, Larkman DJ, Hajnal JV. Generalised SMASH imaging. Magn Res Med 2002;47:160–70.[CrossRef][Medline]
- Magee T, Shapiro M, Williams D, Richard Ramnath R, Simon J. Usefulness of the simultaneous acquisition of spatial harmonics technique during MRI of the shoulder. AJR Am J Roentgenol 2003;181:961–4.[Abstract/Free Full Text]
- Van den Brink JS, Watanabe Y, Kuhl CK, et al. Implications of SENSE MR in routine clinical practice. Eur J Radiol 2003;46:3–27.[CrossRef][Medline]
- Willinek WA, Gieseke J, von Falkenhausen M, Neuen B, Schild HH, Kuhl CK. Sensitivity encoding for fast MR imaging of the brain in patients with stroke. Radiology 2003;228:669–75.[Abstract/Free Full Text]
Top
Abstract
Introduction
