VIBE MRI for Evaluating the Normal and Abnormal Gastrointestinal Tract in Fetuses

Transcription

1 Inaoka et al. MRI of Fetal Gastrointestinal Tract Women s Imaging Clinical Observations 12_07_2063_Inaoka.fm 11/9/07 WOMEN S IMAGING Tsutomu Inaoka 1 Hiroyuki Sugimori 1 Yoshihito Sasaki 2 Koji Takahashi 1 Kazuo Sengoku 2 Nobuhisa Takada 1 Tamio Aburano 1 Inaoka T, Sugimori H, Sasaki Y, et al. Keywords: fetus, gastrointestinal tract, meconium, MR colonography, MRI, obstetrics, pediatrics, volume interpolated breath-hold imaging, women s imaging DOI: /AJR Received February 17, 2007; accepted after revision June 24, Department of Radiology, Asahikawa Medical College, Midorigaoka-Higashi, Asahikawa City, Hokkaido, , Japan. Address correspondence to T. Inaoka (tinaoka@asahikawa-med.ac.jp). 2 Department of Obstetrics and Gynecology, Asahikawa Medical College, Asahikawa City, Hokkaido, Japan. WEB This is a Web exclusive article. AJR 2007; 189:W303 W X/07/1896 W303 American Roentgen Ray Society VIBE MRI for Evaluating the Normal and Abnormal Gastrointestinal Tract in Fetuses OBJECTIVE. The great potential of MRI for assessing gastrointestinal abnormalities in fetuses has been described. T1-weighted images may add additional information to T2-weighted images in diagnosing fetal gastrointestinal abnormalities. The objective of this study was to assess the performance of a 3D volumetric interpolated breath-hold sequence (VIBE) in evaluating the normal and abnormal fetal gastrointestinal tract. CONCLUSION. VIBE provides high-quality T1-weighted and 3D MR colonography images for the evaluation of the normal and abnormal gastrointestinal tract in fetuses, and 3D MR colonography provides excellent delineation of the meconium. renatal diagnosis of fetal gastrointestinal tract malformations, P including bowel dilatation, polyhydramnios, hyperechoic bowel, and ascites, is commonly made with sonographic findings. However, these findings are not specific and may relate to transient normal variants [1 3]. In addition, sonography has some disadvantages, such as operator dependence, low image contrast, and a small field of view [1 3]. Therefore, it is occasionally challenging for obstetric sonographers to assess fetal gastrointestinal abnormalities when accurate recognition of the bowel condition is required to determine fetal and neonatal management [1, 2]. The usefulness of fast MRI, including single-shot fast spin-echo sequences, for evaluating fetal abdominal disorders has been reported with increasing frequency. The great potential of MRI for the assessment of the fetal gastrointestinal tract has been described; in diagnosing abnormal cases, it is important to assess whether meconium is present in the fetal bowel [2 5]. MRI is more sensitive than sonography for detecting the presence of meconium [2]. The location and amounts of meconium in the fetal bowel depend on gestational age; however, the accumulation of meconium may steadily advance from the anal canal in a normal fetus after 20 weeks gestation [2 6]. Meconium exhibits an intermediate or low signal intensity on T2-weighted images and a high signal intensity on T1-weighted images because of its high protein and mineral content [2 4]. T1-weighted images may add additional information to T2-weighted images in diagnosing fetal gastrointestinal abnormalities because meconium is more apparent on T1-weighted than on T2-weighted images [5]. Three-dimensional MR images generated from T1-weighted images, which were obtained using 2D or 3D fast gradient-echo MR sequences, are useful for 3D understanding in the diagnosis and monitoring of fetal gastrointestinal tract malformations [4, 7]. For conventional T1-weighted sequences, a 4-mm or greater section thickness is required to perform imaging through the fetus during a limited acquisition time [2 7]. However, additional thin sections of 3 mm or less are needed to better characterize abnormal findings and to make 3D MR images of sufficient quality because a mean diameter of the intestine in fetuses after 20 weeks gestation is 3 mm or greater [4, 6, 7]. Conventional sequences are still limited in temporal resolution for obtaining T1-weighted images of such thin sections with adequate anatomic coverage for evaluating fetal gastrointestinal abnormalities. Recently, a 3D volumetric interpolated breath-hold sequence (VIBE), which is a modified fast 3D gradient-echo sequence, has been applied to T1-weighted images in clinical body MRI [8 10]. This sequence may provide isotropic or nearly isotropic resolution in three dimensions while preserving wide anatomic coverage in a short acquisition time [8 10]. Motion artifacts are also reduced by the rapid data acquisition [10]. Therefore, 3D MR images of high qual- AJR:189, December 2007 W303

2 Inaoka et al. TABLE 1: Normal Gastrointestinal Tract Features Having High Signal Intensity in Fetuses 19 Weeks 4 Days to 40 Weeks 5 Days Gestational Age Small Intestine Colon Fetal Age and Imaging Type Stomach Duodenum Proximal Distal Ascending Transverse Descending Sigmoid Rectum Overall (n = 28) T2-weighted a 28 (100) 28 (100) 23 (82.1) 19 (67.9) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) T1-weighted b 0 (0) 0 (0) 11 (39.3) 15 (53.6) 26 (92.9) 28 (100) 28 (100) 28 (100) 28 (100) MR colonography c 0 (0) 0 (0) 9 (32.1) 12 (42.9) 26 (92.9) 28 (100) 28 (100) 28 (100) 28 (100) < 32 weeks gestation (n = 12) T2-weighted a 12 (100) 12 (100) 7 (58.3) 6 (50) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) T1-weighted b 0 (0) 0 (0) 5 (41.7) 8 (66.7) 12 (100) 12 (100) 12 (100) 12 (100) 12 (100) MR colonography c 0 (0) 0 (0) 3 (25) 5 (41.7) 12 (100) 12 (100) 12 (100) 12 (100) 12 (100) 32 weeks gestation (n = 16) T2-weighted a 16 (100) 16 (100) 16 (100) 13 (81.3) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) T1-weighted b 0 (0) 0 (0) 6 (37.5) 7 (43.8) 14 (87.5) 16 (100) 16 (100) 16 (100) 16 (100) MR colonography c 0 (0) 0 (0) 6 (37.5) 7 (43.8) 14 (87.5) 16 (100) 16 (100) 16 (100) 16 (100) Note Data are number (%) of patients. a HASTE. b Three-dimensional volumetric interpolated breath-hold fast gradient-echo. c Three-dimensional. ity can be available because of the resultant minimization of partial volume averaging effects and motion artifacts [8, 10]. We attempted to obtain thin-section T1-weighted images through the fetus using VIBE and to generate 3D MR colonography images of the fetus from the VIBE data sets. The purpose of this study was to assess the performance of VIBE in evaluating the normal and abnormal fetal gastrointestinal tract. Materials and Methods Subjects Between June 2004 and September 2006, 45 patients underwent fetal MRI after sonography in our institute. Fetal MRI was limited to cases in which complex anomalies were suspected at second- or third-trimester fetal sonography screening or in which the sonography results were equivocal or undetermined. Before the examination, all subjects were informed about the procedure and the safety of the technique by obstetricians, and all provided their informed consent. Of the 45 participants, 35 who fulfilled the following inclusion criteria were enrolled in this retrospective study. The first criterion was that HASTE T2-weighted and VIBE T1-weighted images were both obtained through the fetus. The second criterion was that diagnoses were confirmed after delivery in our institute. Although there was one twin pregnancy, we examined only the indicated fetus. On the basis of the medical records of postnatal diagnoses, 28 of the 35 fetuses had a normal gastrointestinal tract. The gestational ages of the fetuses with a normal gastrointestinal tract ranged from 19 weeks 4 days to 40 weeks 5 days (mean, 30 weeks 5 days). These fetuses had other abnormalities: CNS abnormalities (n = 10), urogenital abnormalities (n = 7), musculoskeletal abnormalities (n = 6), and hydrops fetalis (n = 1). Four of those fetuses had normal findings. The remaining seven fetuses were confirmed as having gastrointestinal abnormalities, which included duodenal atresia (n = 3), congenital diaphragmatic hernia (n =2), gastroschisis (n = 1), and ileal atresia with meconium peritonitis (n =1). MRI Protocol MRI was performed on a 1.5-T MR unit (Magnetom Sonata, Siemens Medical Solutions) using a phased-array body coil. In the fetal MR examinations, HASTE T2-weighted sequences (TR/TE eff, 1,100/81; flip angle, 150 ; field of view, 400 mm; slice thickness, 5.0 mm; section gap, 1.0 mm; slice number, 18; image matrix, 256; bandwidth, 476 Hz/pixel; turbo factor, 164; 1 signal acquisition; scanning time, 20 seconds) were obtained through the fetus in the transverse, sagittal, and coronal planes. T1-weighted images were obtained through the fetus in the coronal plane using VIBE (TR/TE, 4.45/1.34; flip angle, 15 ; field of view, 400 mm; slice thickness, 3 mm; section gap, mm; number of slices, 32 40; image matrix, 256 [interpolation, 512]; bandwidth, 490 Hz/pixel; 1 signal acquisition; scanning time, seconds). Although the images were basically obtained during a single breath-hold, these were sometimes acquired during quiet respiration. Neither sedatives nor IV gadolinium-based contrast material was used. The pregnant patients were instructed to walk for 10 minutes before the MR examination. MR Image Processing and Evaluation We generated 3D MR colonography images from the VIBE data sets using a volume-rendering algorithm at a computer workstation (Aquarius, version 3.2, Elk Corporation). T2-weighted images, T1-weighted images, and 3D MR colonography images were simultaneously assessed on an image viewer. In the fetuses with a normal gastrointestinal tract, signal characteristics of the stomach, the duodenum, the proximal small intestine, the distal small intestine, the ascending colon, the transverse colon, the descending colon, the sigmoid colon, and the rectum were assessed. The small intestine was subdivided into the duodenum, the proximal small intestine, and the distal small intestine because a distinction between the jejunum and the ileum was not feasible. The identification of the respective intestine was made on the basis of the anatomic location, the continuity, and the presence or absence of haustra. The liver was used as a landmark of the anatomic position in the fetal abdomen. Signal intensity (high or low) of the precise intestinal segments was subjectively determined on both T2- and T1-weighted images. Particularly on T1- weighted images, we compared the signal intensity with that of muscle because the T1-weighted signal in the small and large intestine is greater than that of muscle [4]. On 3D MR colonography images, visualization of the intestine was judged W304 AJR:189, December 2007

3 MRI of Fetal Gastrointestinal Tract TABLE 2: Fetuses with Gastrointestinal Abnormalities Patient No. Gestational Age Diagnosis T1-Weighted VIBE and 3D MR Colonography Findings 1 28 wk 6 d Gastroschisis No intestinal complications (volvulus, atresia, perforation) 2 29 wk 5 d Duodenal atresia No other intestinal complications 3 30 wk 2 d Ileal atresia, meconium peritonitis Ileal atresia 4 33 wk 0 d Left diaphragmatic hernia Colon involved, liver not involved 5 34 wk 0 d Duodenal atresia No other intestinal complications 6 34 wk 2 d Duodenal atresia No other intestinal complications 7 37 wk 5 d Left diaphragmatic hernia Colon involved, liver not involved Note VIBE = volumetric interpolated breath-hold. A B C Fig. 1 MR images of fetus at 35 weeks 4 days gestation show normal gastrointestinal tract. St = stomach, psm = proximal small intestine, dsm = distal small intestine, As = ascending colon, Tr = transverse colon, Ds = descending colon, Sg = sigmoid colon, R = rectum, L = liver, B = urinary bladder. A, Coronal T2-weighted image shows high signal intensity from stomach to small intestine but low signal intensity from transverse colon to sigmoid colon. B, Coronal 3D T1-weighted gradient-echo image shows low signal intensity in stomach and high signal intensity from distal small intestine to colon and rectum. Proximal small intestine shows higher signal intensity than liver does. C, Volume-rendered image in anteroposterior view visualizes through distal small intestine to rectum and also shows liver. to be satisfactory when its signal intensity was distinguished from that of background and when the continuity of the intestine was confirmed. We divided the fetuses with a normal gastrointestinal tract into two age groups of less than 32 weeks gestation (n = 12) and 32 weeks or more gestation (n = 16) on the basis of a previous report [2]. Whether there was herniation, obstruction, dilatation, or narrowing of intestine in the fetal gastrointestinal abnormalities was qualitatively assessed. The diagnosis of herniation was made when the intestine was seen outside the abdomen, whereas the diagnosis of obstruction was made when the intestine was constricted with a resultant dilatation of the bowel proximal to the site of obstruction. Results In all 28 normal and seven abnormal cases, HASTE T2-weighted imaging, VIBE T1- weighted imaging, and 3D MR colonography were performed. A summary of the normal findings is presented in Table 1; abnormal findings are listed in Table 2. In the three cases of duodenal atresia, VIBE imaging and 3D MR colonography did not add to the diagnosis. In the two diaphragmatic hernias and one gastroschisis case, VIBE imaging and 3D MR colonography clearly showed colon involvement in the defect but did not provide additional information. In the case of ileal atresia with meconium peritonitis, the ileal atresia was confirmed (Figs. 1 3). Discussion In fetal MRI, signal intensity of the gastrointestinal tract is basically determined by the location and amounts of swallowed amniotic fluid and meconium [2 4, 6]. Meconium is formed from secretions of the liver and intestinal glands, desquamated intestinal epithelium, and some amniotic fluid after 13 weeks and slowly migrates from the small intestine to the colon and rectum [4]. Therefore, signal intensity of the intestine greatly varies depending on gestational age. According to previous reports [2 4], signal intensity of the stomach through the proximal small intestine is hyperintense on T2- weighted images after weeks gesta- AJR:189, December 2007 W305

4 Inaoka et al. tion, whereas that of the distal small intestine through the colon is intermediate to low signal intensity. On T1-weighted images, A C Fig. 2 MR images of fetus with congenital diaphragmatic hernia at 37 weeks 5 days gestation. R = rectum, L = liver. A and B, Coronal T2-weighted (A) and 3D T1-weighted gradient-echo (B) images show herniated intestine in left thoracic space. Extent of colon into left thoracic space (arrows) is clearly visualized. C and D, Volume-rendered images in anteroposterior (C) and posteroanterior (D) views show anatomic relationship between herniated colon and liver. Colon beyond diaphragm (arrows, C), is clearly shown. Liver has normal shape. Small intestine is not visualized. signal intensity from the sigmoid colon to the rectum is always bright after 23 weeks gestation [2, 3]. The distal small intestine is B D hyperintense in more than half of cases before 32 weeks gestation; thereafter, it remains hyperintense in almost 40% of cases [3]. The visualization of the proximal small intestine on T1-weighted images is limited [2 4]. In our results, the percentages of the respective intestine segments showing a high signal intensity on HASTE T2-weighted and VIBE T1-weighted images were equal or superior to the previously published data; however, the percentage of the proximal small intestine having a high T1 signal was much higher than in previous reports. The rate of the distal small intestine showing a high signal intensity on HASTE T2-weighted images in fetuses at 32 weeks gestation or longer was higher than that in fetuses at less than 32 weeks gestation. The rate of the proximal small intestine showing a high signal intensity on T1-weighted images in fetuses at less than 32 weeks gestation is higher than that in fetuses at 32 weeks gestation or longer. The VIBE sequence, which is a modified 3D fast gradient-echo sequence, uses a symmetric echo in the read direction, partial inplane Fourier sampling in the phase-encoding direction, and asymmetric echo sampling and sinc interpolation along the partition direction [8 11]. This sequence has the ability to provide thinner sections, higher signal-to-noise ratio, higher image contrast, and a shorter acquisition time than conventional sequences while preserving adequate anatomic coverage [8 11]. In fetal MRI, rapid data acquisition is important because it may reduce motion artifacts of both mothers and fetuses. Indeed, VIBE enabled us to obtain sections of 3-mm section thickness through the fetuses with a short acquisition time, and it provided high-quality T1-weighted imaging and 3D MR colonography of the fetuses. Several investigators have suggested that 3D MR images were useful in surgical simulation and treatment planning before birth [7, 12 14]. Rotation of 3D MR colonography images on an image viewer aids in understanding the anatomic position of the intestine and the relation between the intestine and the liver. We have actually presented 3D MR colonography images, with their excellent delineation of the meconium, at family counseling and at conferences with neonatologists and pediatric surgeons. In our results, T1- weighted and 3D MR colonography images of sufficient quality were generated from the same VIBE data sets. The percentages of the recognition of the proximal and distal small W306 AJR:189, December 2007

5 MRI of Fetal Gastrointestinal Tract Fig. 3 MR images of fetus with gastroschisis at 28 weeks 6 days gestation. Sm = small intestine, As = ascending colon, Tr = transverse colon, Ds = descending colon, R = rectum, L = liver, St = stomach. A, Sagittal T2-weighted image shows extraabdominal bowel (arrows) has low signal intensity. B, Coronal 3D T1-weighted gradient-echo image shows absence of colon in abdominal cavity. C and D, Sagittal multiplanar reformatted (MPR) T1-weighted images show eviscerated bowel has high signal intensity, which is consistent with normal colon. Small intestine appears normal. When coronal image sections cannot show abnormal findings, MPR images are useful for evaluation. E H, Volume-rendered images in anteroposterior (E), oblique left-to-right (F), left-to-right (G), and right-to-left (H) views show condition of eviscerated bowel segments of gastroschisis. No volvulus or bowel atresia is seen. Site of abdominal wall defect is predictable because intestine becomes constricted just at defect site (arrows). Small intestine is not visualized. A B D C G E F H AJR:189, December 2007 W307

6 Inaoka et al. intestine on 3D MR colonography images were lower than those on VIBE images. On 3D MR colonography images, the visualization of the small intestine in fetuses at less than 32 weeks gestation was inferior to that in fetuses at more than 32 weeks gestation. We thought that the difference may have been related to the small diameter of the small intestine in fetuses at earlier gestational ages. Our study has some limitations. Because we enrolled a limited number of the fetuses with normal and abnormal gastrointestinal tracts, the sample size is relatively small. Most fetuses were in the third trimester, and the gastrointestinal tract has a different appearance at earlier gestational ages. A large series is required to precisely determine using VIBE the visualization of the respective segments of the fetal gastrointestinal tract at different gestational ages. In addition, we could not determine signal characteristics of fetal gastrointestinal abnormalities, although it has been reported that signal characteristics are more conspicuous in abnormal cases [5]. The impact of VIBE on findings and diagnosis in the abnormal cases is not yet clear. In conclusion, we report the appearances of normal and abnormal fetal gastrointestinal tracts on HASTE T2-weighted imaging, VIBE T1-weighted imaging, and 3D MR colonography. VIBE allowed better visualization of the fetal gastrointestinal tract than techniques used in previous reports [2 4] despite the fact that this sequence provided thinner slice sections through the fetus. We believe that the routine use of VIBE for fetal MRI may offer T1-weighted imaging and 3D MR colonography of high quality for the evaluation of fetal gastrointestinal abnormalities. References 1. Frates MC, Kumar AJ, Benson CB, Ward VL, Tempany CM. Fetal anomalies: comparison of MR imaging and sonography for diagnosis. Radiology 2004; 232: Saguintaah M, Couture A, Veyrac C, Baud C, Quere MP. MRI of the fetal gastrointestinal tract. Pediatr Radiol 2002; 32: Veyrac C, Couture A, Saguintaah M, Baud C. MRI of fetal GI tract abnormalities. Abdom Imaging 2004; 29: Brugger PC, Prayer D. Fetal abdominal magnetic resonance imaging. Eur J Radiol 2006; 57: Farhataziz N, Engels JE, Ramus RM, Zaretsky M, Twickler DM. Fetal MRI of urine and meconium by gestational age for the diagnosis of genitourinary and gastrointestinal abnormalities. AJR 2005; 184: Trop I, Levine D. Normal fetal anatomy as visualized with fast magnetic resonance imaging. Top Magn Reson Imaging 2001; 12: Sasaki Y, Miyamoto T, Hidaka Y, et al. Three-dimensional magnetic resonance imaging after ultrasonography for assessment of fetal gastroschisis. Magn Reson Imaging 2006; 24: Rofsky NM, Lee VS, Laub G, et al. Abdominal MR imaging with a volumetric interpolated breath-hold examination. Radiology 1999; 212: Bader TR, Semelka RC, Pedro MS, Armao DM, Brown MA, Molina PL. Magnetic resonance imaging of pulmonary parenchymal disease using a modified breath-hold 3D gradient-echo technique: initial observations. J Magn Reson Imaging 2002; 15: Semelka RC, Balci NC, Wilber KP, et al. Breathhold 3D gradient-echo MR imaging of the lung parenchyma: evaluation of reproducibility of image quality in normals and preliminary observations in patients with disease. J Magn Reson Imaging 2000; 11: Biederer J, Liess C, Charalambous N, Heller M. Volumetric interpolated contrast-enhanced MRA for diagnosis of pulmonary embolism in an ex vivo system. J Magn Reson Imaging 2004; 19: Schierlitz L, Dumanli H, Robinson JN, et al. Threedimensional magnetic resonance imaging of fetal brain. Lancet 2002; 357: Luks FI, Carr SR, Ponte B, Rogg JM, Tracy TF. Preoperative planning with magnetic resonance imaging and computerized volume rendering in twin-totwin transfusion syndrome. Am J Obstet Gynecol 2001; 185: Hata N, Wada T, Chiba T, Tsutsumi Y, Okada Y. Dohi T. Three-dimensional volume rendering of fetal images for the diagnosis of congenital cystic adenomatoid malformation. Acad Radiol 2003; 10: W308 AJR:189, December 2007