European Radiology

, Volume 16, Issue 3, pp 634–641

MRI of the pelvis at 3 T: very high spatial resolution with sensitivity encoding and flip-angle sweep technique in clinically acceptable scan time

Authors

    • Department of RadiologyUniversity of Bonn
  • Jürgen Gieseke
    • Department of RadiologyUniversity of Bonn
    • Philips Medical Systems
  • Christiane Kuhl
    • Department of RadiologyUniversity of Bonn
  • Götz Lutterbey
    • Department of RadiologyUniversity of Bonn
  • Marcus von Falkenhausen
    • Department of RadiologyUniversity of Bonn
  • Frank Träber
    • Department of RadiologyUniversity of Bonn
  • Tjoung-Won Park-Simon
    • Department of Gynecology and ObstetricsUniversity of Bonn
  • Oliver Zivanovic
    • Department of Gynecology and ObstetricsUniversity of Bonn
  • Hans H. Schild
    • Department of RadiologyUniversity of Bonn
Urogenital

DOI: 10.1007/s00330-005-0016-1

Cite this article as:
Morakkabati-Spitz, N., Gieseke, J., Kuhl, C. et al. Eur Radiol (2006) 16: 634. doi:10.1007/s00330-005-0016-1

Abstract

Objective

The higher signal at 3.0-T allows spatial resolution to be increased without loss in image quality. We evaluated a T2-weighted turbo spin-echo sequence with high spatial resolution (3T-HR) to determine whether this provides clinically useful pelvic MRI.

Materials and methods

We designed a sequence with high spatial resolution (3T-HR) (0.45×0.46×4 mm) that was combined with parallel imaging and the variable refocusing angle technique (8.06 min). We examined 23 patients with gynecological disorders using 3T-HR and a standard sequence (3T-SP; 4.03 min; equivalent to 1.5 T). Two radiologists analyzed tissue contrast, signal to noise, detail delineation and artifact level.

Results

Tissue contrasts and signal to noise were rated equal. Motion artifacts occurred more often with 3T-SP despite the longer scanning time of 3T-HR. The higher spatial resolution provided additional information in four patients. In two patients small myomas were detected, in one patient a lymph node metastasis was apparent, and in one patient 3T-HR excluded tumor invasion.

Conclusions

High spatial resolution pelvic studies with high image quality can be obtained at 3 T in acceptable scan time. The higher spatial resolution that is feasible at 3 T also provides more clinically relevant information.

Keywords

High-field magnetic resonance imagingSpatial resolutionParallel imagingVariable refocusing angleSpecific absorption rate reduction

Introduction

Magnetic resonance (MR) imaging plays a major role in staging of gynecological tumors, in the further workup of infertility, and in the differential diagnosis of ovarian disorders. In general the assessment of gynecological disorders requires MR pulse sequences with high spatial resolution. The now available high-field MR systems offer high intrinsic signal thus allowing a further increase in spatial resolution. For neurological applications it has been demonstrated that the higher spatial resolution improves detail delineation and thus diagnostic accuracy [15]. Whole-body applications are now being evaluated at 3 T [615]. There have been only few reports in the literature dealing with whole-body MR applications at 3 T with high spatial resolution [11]. The main challenge for the pulse sequence design at higher magnetic field is the increased radiofrequency energy deposition, which requires new strategies to reduce the specific absorption rate (SAR) [16]. Further technical challenges at 3 T include the increased T1 relaxation time [17, 18], decreased T2 relaxation time [19, 20], and insufficient radiofrequency power penetration [21], stronger susceptibility effects, and a larger chemical shift.

The purpose of this study was to evaluate the diagnostic potential of a new T2-weighted turbo spin-echo (TSE) pulse sequence with very high spatial resolution (3T-HR) for MR imaging of the female pelvis.

Materials and methods

Study design

Between August 2003 and September 2004 we performed a prospective intraindividual study of women referred for MR imaging of the pelvis. We compared our new T2-weighted TSE pulse sequence with very high spatial resolution (3T-HR) and a 3.0 T standard sequence (3T-SP) which has shown to be equivalent to 1.5 T [16]. The two T2-weighted TSE pulse sequences were acquired in a randomized order. The study design was approved by our institutional review board, and all 23 patients gave informed consent to be examined after the nature of the procedure had been fully explained to them.

Patients

We included 23 consecutive patients (mean age 46±13 years, range 15–60; Table 1). The patients had been referred with the diagnosis of ovarian cancer (n=4), myoma (n=8), carcinoma of the cervix (n=7), uterine malformation (n=1), indeterminate high-frequency ultrasound of the pelvis (n=1), pelvic pain (n=1), and pelvic metastasis (n=1). Two of the patients had a normal pelvic MRI. Hysterectomy had been performed in two patients.
Table 1

Referral and final MR diagnoses

Referral diagnosis

Final MR diagnosis

Ovarian cancer (n=4)

Ovarian cancer (n=4)

Uterine myomas (n=8)

Uterine myomas (n=8)

Cancer of the cervix (n=7)

Cancer of the cervix (n=7)

Uterine malformation (n=1)

Uterine malformation (n=1)

Pelvic metastasis (n=1)

Pelvic metastasis (n=1)

Unclear abdominal pain (n=1)

Normal pelvic MRI study (n=1)

Indeterminate pelvic US (n=1)

Normal pelvic MRI study (n=1)

MR imaging technique

Studies were performed on a clinical 3.0 T MR scanner (Intera 3.0T, Philips Medical Systems, Best, The Netherlands; maximal gradient amplitude 30 mT, slew rate 150 T m−1 s−1) equipped with a transmit-receive quadrature body coil. Pelvic imaging was performed with a six-element cardiac-synergy-coil. To reduce peristalsis n-butyl-scopolamine (20 mg) was given intravenously to all patients prior to the study. In addition, a regional saturation technique was placed on the anterior abdominal wall to minimize ghosting artifacts. The new T2-weighted pulse sequence was designed to have significantly higher spatial resolution (at least threefold increase) than 3T-SP. To reduce the energy deposition 3T-HR was combined with parallel imaging (sensitivity encoding, SENSE) and a new variable refocusing angle technique [2225] called flip-angle sweep (FAS). This technique uses RF pulses with lower power deposition. In detail, flip angles are reduced as a sweep along the echotrain starting with high flip angles resulting in a reduction in the RF power. The variable refocusing angle technique produces a higher amount of stimulated echoes than techniques using 180° refocusing pulses only. To collect a considerable amount of stimulated echoes we increased TR, thus increasing the signal and yielding adequate T2 weighting. To compensate for a potential signal loss caused by SENSE the number of signals averaged (NSA) was increased at 3T-HR. 3T-HR differed from 3T-SP in the following aspects (see Table 2): The spatial resolution of 3T HR was 3.6-fold higher while the scan time was only doubled (8.06 min instead of 4.03 min). 3T-SP was combined with the CLEAR technique (homogeneity correction technique) [14]. SENSE (SF 3) and FAS (130°) were combined with 3T-HR, NSA was four (instead of one1), TR was 3958 ms (instead of 2705 ms), TE was 70 ms (instead of 80 ms), and TF was 18 (instead of 25).
Table 2

Imaging parameters of 3T-SP sequence as compared to the 3T-HR sequence

 

3T-SP

3T-HR

Acquisition time

4.03 min

8.06 min

Measured voxel size (mm3)

0.7/1.07/4

0.45/0.46/4

Sensitivity encoding

No

Yes (SF 3)

Specific absorption rate (W/kg)

3.4

4

Bandwidth per pixel (Hz)

2.091 (207.8)

2.364 (183.8)

TR/TE (ms)

2705/80

3958/70

Slices axial

29 (4 mm)

29 (4 mm)

Number of signals averaged

1

4

Turbo factor

25

18

Half scan

Yes (0.625)

Yes (0.725)

Matrix

512×512

800×1024

Field of view

360 mm

360 mm

Scan percentage

80

100

Image analysis

First we evaluated the signal-to-noise relationship. Because images were obtained with SENSE and CLEAR [19, 14], we performed a mere qualitative analysis of signal-to-noise. Two radiologists blinded to the type of sequence performed a direct visual comparison (in consensus) of 3T-HR and 3T-SP images evaluating apparent signal and image noise. Images of 3T-HR and 3T-SP were rated on a three-point scale with 3 points for superior signal-to-noise at 3T-HR, 2 for equal signal-to-noise, and 1 for superior signal-to-noise with 3T-SP.

Tissue contrasts were analyzed qualitatively regarding anatomical structures such as the zonal anatomy of the uterus (except in patients with hysterectomy and in the case of tumor invasion into the uterus) and gynecological disorders. Again, a three-point scale was used as described above. In addition, region of interest based quantitative measurements were performed to assess different tissue contrasts (C) according to the following relation: C=(A−B)/(A+B), where A represents the signal of tissue A and B the signal of tissue B. We analyzed the contrast between muscle and solid tumors, between muscle and cystic tumors, between muscle and urine, between the junctional zone and the myometrium of the uterus (zonal uterine anatomy), and between ovarian cysts and ovarian stroma. The degree of artifacts due to ghosting of the abdominal wall and peristalsis was analyzed in consensus by two radiologists using a five-point scale. 3T-HR and the corresponding 3T-SP images were evaluated separately. We assigned 1 point if no artifacts were present, 2 for minor artifacts, 3 for moderate (not diagnostically relevant) artifacts, 4 for stronger artifacts (diagnostically relevant), and 5 for severe artifacts (nondiagnostic study).

Delineation and detectability of image details were evaluated with respect to the visualization of small anatomical details (fine septae in the fat tissue, small vessels, if present small ovarian cysts and the delineation of the zonal anatomy of the uterus). Furthermore, the detectability of gynecological disorders was evaluated. We evaluated how often the higher spatial resolution provided additional information compared to the standard sequence. We evaluated tumor borders and signs of infiltration, detectability of lymph nodes, and detectability of small benign and malignant tumors. Two radiologists compared the MR images of 3T-HR and 3T-SP directly (consensus) using a three-point scale (as described above). We compared the final MR imaging diagnoses that were obtained with the standard pulse sequence (3T-SP) and the high spatial resolution pulse sequence (3T-HR) using the regular MR imaging criteria for gynecological disorders, for example, the International Federation of Gynecology and Obstetrics (FIGO) classification for tumor staging and the American Fertility Society classification of Müllerian duct anomalies in the case of uterine malformation.

Statistical analysis

For statistical analysis the SPSS software package (SPSS, Chicago, Ill., USA) was used to calculate mean values and standard deviations for contrast measurements. To test for statistical significance we used Wilcoxon's paired test for the quantitative analysis of tissue contrasts. Differences at a P level of 0.05 were considered significant. The marginal homogeneity test was used to test for statistically significant differences in artifact levels between 3T-HR and 3T-SP. Again, a P level of 0.05 was considered significant.

Results

Concerning the visual signal to noise the qualitative image analysis did not reveal a significant difference between 3T-SP and 3T-HR. We could not analyze tissue contrast of the zonal anatomy in 7 of the 23 patients. Two of these seven had had hysterectomy, and in five the zonal anatomy was destroyed by tumor invasion. In all the remaining 16 patients the qualitative analysis revealed comparable tissue contrast regarding the delineation of the zonal uterine anatomy at 3T-SP and 3T-HR. Tissue contrasts of the various gynecological disorders (n=21; see Table 2) were also comparable at 3T-HR and 3T-SP.

Concerning quantitative contrast analysis (Table 3) Wilcoxon's paired test revealed no statistically significant differences between 3T-SP and 3T-HR. With regard to the amount of artifacts only minor to moderate artifacts caused by motion of the abdominal wall were observed both with 3T-HR (1.74±0.54; range 1–3) and 3T-SP (1.96±0.64; range 1–3). In 18 of the 23 studies artifacts were rated equal with 3T-HR and 3T-SP, and in 5 studies the degree of artifacts was even rated worse with 3T-SP. This difference (Fig. 1) was statistically significant (marginal homogeneity test: P=0.025). In two of the five pelvic MR studies artifacts were upgraded from absent with 3T-HR to minor with 3T-SP; in three studies artifacts were upgraded from minor with 3T-HR to moderate with 3T-SP. With intravenous n-butyl-scopolamine artifacts due to peristalsis did not occur. Chemical shift or dielectric artifacts were not observed in this study.
Table 3

Mean contrast between various tissues at 3T-SP and 3T-HR. Wilcoxon's paired test was used to test for statistical significance

Mean tissue contrast

3T-SP

3T-HR

P

Muscle-solid tumors

0.34±0.19

0.33±0.19

0.79

Muscle-cystic tumors

0.55±0.11

0.57±0.15

0.71

Muscle-bladder

0.62±0.15

0.58±0.21

0.83

Junctional zone-myometrium

0.29±0.18

0.27±0.18

0.52

Ovarian cysts-ovarian stroma

0.26±0.13

0.28±0.11

0.96

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-005-0016-1/MediaObjects/330_2005_16_Fig1_HTML.gif
Fig. 1

Comparison of motion artifact level with 3T-HR and 3T-SP

The delineation and detectability of small anatomical details such as fine septae in the fat tissue, small ovarian cysts or small vessels was rated superior with 3T-HR in all pelvic MR studies (Fig. 2). The detectability of gynecological disorders was also evaluated. MR imaging diagnoses differed in 4 of the 23 patients in whom the higher spatial resolution provided additional clinically relevant information. This added value of 3T-HR was not based on a significant lower artifact level at 3T-HR. In two of the four patients with multiple myomas small additional myomas were detected with 3T-HR (Fig. 3). Both studies showed identical artifact levels at 3T-HR and 3T-SP (minor or moderate artifacts for the same patient). In one of the four patients with cervical carcinoma of FIGO stage IIB a small lymph node metastasis was clearly visible on 3T-HR (Fig. 4) which had not been diagnosed with 3T-SP. In this patient the artifact level was rated moderate with 3T-SP and minor with 3T-HR. In one of the four patients with a pelvic metastasis (breast cancer) invading the bladder wall 3T-HR (Fig. 5) excluded tumor invasion into the vagina due to the presence of a small fat layer between the metastasis and the bladder wall. In this pelvic study motion artifacts were absent both for 3T-HR and 3T-SP. In the remaining 19 patients the delineation and detectability of the gynecological disorders was rated equal; thus MR imaging diagnoses did not differ.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-005-0016-1/MediaObjects/330_2005_16_Fig2_HTML.jpg
Fig. 2

Delineation and detectability of small anatomic details with 3T-SP (a) and 3T-HR (b). Small anatomic details such as fine septae in the fatty tissue are much better visualized with the very high spatial resolution

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-005-0016-1/MediaObjects/330_2005_16_Fig3_HTML.jpg
Fig. 3

Patient with multiple uterine myomas. The detection of a small additional intramural myoma was possible with 3T-HR (long arrow) whereas it could not be visualized with 3T-SP

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-005-0016-1/MediaObjects/330_2005_16_Fig4_HTML.jpg
Fig. 4

Patient with carcinoma of the cervix FIGO IIB. 3T-HR allows the detection of a small lmyph node metastasis (long white arrow) which had not been diagnosed prospectively with 3T-SP. Please note that tissue contrasts are comparable

https://static-content.springer.com/image/art%3A10.1007%2Fs00330-005-0016-1/MediaObjects/330_2005_16_Fig5_HTML.jpg
Fig. 5

Patient with a pelvic metastasis of breast cancer. Bladder wall invasion was diagnosed with 3T-HR and 3T-SP. 3T-HR allowed to delineate the wall of the vagina and thus to exclude tumor infiltration into the vagina whereas this could not be excluded with certainty at 3T-SP (long white arrow)

Discussion

MR imaging is gaining increasing importance for the work-up of female pelvic disorders [2629]. In the field of gynecological applications MR pulse sequences with high spatial resolution are required to assess tumor margins correctly and to detect small gynecological disorders. With the available field strength at 1.5 T a further increase in spatial resolution is limited by an unacceptable increase in scan time or decrease in signal-to-noise ratio, thus impairing image quality. Due to the high intrinsic signal the new high field MR systems generally enable MR pulse sequences with very high spatial resolution [1].

This study compared a new 3-T T2-weighted TSE sequence (3T-HR) with very high spatial resolution to a 3-T standard T2-weighted TSE pulse sequence (3T-SP) in women undergoing MR imaging of the pelvis due to various gynecological disorders. We wanted to answer the following questions: Does the increased signal at 3.0T allow a further increase in spatial resolution of a TSE pulse sequence for MR imaging of the female pelvis? Does this sequence still allow clinically useful pelvic MR imaging? Is scan time acceptable for patients? Does this pulse sequence maintain the diagnostic image quality of a 3-T standard pulse sequence with respect to the signal to noise, the level of artifacts, and the familiar tissue contrasts of anatomical structures and of various gynecological disorders? Does the further increased spatial resolution of this new TSE sequence improve the visualization of anatomical details and of small pathologies and thus lead to more clinically diagnostic relevant information?

Concerning the first question, it must be stated that in principle the higher signal does allow a further increase in the spatial resolution. However, it should also be noted that several technical modifications must be considered because SAR limits are reached earlier at high magnetic field. Thus, increasing spatial resolution of a TSE pulse sequence (and thus the number of 180° RF pulses resulting in high energy deposition) limits the possible echo train length (ETL) in order to avoid heating. As a consequence scan time increases. Therefore strategies to reduce SAR levels and scan time are needed. As described above, parallel imaging (SENSE) reduces the energy deposition and reduces ETL and thus energy deposition, scan time, and the amount of blurring artifacts. We did not use a higher sense factor because SENSE also results in a further signal decrease and may lead to infolding artifacts thus impairing image quality. To compensate for a potential signal loss caused by SENSE the NSA and TR were increased.

Because energy deposition and scan time were still too high, we also used a new variable refocusing angle technique [25]. The use of an even smaller FAS value than 130° might have reduced scan time further, but it must be considered that the FAS technique also alters image contrast [2225]. The consideration of applying only a moderate reduction in the refocusing angle was based on prior measurements [30] in which FAS 130° was the threshold which maintained the familiar tissue contrasts at given contrast parameters and still allowed very high spatial resolution. Because experience with this technical feature was limited at the time that this study was undertaken, we applied this moderate FAS reduction. In summary, the higher magnetic field allows a further increase in spatial resolution with the above technical modifications.

Concerning the question of whether the new high-resolution pulse sequence also allows clinically useful pelvic MR imaging, it must first be noted that although the spatial resolution was increased 3.6-fold, scan time was only doubled compared to the standard sequence. We did not further increase the spatial resolution of the 3T-HR sequence because experience shows that longer acquisition times may be associated with impaired image quality due to the increasing risk of patient movement. As our data show, scan time still seems to be acceptable from this point of view. We asked whether the high resolution sequence (3T-HR) maintains diagnostic image quality with regard to signal to noise, level of artifacts, and familiar tissue contrasts compared to the standard of reference (3T-SP). As our data show, the high spatial resolution TSE sequence was technically successful in all 23 patients. The visual signal to noise was comparable for 3T-SP and 3T-HR. Despite the elongated scan time at higher spatial resolution we did not encounter stronger motion artifacts with 3T-HR. On the contrary, motion artifacts were significantly stronger with 3T-SP, although still minor to moderate. This observation may be explained by the increased NSA with 3T-HR which were able to average motion artifacts. Further artifacts such as chemical shift, blurring, and dielectric artifacts were observed neither with 3T-HR nor with 3T-SP. The next issue is whether 3T-HR provides familiar tissue contrasts, which is the fundamental basis for the detection and correct classification of gynecological disorders. As our data show, tissue contrasts of both anatomical structures and pelvic pathologies were comparable with 3T-HR and 3T-SP. In summary, the high spatial resolution sequence does maintain diagnostic image quality.

With regard to the diagnostic potential of the new pulse sequence we asked whether the higher spatial resolution improves the visualization of anatomical details and the detection and delineation of small pathologies, and if so, whether the additional information is also clinically relevant. As our data show, the higher spatial resolution provides significantly better delineation of small anatomical details in all patients.

All benign (n=9) and malignant (n=12) pelvic disorders were detected with both 3T-SP and 3T-HR, but the higher spatial resolution provided additional clinically relevant information in 4 of the 21 patients with gynecological disorders. Although in this clinical setting the detection of additional small myomas in two patients of the four did not have a direct impact on patient treatment, it has been demonstrated that the high spatial resolution is able to detect small pathologies. On the other hand, in two patients with malignant pelvic disorders the higher spatial resolution significantly improved tumor staging with the diagnosis of a lymph node metastasis in one patient with carcinoma of the cervix of FIGO stage IIB and exclusion of tumor invasion into the vagina in another patient with a pelvic metastasis of breast cancer invading the bladder. In summary, very high spatial resolution pelvic MR studies with high image quality are feasible in acceptable scan time on a 3 T scanner. Our preliminary data suggest that the higher spatial resolution also seems to be clinically advantageous for MR imaging of the pelvis providing additional relevant information.

To our knowledge, this is the first report on very high spatial resolution MR imaging of the female pelvis with 3 T. Our study has the following limitations. Because data analysis was performed in consensus we cannot provide data on interobserver variability. We are aware of the rather small size of our patient cohort. Although in this clinical setting 3T-HR did not alter FIGO staging of malignant tumors or substantially alter patients' treatment, our data are still promising. It is likely that in a larger group of patients the additional clinically relevant information provided by the higher spatial resolution would also significantly improve diagnostic confidence and diagnostic accuracy and thus influence patients' management and treatment. This is being evaluated in ongoing studies. Furthermore we are not able to provide confirmation by surgery or histology in the four patients with added value. However, we want to point out that in the general clinical setting, MR imaging is well accepted as method of choice and standard of reference in tumor staging and is the basis for the choice of therapy. Concerning myomas, this diagnosis is a mere MR imaging diagnosis and does not require histological confirmation.

In conclusion, high spatial resolution pelvic studies with high image quality can be obtained with a 3T scanner in acceptable scan time. Our preliminary data suggest that the higher spatial resolution at 3 T leads to more clinically relevant information.

Copyright information

© Springer-Verlag 2005