| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Beth Israel Deaconess Medical Center, 330 Brookline
Ave., Boston, MA 02215.
2 Department of Obstetrics and Gynecology, Beth Israel Deaconess Medical Center,
Boston, MA.
3 Department of Radiology, Children's Hospital, Boston, MA.
4 GE Healthcare, Menlo Park, CA.
Received August 8, 2005;
accepted after revision November 4, 2005.
This study was supported in part by National Institutes of Health (NIH)
grant NIBIB 01998.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Part 1 of the study was a prospective analysis using a questionnaire rating the ease of use and utility of RT imaging. Part 2 of the study was a retrospective analysis of 69 fetal MRI studies with RT sagittal midline imaging of the head, face, or both. Standard and RT SSFSE image sets were de-identified, randomized, and shown to three pediatric neuroradiologists who rated on a 5-point scale whether the images were midline and how well they could see and characterize as normal the corpus callosum and secondary palate. The imaging results were correlated with postnatal diagnosis. Statistical methods included the Wilcoxon's signed rank test, McNemar chi-square test, and analysis of variance.
RESULTS. Prospectively, the RT SSFSE technique was ranked as excellent in all the categories assessed. Retrospective analysis showed that the midline view obtained with RT SSFSE imaging was helpful in diagnosing the normal and abnormal secondary palate, allowing improved diagnosis of 19 (30.6%) of 62 cases of normal palate and four (57.1%) of seven cases of abnormal palate, when compared with the standard SSFSE technique. RT SSFSE imaging improved the ability to diagnose a normal corpus callosum on the midline view in 13 (27.6%) of 47 fetuses of 20 or more weeks gestational age.
CONCLUSION. The RT SSFSE technique can aid in obtaining images in planes that are critical to the evaluation of a moving fetus, particularly when a midline sagittal view of the corpus callosum or palate is required. The use of this technique may lead to improved diagnosis of CNS or orofacial abnormalities in fetuses.
Keywords: cleft palate • CNS • corpus callosum • fetal imaging • MRI • palate
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Real-time (RT) MRI with a T2-weighted SSFSE sequence (RT SSFSE) is a recent software advancement that allows rapid reconstruction of images and on-the-fly selection of plane of study [7]. In the RT mode, the position, orientation, field of view, and thickness of the tomographic section are controlled in an interactive fashion. A new SSFSE image can be obtained every 1-2 seconds, and the imaging parameters can be changed as needed. To accelerate T1 recovery, the transverse magnetization remaining at the end of the SSFSE readout is realigned with the main field. This "driven equilibrium" technique has been shown to improve T2 contrast at a given TR and TE [7]. Wiener demodulation can also be enabled to reduce blurring [7]. The Wiener filter operates on k-space data before reconstruction, boosting the signal of high-frequency phase-encoded lines to counter the reduction in signal caused by T2 decay. This diminishes the blurring typically observed in single-shot sequences; since the Wiener filter is constrained, noise is amplified less than with other inverse filters.
MRI has been found to be helpful in assessing the CNS and secondary palate in fetuses discovered to have cleft lip [8, 9]. In this study, we prospectively assessed the ease of use of RT imaging and then performed a retrospective analysis comparing the visibility of the fetal corpus callosum and soft palate on standard SSFSE versus RT SSFSE imaging. These structures were chosen because they are important for clinical diagnosis, and when present (and sufficiently developed in the case of the corpus callosum) should be visualized on a midline image.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Imaging Technique
The MR examinations were performed on an 8-channel 1.5-T superconducting
system (Signa Excite, GE Healthcare) using an 8-element phased-array surface
coil. SSFSE imaging was performed in the fetal sagittal, coronal, and axial
planes using the following parameters: TR/TE, single shot/60; field of view,
32 x 32 cm; matrix, 256 x 256; slice thickness, 4 mm; and 85
milliseconds per slice. These parameters were altered at the discretion of the
radiologist who selected the protocol for the examination. Sequences were
repeated as needed for diagnosis.
After all sequences needed for diagnosis had been performed, RT imaging was performed with the following parameters: TR/TE, single shot/90; field of view, 30 x 30 or 32 x 32 cm; matrix, 256 x 256; slice thickness, 4 mm; and 85 milliseconds per slice. These parameters were altered at the discretion of the radiologist monitoring the examination who selected the appropriate image plane and repeated sequences as needed.
The average number of standard sagittal SSFSE sequences was 1.8 ± 1.0 (SD) (range, 1-7 sequences), and the average number of sagittal RT SSFSE images was 7.8 ± 4.1 (range, 1-25 images). The most RT image acquisitions were required for fetuses being studied for potential cleft palate because repeated imaging was performed to view the oropharynx during swallowing. The time to perform sagittal midline RT SSFSE imaging was retrospectively assessed by evaluating the time signature on the images. The time spent on the midline sagittal view, on the scout images to obtain that view, and on both combined were calculated.
Part 1: Prospective Analysis
Part 1 of the study was a prospective analysis of the ease of use and
utility of RT imaging. The radiologist performing the MR examination completed
a questionnaire regarding his or her subjective impression of the following
categories: ease of use, image quality, suitability of RT parameters, image
contrast, orientation of scanning plane, and time to complete the study. A
7-point scale was used (1-2 = terrible, 3 = bad, 4 = neutral, 5 = good, 6-7 =
excellent), consistent with prototype evaluation standards used by the vendor
that developed the software.
This questionnaire was filled out about the first 20 patients undergoing RT imaging at our institution; these examinations were performed from June 11, 2003, through October 14, 2003.
Part 2: Retrospective Analysis
Patients—Part 2 of the study was a retrospective analysis of
consecutive fetal MRI studies that included RT imaging performed from June 16,
2003, through January 11, 2005. Exclusion criteria were fetuses with
indications for imaging that do not necessitate a midline view of the brain or
face.
The indication for examination and the prenatal imaging diagnoses were recorded. Gestational age was recorded as the age from last menstrual period if that was in agreement with the first-trimester sonogram or if it was within 2 weeks of the estimate based on the second-trimester sonogram. If gestational age from last menstrual period was determined to be inaccurate on the basis of an early sonogram, then age by the earliest sonogram available during the pregnancy was used to determine gestational age.
Cases of polyhydramnios (as assessed on sonography before MRI) were recorded.
Image selection—The number of standard SS-FSE sequences in the fetal sagittal plane and the number of RT SSFSE image acquisitions in the fetal sagittal plane were recorded. For each standard and RT SSFSE examination, one of the authors selected up to four images from each of the sequences that best showed midline sagittal anatomy. The standard SSFSE and RT image sets from the 69 MRI studies were then randomized into 138 cases.
Image review—The groups of randomized images were shown to three board-certified pediatric neuroradiologists for review. The reviewers were blinded to the indication for the study; the imaging technique used, RT or standard SSFSE; and gestational age. For each set of images, the reviewers rated how well they could see the corpus callosum and palate (1 = not at all, 2 = minimally, 3 = moderately, 4 = mostly, 5 = clearly and completely), indicated whether the images of the corpus callosum and palate were midline (1 = not at all, 2 = minimally, 3 = moderately, 4 = mostly, 5 = perfect), and assessed whether the corpus callosum and palate were normal (1 = definitely abnormal, 2 = probably abnormal, 3 = can't tell, 4 = probably normal, 5 = definitely normal).
Postnatal follow-up—Postnatal follow-up was physical examination (n = 54) or imaging (n = 28). Fifteen patients had no follow-up. Six patients either terminated or miscarried their pregnancy, and autopsy was not available for these cases. Nine patients were lost to follow-up. An abnormal palate was seen in seven fetuses and an abnormal corpus callosum, in six fetuses. In each of the abnormal cases except one, a case of an abnormal palate in a fetus with agnathia microstomia (the patient underwent a termination procedure without autopsy), postnatal confirmation of the prenatal diagnosis was available. In each of the other abnormal cases of corpus callosum dysgenesis and palatal abnormality, the postnatal diagnosis was concordant with the prenatal diagnosis. In 14 cases of normal corpus callosum and normal palate without a postnatal follow-up for comparison with the prenatal diagnosis (which was based on analysis of complete sonography and MRI examinations of fetuses), the prenatal normal diagnoses were used for statistical analysis.
Statistical Analysis
Wilcoxon's signed rank test was used to determine whether individual
reviewers' scores were higher for RT SSFSE imaging or for standard SSFSE
imaging. Wilcoxon's signed rank test was also used to determine whether the
median scores for RT SS-FSE imaging were higher than those for standard SS-FSE
imaging of fetuses with a normal palate (n = 62) and a normal corpus
callosum with a gestational age of 20 weeks or more (n = 47). This
20-week threshold for assessing the normal or abnormal corpus callosum was
used because it is expected to be neither fully formed nor completely
visualized on MRI before that gestational age
[10,
11].
The McNemar chi-square test (chi-square test for correlated proportions) was used to compare the proportion of RT SSFSE and standard SSFSE cases for which the median of the reviewers' scores appropriately classified the palate and corpus callosum as normal (score of 4 or 5) or abnormal (score of 1 or 2) and to appropriately classify the cases overall.
Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were assessed for the median reviewers' score for RT SSFSE imaging and then routine SSFSE imaging individually using a score of 1 or 2 (definitely to probably abnormal) compared with the final diagnosis as normal or abnormal. These performance measures were calculated for visualization of the secondary palate in all fetuses and of the corpus callosum in fetuses of 20 or more weeks gestational age.
Analysis of variance was performed to assess the impact of gestational age on the scores of individual reviewers for visualization of the normal palate, midline position for visualization of the palate, and ability to determine whether the palate was normal. Similar assessments were performed for the corpus callosum. The analysis of variance was also used to assess the relationship between gestational age and appropriate classification (i.e., as normal or abnormal) of palate and corpus callosum.
Interobserver kappa statistics were assessed: 0.40-0.59 indicated moderate agreement; 0.60-0.79, substantial agreement; and 0.80-0.89, near perfect agreement.
The percentage of cases for which scores of 1, 2, or 3 on SSFSE imaging increased to 4 or 5 on RT SSFSE imaging was graphed with respect to gestational age. The McNemar chi-square was used to compare fetuses of less than 20 weeks gestational age with those of 20 or more weeks gestational age for which scores of 1, 2, or 3 on SSFSE imaging increased to 4 or 5 on RT SSFSE imaging.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Part 2
Sixty-nine standard and RT SSFSE fetal MR examinations were performed in 66
patients. One patient had twins, each referred for evaluation of
ventriculomegaly (thus, two fetal examinations), and two patients were studied
twice. The indications for MR examination (each fetus and each fetal
examination counted separately) were ventriculomegaly (n = 39);
intraventricular hemorrhage (n = 3); posterior fossa abnormality
(n = 5); cleft lip or palate (n = 6); partial or complete
agenesis of the corpus callosum (n = 4); lymphatic malformation
(n = 2); arachnoid cyst (n = 2); and one case each of
micrognathia, absent mandible, macrocephaly, microcephaly, unexplained
polyhydramnios, midface anomaly, growth restriction with elevated
-fetoprotein level, and abnormal quadruple screen with increased risk
for trisomy 21.
The time spent on RT imaging in the midsagittal plane ranged from 16 to 122 seconds (mean, 43.5 ± 20.0 seconds). The time spent on the scout images for these midline images ranged from 13 to 36 seconds (mean, 23.5 ± 5.3 seconds). The total time ranged from 34 to 147 seconds (67 ± 22.1 seconds). These times were longer for the swallowing studies.
Gestational ages ranged from 17 to 37 weeks, with a mean of 25.0 ± 5.9 weeks. The six abnormal corpus callosum and seven abnormal palate cases are detailed in Tables 2 and 3, and representative images of these cases are shown in Figures 1A, 1B, 2A, 2B, 3A, and 3B.
|
|
|
|
|
|
|
|
The Wilcoxon's signed rank test showed that each reviewer's scores for normal palate on RT SSFSE images were significantly higher than for normal palate on standard SSFSE images (p = 0.004 to < 0.0001 for each individual test result). RT SSFSE image scores were higher also for visualizing the midline corpus callosum (p < 0.0001) and the region of the corpus callosum (p = 0.007 to < 0.0001). However, only two of the three reviewers had significantly higher scores for determining the corpus callosum was normal on RT SSFSE imaging than on standard SSFSE imaging (p = 0.001 and p < 0.0001), and one reviewer's scores showed no significant difference. Evaluating the median scores always showed a significant difference between RT SSFSE imaging and standard SSFSE imaging (p = 0.016 to < 0.0001).
Standard SSFSE imaging versus RT SSFSE imaging allowed appropriate classification of normal palate (graded as probably normal or normal) in 21 (33.9%) and 40 (64.5%) of 62 fetuses with normal palate, respectively (p = 0.001, Tables 4 and 5). There was no case in which the normal palate was graded as normal on standard SSFSE imaging and could not be adequately visualized on RT SSFSE imaging.
|
|
The abnormal palate (definitely abnormal or probably abnormal) was found in two (28.6%) and six (85.7%) of the seven abnormal cases (p = not significant) on standard and RT SSFSE imaging, respectively. The overall classification was appropriate in 23 (33.3%) and 46 (66.7%) of the 69 cases (p < 0.001) using standard and RT SSFSE imaging, respectively. These data yield a sensitivity of 28.6% and 85.7%, specificity of 98.4% and 100%, accuracy of 91.3% and 98.6%, positive predictive value of 66.7% and 100%, and negative predictive value of 92.4% and 98.4% for the standard and RT SSFSE techniques, respectively.
When using the RT SSFSE sequence, the ability to determine whether the palate was normal was significantly influenced by gestational age; in these cases, a median score of 3 was rendered at a mean gestational age of 22.1 ± 6.0 weeks and a median score of 4 or 5 was rendered at a mean gestational age of 26.5 weeks (p < 0.0001). In contrast, this relationship between gestational age and palate visualization was not found for routine SSFSE imaging; the mean gestational age for a median score of 3 on standard SSFSE images was 24.2 ± 6.1 weeks compared with 26.7 ± 5.7 weeks for a score of 4 or 5 (p = 0.16).
The cases misclassified on the basis of both imaging techniques were in fetuses at significantly lower gestational ages than those classified correctly using both techniques (p =0.02) and those classified correctly using RT SSFSE imaging and incorrectly using standard SSFSE imaging (p = 0.02). RT SSFSE imaging had significantly higher scores than standard SSFSE imaging for visualization of the normal palate (p < 0.0001), being midline (p < 0.0001), and being normal (p < 0.0001).
Standard versus RT SSFSE imaging allowed the appropriate classification of normal corpus callosum (graded as probably normal or normal) in 24 (51.1%) of 47 and in 36 (76.6%) of 47 fetuses of 20 or more weeks gestational age, respectively (p < 0.01). In this group of fetuses, standard and RT SSFSE imaging allowed identification of four (80%) of five cases of abnormal corpus callosum, leading to a sensitivity of 80%, specificity of 100%, accuracy of 98.1%, positive predictive value of 100%, and negative predictive value of 97.9% for both methods. Assessing only those fetuses with normal corpus callosum, the ability to determine the corpus callosum was normal differed significantly with respect to gestational age, with mean gestational age of 20.0 ± 2.6 weeks for a score of 3 versus a mean gestational age of 27.6 ± 5.2 weeks for a score of 4 or 5 using RT SSFSE imaging (p < 0.0001). This relationship between gestational age and corpus callosum visualization was also found for routine SSFSE imaging in which the mean gestational age for a score of 3 was 22.5 + 4.9 weeks compared with 28.1 ± 5.3 weeks for a score of 4 or 5 (p < 0.0001). In cases in which the findings of both techniques were incorrect, gestational age was lower than when findings of both techniques were correct (p = 0.001). RT SSFSE imaging had significantly higher scores than standard SSFSE imaging for visualization of the normal corpus callosum (p < 0.0001), being midline (p < 0.0001), and being normal (p = 0.02).
Analysis of variance for standard SSFSE imaging showed that for median scores, visualization of the region of the secondary palate, ability to determine whether the palate was normal, and midline image plane did not change with respect to gestational age. For RT SSFSE imaging, median data showed that a statistically significant impact of gestational age was seen for visualization of the region of the palate (p = 0.002), ability to determine whether the palate was normal (p = 0.02), and midline image plane (p = 0.002).
|
Weighted kappa statistics showed moderate to almost perfect agreement for the three reviewers' scores in all categories.
There were three cases of polyhydramnios. In each of these cases (fetuses at gestational ages of 26, 29, and 32 weeks), visualization of the palate and corpus callosum regions was improved with RT SSFSE imaging compared with standard imaging (scores of 1 increasing to 4 in two cases and a score of 2 increasing to 3 in one case).
Figure 4 shows a graph of the percentage of cases with respect to gestational age in which scores of 1, 2, or 3 on SSFSE imaging increased to 4 or 5 on RT SSFSE imaging. Sixteen fetuses of less than 20 weeks gestational age were examined. The McNemar chi-square analysis showed no statistically significant difference in scores for fetuses of less than 20 weeks gestational age and showed statistically significant differences (p < 0.01 and p < 0.001) for fetuses of 20 or more weeks gestational age in each of the categories assessed.
In each of the fetuses with abnormal corpus callosum and in each of the fetuses with abnormal palate, the correct diagnosis was made at the time of prenatal imaging.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In part 2 of the study, we found that the midline view obtained with RT SSFSE imaging was helpful in diagnosing the normal and abnormal secondary palate and allowed improved diagnosis in 19 (30.6%) of 62 fetuses with a normal palate and four (57.1%) of seven fetuses with an abnormal palate compared with standard SSFSE imaging. The use of RT sequences in these patients allowed multiple images to be obtained in the fetal sagittal plane for clear visualization of the oropharynx. The diagnosis of cleft palate was made prospectively in each case when the images were reviewed along with the patient's history, sonographic findings, and complete MR examination. The sagittal midline plane is important for assessing the soft palate because it can be viewed in the entire anterior to posterior extent (Fig. 2A). In our experience, coronal views can be angled and may therefore miss a small posterior defect. Axial views are difficult to interpret because the normal arch of the palate can mimic a cleft.
We also found that the midline view obtained with RT SSFSE imaging is helpful for visualizing the normal corpus callosum. Use of the RT SSFSE images improved the reviewers' ability to diagnose a normal corpus callosum on the basis of the midline view in 13 (27.6%) of 47 cases in fetuses of 20 or more weeks gestational age. This improved ability was significantly affected by gestational age. This finding is explained by the fact that, in fetuses at early gestational ages, the corpus callosum is relatively thin and that partial volume averaging limits our ability to distinguish it as normal, even on midline images.
In fetuses of less than 20 weeks gestational age, significantly less incremental benefit in visualization of the palate and corpus callosum on RT SSFSE imaging compared with standard SSFSE imaging was detected, as shown in Figure 4. This is not merely due to increased fetal motion at an early gestational age because visualization of midline anatomy was improved in fetuses with polyhydramnios (where motion frequently limits our ability to visualize anatomy).
In this study, we did not assess the accuracy of MRI in the diagnosis of cleft palate or of agenesis of the corpus callosum. Only sagittal images were reviewed. The fact that not all cases were appropriately diagnosed in this study is largely due to three elements of the study design. First, the blinded nature of the study required that the reviewers did not know patient history. Second, review of only sagittal midline images limited the reviewers' ability to check for other clues to the diagnosis; for example, in cases of agenesis of the corpus callosum, the ventricular contour and paramidline sulci are crucial in making the diagnosis, and in cases of cleft lip, review of axial and coronal images would show the cleft lip, thus leading to a more careful search of the palatal region. Third, the MR examinations were performed to answer a clinical question and thus were not always optimal for imaging the palate in fetuses with ventriculomegaly or the corpus callosum in fetuses with cleft lip.
The diagnosis of agenesis of the corpus callosum or of cleft secondary palate was made during the prospective clinical evaluation in each of the abnormal cases, thus illustrating the importance of reviewing images obtained in additional image planes and patient history for making the appropriate diagnosis. Our findings also emphasize that the midline view alone is not sufficient for diagnosis of agenesis of the corpus callosum, especially in the first half of the second trimester.
Devising a study in which standard SSFSE imaging and RT SSFSE imaging could be compared in an unbiased fashion was difficult. Our standard fetal MR protocol consists of series of images through the fetus in three planes orthogonal to fetal anatomy. Sequences are repeated if motion has degraded the image, but perfect midline views are not obtained of every fetus. When we use the RT SSFSE technique, we focus on a single region of anatomy and repeatedly image it until the appropriate view is obtained. For example, in fetuses with a risk for cleft palate, the image is typically repeated until the fetus swallows and fluid outlines the oropharynx. In this study, we assumed that obtaining images in specific planes will continue to be the manner in which RT SSFSE imaging is performed because that is how we typically use RT SSFSE imaging when scanning the fetal CNS and secondary palate.
There are three limitations to this study. The first limitation is that knowledge that RT SSFSE images were available could have decreased the willingness of the radiologist performing the examination to repeat standard SSFSE imaging when the standard SSFSE images were not sufficient for diagnosis. The second limitation is that there could have been unintentional bias in selection of the images for review. Every attempt was made to select the best sagittal midline images, regardless of imaging mode. The third limitation is that midline imaging was attempted many more times with the RT SSFSE sequence than with the standard SSFSE sequence. For example, it is likely that the diagnosis of cleft palate could have been made in the affected fetuses without RT SSFSE imaging if the standard SSFSE sequence had been repeated often enough. However, one of the benefits of RT SSFSE imaging is the speed of repeated image acquisition (one image every 1-2 seconds) that allows rapid change in image plane as the fetus moves and that can be used to assess swallowing.
The use of repeated sagittal midline MR images allows visualization of the tongue moving above the palatal shelf, even in fetuses with a cleft of only the soft palate. Because the lingual elevation can be intermittent, the use of multiple midline images is helpful. This allowed prospective diagnosis of clefts of the palate associated with cleft lip and for isolated cleft of the soft palate. Although not assessed in this study, the use of the RT SSFSE sequence for repeated imaging in a single plane has the potential to decrease examination time when a specific structure is being evaluated because it streamlines the series of events performed in standard SSFSE imaging—namely, obtaining a series of images, reconstructing those images, viewing the images, choosing an image plane, proscribing the slice orientation, prescanning, and then repeating the scan. The advantages of RT SSFSE imaging need to be viewed in the context of its increased requirement for physician or technologist involvement during the acquisition.
In summary, RT SSFSE imaging allows significantly improved plane positioning for the assessment of the normal and abnormal palate and the normal corpus callosum in the midline sagittal plane of the moving fetus. Use of this technique may lead to improved diagnosis in fetuses with CNS and orofacial abnormalities.
Acknowledgments
We thank Hugh Wheeler for his support with statistical analysis.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Limperopoulos, R. L. Robertson Jr., O. S. Khwaja, C. D. Robson, J. A. Estroff, C. Barnewolt, D. Levine, D. Morash, L. Nemes, L. Zaccagnini, et al. How Accurately Does Current Fetal Imaging Identify Posterior Fossa Anomalies? Am. J. Roentgenol., June 1, 2008; 190(6): 1637 - 1643. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
