British Journal of Radiology (2006) 79, 873-879
© 2006 British Institute of Radiology
doi: 10.1259/bjr/36989440
The development and optimization of high spatial resolution MRI for imaging the oesophagus using an external surface coil
A M Riddell, BSc, FRCS, FRCR
C Richardson, DCRr
E Scurr, BSc, DCRr
and
G Brown, MD, MRCP, FRCR
Department of Radiology, Royal Marsden Hospital, Fulham Road, London SW3 6JJ, UK
 |
Abstract
|
|---|
This paper describes the development and optimization of an innovative technique using an external surface coil to obtain high resolution, thin section MR images of the oesophagus using volunteers. T2 weighted fast spin echo sequences were performed with and without cardiac gating. The field of view (FOV), matrix size, slice thickness, number of signal averages (NSA), and repetition time (TR)/echo time (TE) were altered to optimize signal to noise ratio (SNR) whilst maintaining spatial resolution. The effect of cardiac gating was also investigated. Workstation images were evaluated on the ability to visualize: individual oesophageal wall layers; perioesophageal fat; the azygos vein and wall of the descending aorta, giving qualitative assessment of image clarity. The optimum sequence enabled the layers of the oesophageal wall and perioesophageal tissues to be demonstrated in an acceptable scan time of 7.07 min. A FOV of less than 250 mm degraded image quality so that individual oesophageal wall layers could not be depicted and noise within the image impaired visualization of posterior mediastinal structures. The results indicate that high resolution imaging of the oesophagus using an external surface coil can depict anatomic structures clearly and that the use of cardiac gating improves image clarity. The technique offers an alternative, non-invasive method of detailed imaging of the oesophagus.
|
Introduction
|
|---|
Endoscopic ultrasound is currently used for local staging of oesophageal cancer, but has technical limitations, with a failure rate in up to 17–24% of patients due to stenotic tumours preventing the passage of the endoscope [1, 2]. It is also operator dependent, with a recognized learning curve [3]. The high frequency probe (12 MHz) used to achieve high spatial resolution, has a limited sonographic range and, as a consequence, the relationship of the oesophagus to the surrounding perioesophageal tissues can be difficult to visualize with clarity.
There have been few studies that have evaluated MRI of the oesophagus. Early in vivo studies used an integrated body coil and a low field strength magnet (0.35–0.5 T). As a consequence, the images generated had a low in-plane spatial resolution [4, 5]. More recent studies have concentrated on the use of an endoluminal surface coil attached to the tip of a modified endoscope. Preliminary results have been encouraging, with confirmation of the ability to depict the layers of the oesophageal wall [6–8]. However, the technique has recognized limitations, namely: the inability of the coil to traverse strictures, limiting the use for the evaluation of oesophageal cancers; a short radius for receiving signal (3–4 cm), which necessitates repositioning of the endoscope for evaluation of long tumours therefore increasing the overall scan time; and the action of peristalsis, resulting in motion artefact within the images acquired and possible coil migration.
The only study to date using a 1.5 T magnet and an external surface coil examined oesophageal specimens in vitro [9]. The study concluded that individual layers of the oesophageal wall were more clearly depicted on T2 weighted rather than T1 weighted images.
The aim of this study was to use healthy volunteers to develop the technique of high resolution thin slice MRI, using an external surface coil in vivo, as a non-invasive method of imaging the oesophagus and perioesophageal tissues; and to both describe the methods used to optimize the signal to noise ratio (SNR) and to assess the effect of cardiac gating.
|
Methods and materials
|
|---|
Four volunteers were imaged using a 1.5 T magnet (Philips Intera, software version 9.5.2, Andover, MA) and an external 5 channel surface coil (Philips Sense CardiacTM). The study evaluated images of the lower oesophagus, as clinically the majority of patients at our institution who are considered for surgical resection are those with adenocarcinoma of the lower oesophagus and gastro-oesophageal junction.
The study objectives were to:
- Maximize spatial resolution by alterations to the feld of view (FOV), matrix size and slice thickness, whilst maintaining acceptable image quality.
- Maximize SNR by alterations to the echotime (TE) and number of signal averages (NSA), whilst maintaining acceptable image quality and scan time.
- Minimize the effects of cardiac motion, by the use of a saturation band placed over the heart and also the use of cardiac gating.
- Maintain an acceptable scan time. The maximum limit for an individual sequence was set at 7 min 30 s for 22 axial images.
T2 weighted sequences were used as it had previously been established that these are optimum for delineating the layers of the oesophageal wall [9, 10]. A sagittal sequence was performed initially to localize the oesophagus and to ensure optimum coil placement. Axial images were acquired perpendicular to the long axis of the oesophagus.
A standard fast spin echo T2 weighted sequence used for pelvic imaging [11] was modified and used as a baseline sequence for the studies without cardiac gating, and the baseline parameters used when cardiac gating was employed were modified from a standard cardiac T2 weighted sequence, with a reduction in the FOV. In both instances, the baseline sequence was given a relative SNR of 1. Changes in the SNR due to alterations in sequence parameters were expressed as a ratio from this baseline.
Assessment of image quality
For each of the study objectives, the images acquired using specific sequence parameters were viewed on a workstation (eFilm workstationTM Version 1.9.3) and were scored objectively by one radiologist (AMR) as good, moderate or poor, on the basis of the ability to visualize individual layers of the oesophageal wall, the perioesophageal fat, the azygos vein and wall of the descending aorta. This provided a qualitative evaluation of image clarity. Using the same parameters, the effect of cardiac gating on image quality was assessed by comparing equivalent images with and without cardiac gating.
|
Results
|
|---|
The sagittal T2 weighted sequence provided valuable information to optimize coil placement. Figure 1
illustrates the boundary of acceptable signal and marked signal drop off outside this region. The individual layers of the oesophageal wall could be depicted clearly on the axial high resolution T2 weighted images. The mucosa returned intermediate to low signal. This was surrounded by high signal intensity submucosa and the outer low signal intensity muscularis propria, as shown in Figure 2. The perioesophageal fat returned high signal. The volume of fat varied between subjects and for those subjects with less fat, the clarity of the oesophageal wall layers and perioesophageal fat was graded as moderate, compared with a grading of good for subjects with a greater volume of perioesophageal fat.
View larger version (181K):
[in this window]
[in a new window]
|
Figure 1. The margins of acceptable signal to noise, confined within the white lines. A saturation band is placed over the heart(X) to reduce artefact from cardiac motion. The oesophagus is clearly seen anterior to the vertebral column (white arrows).
|
|
View larger version (145K):
[in this window]
[in a new window]
|
Figure 2. The main oesophageal wall layers: low signal mucosa (white arrow), surrounded by high signal submucosa (black arrow) and low signal muscularis propria (arrowheads). The descending thoracic aorta (A) and vertebral body (V) are marked.
|
|
Maximizing spatial resolution
For the T2 weighted sequence without cardiac gating a relative SNR, when compared with the initial baseline sequence parameters, of less than 0.6 resulted in unacceptable image quality. The highest spatial resolution achieved with an acceptable SNR was obtained using a 256 mmx256 mm matrix, with 240 mm FOV, 3 mm slice thickness and 6 NSA. These parameters gave an in-plane resolution of 0.94 mmx0.94 mm and a voxel size of 2.63 mm3. A matrix size of 512 mmx256 mm produced higher resolution images, but unacceptable noise within the image. Image quality was also significantly impaired by reducing the FOV to less than 240 mm. The optimum sequence parameters are shown in Table 1.
Using cardiac gating, the highest achievable spatial resolution with acceptable signal to noise was achieved by employing a 250 mm FOV, a 301 mmx512 mm matrix and a slice thickness of 4 mm, giving an in-plane spatial resolution of a 0.5 mmx0.8 mm and a voxel size of 1.62 mm3, as shown in Table 1
.
Maximizing SNR whilst maintaining spatial resolution
The optimized sequence parameters, without cardiac gating, achieved from the first study were used as the baseline for technique refinement. A decrease in TE resulted in an increase in the relative signal to noise (Figure 3), whist maintaining the T2 weighting. As a consequence, by utilizing a shorter TE, a smaller field of view could be used to improve spatial resolution. Increasing the NSA improved the SNR (Figure 4); an NSA of 6 was considered optimum as the scan time was increased to 9.28 min for 22 slices with an NSA of 8, exceeding the maximum limit. The optimum sequence parameters are shown in Table 1.
View larger version (67K):
[in this window]
[in a new window]
|
Figure 3. Images at the same level of the oesophagus in one volunteer. The echo time(TE) is reduced from (a) 120 ms to (b) 80 ms. All other parameters are maintained (matrix 256 mmx256 mm, repetition time (TR) 5446 ms, field of view (FOV) 225 mm, number of signal averages (NSA) 6, turbo spin echo (TSE) 16). The signal is increased, improving the conspicuity of the oesophagus (white arrow), the wall of the aorta (arrowhead) and the perioesophageal fat (double arrow heads) with the shorter TE.
|
|
View larger version (70K):
[in this window]
[in a new window]
|
Figure 4. Images at the same level of the oesophagus in one volunteer. The number of signal averages(NSA) is increased from (a) 4 to (b) 8. All other parameters are maintained (matrix 256 mmx256 mm, repetition time (TR) 5446 ms, field of view (FOV) 225 mm, echo time (TE) 80 ms, turbo spin echo (TSE) 16). The higher NSA improves image quality by improving conspicuity of the oesophageal wall layers: the high signal submucosa (white arrow) and lower signal muscularis propria (arrowhead); the perioesophageal tissues (black arrow) and wall of the aorta.
|
|
Minimizing the effects of cardiac motion
The images acquired with a saturation band alone remained subject to motion artefact, primarily due to cardiac pulsation. Image quality was superior for the sequence with cardiac gating, when directly compared with the equivalent image without cardiac gating (Figure 5). Specifically, the clarity of the perioesophageal tissues and wall of the descending thoracic aorta was improved with the use of cardiac gating, with the majority of images with cardiac gating being scored as good quality. Depiction of the individual oesophageal wall layers was also superior with cardiac gating, although the improvement was not as marked.
View larger version (81K):
[in this window]
[in a new window]
|
Figure 5. Images taken at the same level of the oesophagus in one volunteer, using the optimized high resolutionT2 weighted sequence (a) without and (b) with cardiac gating. The sequence with cardiac gating provides improved image quality. This is illustrated by the clarity of the oesophageal wall submucosa (black arrow), the aortic wall (white arrow) and right pleural reflection (fine black arrow).
|
|
Initially, the RF pulse was set to trigger on every third R wave (3 beats), giving an effective TR of 2250 ms. This gave adequate T2 weighting to the image. Increasing the effective TR to 4000 ms (6 beats) resulted in a reduction in image quality, in particular the clarity of the perioesophageal tissues (Figure 6).
View larger version (93K):
[in this window]
[in a new window]
|
Figure 6. Images taken at the same level of the oesophagus in one volunteer. Increasing the repetition time(TR) from (a) 3 beats (effective TR 2250 ms) to (b) 6 beats (effective TR 4000 ms) caused blurring within the image. The oesophageal wall submucosa (black arrowhead), muscularis propria (black arrow), and right pleural reflection (white arrow) are more clearly seen with an effective TR of 2250 ms (3 beats).
|
|
When the T2 weighted sequence without cardiac gating was modified to utilize the sequence parameters optimized with cardiac gating, there was a 37% drop in relative signal to noise and the scan time increased from 7.07 min to 10.0 min for 22 slices, exceeding the acceptable scan time limit.
Scan duration
Using the optimized sequence parameters without cardiac gating, an acceptable scan time of 7.07 min for 22 slices was achieved. The use of cardiac gating did not significantly increase the scan duration. The scan time using cardiac gating varied slightly between volunteers due to differences in resting heart rate, but for the optimized sequence parameters did not exceed the scan time limit.
|
Discussion
|
|---|
High resolution thin slice MRI of the oesophagus using an external surface coil is potentially challenging due to the relatively low signal to noise returned from the structure due to its small size and location within the thorax. This study shows that the individual layers of the oesophageal wall and surrounding anatomical structures can be delineated using the technique. The signal returned from the individual wall layers in this study is in accordance with the description given by Yamada et al [9, 10]. Endoscopic ultrasound can also delineate the layers of the oesophageal wall. However, the ability of the technique to delineate the perioesophageal tissues is limited by the relatively short ultrasound range of the probe (in the order of 3–4 cm). The axial images acquired of the posterior mediastinum obtained using high resolution MRI clearly delineate the relationship of the oesophagus to all its surrounding anatomical structures, providing improved detail regarding potential resection margins for those patients with oesophageal carcinoma who are being considered for surgery. The variation in the amount of perioesophageal fat between subjects resulted in a reduction in image clarity in subjects with a small amount of perioesophageal fat, which could be considered a limitation of the technique.
The non-invasive nature of the technique will allow optimum visualization of stenotic tumours, not amenable to endoscopic evaluation. In many instances these bulky tumours are not considered surgically resectable. However, a technique which allows for detailed initial evaluation of the local extent of tumour will provide the most accurate method of local staging of the tumour. It would also provide a basis for alternative treatment planning and a baseline for future imaging to fully assess treatment response.
The perioesophageal tissues are visualized with CT, but recent studies have shown that the ability of the technique to determine resectability is relatively low. Markland et al reported a sensitivity of between 0% and 66.7% [12] and a further study of 51 patients showed that 59% had more advanced disease than was appreciated on pre-operative CT, reinforcing the need to refine the techniques for the local staging of oesophageal carcinoma to improve patient selection for radical therapy [13]. The superior contrast resolution achieved with MRI would suggest that the technique is likely to be useful in differentiating tumour abutting surrounding structures (stage T3) from direct invasion of tumour into the surrounding tissues (stage T4).
Spatial resolution
Previous attempts to achieve high resolution, thin slice imaging of the oesophagus have been limited to in vitro studies of resected oesophageal specimens from patients with known oesophageal carcinoma [9, 10]. The studies used a 4 cm loop surface coil and a 1.5 T or 4.5 T magnet, with FOV of between 40 mm and 60 mm, and as such were able to achieve a voxel size of 0.08–0.16 mm3. Given this high resolution, 6–8 layers of the oesophageal wall were visualized and the technique was able to identify tumour which was confined to the mucosa. Prior in vivo studies using surface coils used low field strength magnets (0.35–0.5 T) and acquired images with a slice thickness of 8–10 mm, to achieve adequate signal to noise. As a consequence, these studies concluded that MRI was no better than CT at providing detailed analysis of the oesophageal wall and surrounding anatomical structures [4, 5]. Investigations using an endoluminal coil have achieved an in plane resolution of 0.469 mmx0.625 mm (voxel size 1.17 mm3), enabling visualization of the three main mural layers (mucosa, submucosa and muscularis) [14]. Our study shows that using fast spin echo sequences, high resolution images can be acquired with an external surface coil, with a voxel size of 1.62 mm3. At this resolution the main component layers of the oesophageal wall were readily delineated, indicating that it would be possible to use the technique for the local staging of oesophageal tumours.
Signal to noise
The benefit of using an endoluminal surface coil is that the coil is closer to the region of interest, thus improving signal to noise. As a consequence, the spatial resolution can be maximized. There is, however, rapid signal decay beyond a 3–4 cm radius of the endoluminal coil and therefore the technique has limited ability to assess the perioesophageal tissues and determine tumour resectability. Previous MRI studies with an external coil used a body coil integrated into the bore of the magnet. The introduction of multi-channel external surface phased array coils, as used in our study, has improved the achievable SNR, by the use of parallel imaging techniques. The technology allows for faster sequences enabling higher resolution imaging to be acquired with increased signal to noise.
Cardiac gating
Several of the previous studies using endoluminal and external surface coils have used cardiac gating, but no previous study has directly compared the effect of using the technique on image quality. The acquisition of imaging data when using cardiac gating is synchronized with and limited to specific times or phases of the cardiac cycle. The R wave on the electrocardiogram (ECG) is used to trigger each radiofrequency (RF) pulse. The RF signal is therefore generated at the same point of subsequent cardiac cycles. Since the R to R interval is controlled by the patient's heart rate – the TR (and as a consequence the image weighting and number of slices) depends on the heart rate. To enable variation in the TR value, the system can be adapted to trigger an RF pulse at every second or third R wave; in this way the "effective TR" is lengthened, allowing for T2 weighted imaging. Our study would indicate that the use of cardiac gating does improve image quality.
Scan time
Our experience is that, in general, patients are able to lie still for an individual sequence which lasts for between 7 min and 7
min. Our study used 7 min as the maximum acceptable scan time for 22 axial images. Although image quality was improved by increasing the NSA from 6 to 8, for the T2 weighted sequence without cardiac gating, this exceeded our upper time limit. These scan times compare favourably with endoluminal MRI studies, which have documented scan times of between 7 min and 10 min for 8–12 images. No previous study has specifically assessed the affect of cardiac gating on scan time; these results show that its use does not significantly increase the overall scan duration.
We conclude that our study indicates that the technique of high resolution, thin slice MRI using and external surface coil is a reproducible, non-invasive method of imaging the oesophagus, which could potentially provide an alternative method for the local staging of oesophageal carcinoma.
Received for publication February 9, 2006.
Revision received March 27, 2006.
Accepted for publication April 24, 2006.
|
References
|
|---|
- Wakelin SJ, Deans C, Crofts TJ, Allan PL, Plevris JN, Paterson-Brown S. A comparison of computerised tomography, laparoscopic ultrasound and endoscopic ultrasound in the preoperative staging of oesophago-gastric carcinoma. Eur J Radiol 2002;41:161–7.[CrossRef][Medline]
- Preston SR, Clark GW, Martin IG, Ling HM, Harris KM. Effect of endoscopic ultrasonography on the management of 100 consecutive patients with oesophageal and junctional carcinoma. Br J Surg 2003;90:1220–4.[CrossRef][Medline]
- Schlick T, Heintz A, Junginger T. The examiner's learning effect and its influence on the quality of endoscopic ultrasonography in carcinoma of the esophagus and gastric cardia. Surg Endosc 1999;13:894–8.[CrossRef][Medline]
- Quint LE, Glazer GM, Orringer MB. Esophageal imaging by MR and CT: study of normal anatomy and neoplasms. Radiology 1985;156:727–31.[Abstract/Free Full Text]
- Lehr L, Rupp N, Siewert JR. Assessment of resectability of esophageal cancer by computed tomography and magnetic resonance imaging. Surgery 1988;103:344–50.[Medline]
- Inui K, Nakazawa S, Yoshino J, Yamao K, Yamachika H, Wakabayashi T, et al. Endoscopic MRI: preliminary results of a new technique for visualization and staging of gastrointestinal tumors. Endoscopy 1995;27:480–5.[Medline]
- Kulling D, Feldman DR, Kay CL, Hoffman BJ, Reed CE, Young JW, et al. Local staging of esophageal cancer using endoscopic magnetic resonance imaging: prospective comparison with endoscopic ultrasound. Endoscopy 1998;30:745–9.[Medline]
- Dave UR, Williams AD, Wilson JA, Amin Z, Gilderdale DJ, Larkman DJ, et al. Esophageal cancer staging with endoscopic MR imaging: pilot study. Radiology 2004;230:281–6.[Abstract/Free Full Text]
- Yamada I, Izumi Y, Kawano T, Yoshino N, Tetsumura A, Ohashi K, et al. Superficial esophageal carcinoma: an in vitro study of high-resolution MR imaging at 1.5T. J Magn Reson Imaging 2001;13:225–31.[CrossRef][Medline]
- Yamada I, Murata Y, Izumi Y, Kawano T, Endo M, Kuroiwa T, et al. Staging of esophageal carcinoma in vitro with 4.7-T MR imaging. Radiology 1997;204:521–6.[Abstract/Free Full Text]
- Brown G, Daniels IR, Richardson C, Revell P, Peppercorn D, Bourne M. Techniques and trouble-shooting in high spatial resolution thin slice MRI for rectal cancer. Br J Radiol 2005;78:245–51.[Abstract/Free Full Text]
- Markland CG, Manhire A, Davies P, Beggs D, Morgan WE, Salama FD. The role of computed tomography in assessing the operability of oesophageal carcinoma. Eur J Cardiothorac Surg 1989;3:33–6.[Abstract]
- Sariego J, Mosher S, Byrd M, Matsumoto T, Kerstein M. Prediction of outcome in "respectable" esophageal carcinoma. J Surg Oncol 1993;54:223–5.[Medline]
- Feldman DR, Kulling DP, Hawes RH, Kay CL, Muckenfuss VR, Cotton PB, et al. MR endoscopy: preliminary experience in human trials. Radiology 1997;202:868–70.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
BJR review of the year - 2006
Br. J. Radiol.,
March 1, 2007;
80(951):
147 - 151.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
