| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Obstetric Imaging |
1 From the Department of Obstetrics and Gynecology, Chiba University Hospital, Chiba University School of Medicine (H.O., K.K., K.M., Y.I.) and Department of Reproductive Medicine, Graduate School of Medicine, Chiba University (K.S., S.S.), 1–8-1 Inohana, Chuo-ku, Chiba-shi, Chiba 260-8677 Japan. Received December 19, 2002; revision requested February 7, 2003; final revision received September 10; accepted October 21. Address correspondence to H.O. (e-mail: hosada@mue.biglobe.ne.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A total of 87 fetuses evaluated with MR imaging at 24–39 weeks of gestation were classified into a control group with good respiratory outcome (group A, n = 58) or a poor outcome group with severe respiratory disturbance after birth (group B, n = 29). Planimetric measurement of total lung volume and calculation of the ratio of lung signal intensity to spinal fluid signal intensity (L/SF) were performed on MR images by using region-of-interest analysis. Regression analysis, analysis of covariance, analysis of variance, and receiver operating characteristic (ROC) analysis were performed.
RESULTS: The best fit for group A lung volume was represented by the regression line V = (2.41 x G) – 37.6 (r = 0.537, P < .001), in which V is lung volume and G is gestational weeks; that for group B, by V = (0.97 x G) – 14.0 (r = 0.378, P < .05). Results of analysis of covariance with gestational weeks used as a covariate showed a significant difference in lung volume between the two groups (P < .001). Mean ± SEM for L/SF ratio was 0.817 ± 0.013 and 0.598 ± 0.019 in groups A and B, respectively (P < .001). For prediction of postnatal respiratory outcome, the area under the ROC curve for lung volume and L/SF ratio combined was 0.990, significantly higher than that for lung volume alone (P < .05).
CONCLUSION: Simultaneous measurement of fetal lung volume and signal intensity on MR images is a promising method for predicting fetal pulmonary hypoplasia.
© RSNA, 2004
Index terms: Fetus, respiratory system, 80.8755 • Lung, abnormalities, 80.8755 • Lung, MR, 60.121412, 60.121419, 60.12146
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Measurements of thoracic circumference, lung length, and ratio of thoracic circumference to abdominal circumference with conventional ultrasonography (US) have been proposed as methods for determining fetal lung development (1–3). However, these parameters are not sensitive or specific enough to be used in clinical decision making, although the approach is attractive because it is noninvasive and rapid. Other studies have demonstrated the usefulness of fetal lung volume measurement with three-dimensional US (4–9). While this technique allows access to all possible views and accurate determination of organ volume, it may be affected by fluid volume or maternal obesity and does not provide any additional physiologic information relating to potential function.
Magnetic resonance (MR) imaging is a relatively new but well-established medical imaging technique. The development of fast pulse sequences, such as those used in echo-planar imaging (10) and single-shot rapid acquisition with relaxation enhancement (11–13), has brought major advances in fetal MR imaging. Clearer fetal MR images with fewer motion artifacts can be acquired without fetal immobilization. These MR imaging techniques allow detailed anatomic assessments, including estimation of lung volume, and can be used in place of US. The feasibility of using volumetric measurements on MR images to evaluate fetal pulmonary hypoplasia has been reported previously (14–18). Furthermore, the possibility that the MR signal intensity of fetal lung is a good indicator of fetal lung maturation has been shown by MR relaxation time measurement (19).
In the present study, our purpose was to measure both the volume and the signal intensity of the fetal lung with MR imaging and to evaluate the possibility of clinical application of this procedure to predict fetal pulmonary hypoplasia.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The subjects were classified into two groups by three authors (K.K., K.M., Y.I.) using the medical record information, as follows: Group A (n = 62), the control group, included fetuses with good postnatal respiratory outcome that either did not require endotracheal intubation or required only short-term mechanical ventilation. Group B (n = 29) included fetuses with poor outcome, with severe respiratory disturbance after birth. In this study, severe respiratory disturbance was defined as respiratory failure resulting in neonatal death, use of high-frequency ventilation or extracorporeal membrane oxygenation, or long-term mechanical ventilation for more than 4 weeks. In group A, the most common MR diagnoses involved cerebrospinal diseases (16 cases), followed by kidney diseases (14 cases) and abdominal wall diseases (10 cases). Although four subjects with chest diseases comprising cystic adenomatoid malformation of the lung type 1 (three cases) and congenital diaphragmatic hernia (one case) also fulfilled the criteria for good postnatal respiratory outcome, they were judged inappropriate as controls and were excluded from the study. Therefore, the final number of subjects in group A was 58. Group B consisted of 11 subjects with congenital diaphragmatic hernia, seven with hydrothorax, three with bilateral polycystic kidney, two with bilateral renal agenesis, two with oligoamnios, one with cystic adenomatoid malformation of the lung, one with bronchopulmonary sequestration, one with osteogenesis imperfecta, and one with short-limbed skeletal dysplasia.
Gestational age was confirmed and determined with routine first-trimester US. Fetal weight was estimated with the Hadlock formula by using US measurement performed on the same day as MR imaging by three authors (K.K., K.M., Y.I.).
MR Imaging
MR imaging was performed with a 1.5-T superconducting magnet (Signa; GE Medical Systems, Milwaukee, Wis) and a four-element phased-array surface coil (GE Medical Systems). T1-weighted images were obtained by using breath holding and a spoiled gradient-echo sequence with repetition time msec/echo time msec of 100–150/4.2, flip angle of 70°–90°, matrix of 256 x 128–192, and one signal acquired. T2-weighted images were obtained by using single-shot rapid acquisition and relaxation enhancement (
/90 [effective]) with a matrix of 256 x 160–256. A minimal bandwidth was used for all sequences. Acquisition time was 15–30 seconds. Section thickness and intersection gap were 4–6 mm and 0–1 mm, respectively. T2-weighted images were acquired in transverse, sagittal, and coronal planes with respect to the fetal thorax. In the presence of fetal central nervous system anomalies, additional MR imaging of the fetal thorax in a transverse plane was performed. If the fetal thorax was not imaged completely or fetal movement was observed, the acquisition was repeated. There were no cases in which images could not be used because of fetal motion.
Image Evaluation
One reader (H.O.), who did not participate in obtaining medical record information, performed planimetric measurement of lung volumes in all cases according to the methods of Coakley et al (16). Therefore, the reader was generally blinded to outcome, with certain exceptions, such as in cases in which bilateral renal anomalies with oligohydramnios were depicted on images. The reader selected images acquired with single-shot rapid acquisition and relaxation enhancement sequences that depicted the complete fields of both lungs without motion or section misregistration artifacts. The fetal lungs were well depicted at MR imaging and exhibited high signal intensity on T2-weighted images, presumably because the fetal lungs in utero are filled with amniotic fluid and not air. Even in cases of severe hypoplasia or large lung mass, lung borders could be distinguished. The cross-sectional area of the lung was measured on each transverse section by using a picture archiving and communication system workstation (Advantage Windows RP version 2.0; GE Medical Systems). In this step, the hila of the lung, and lung lesions if present, were excluded from lung volume measurement. To calculate the volume for each section, the value of the cross-sectional area was multiplied by that of the section thickness and intersection gap combined. The volumes of all of the sections were then summed to obtain the volume of the entire lung. The calculation was repeated for the contralateral lung, and then the volumes of both lungs were added to obtain the total lung volume.
Furthermore, the same reader (H.O.) calculated the ratio of the lung signal intensity to the spinal fluid signal intensity (L/SF) by using region-of-interest (ROI) analysis (Fig 1). With the same computer program used for image analysis, the mean signal intensity (pixel value) of an ROI that covered the entire cross-sectional area of the right and left lungs on each transverse section was measured. Measurements from a circular ROI of spinal fluid, ranging from 3 to 10 mm2, were also obtained; the ROI was placed to include just fluid, and not the spinal cord. For each section, the mean signal intensity of the left and right lungs was divided by the signal intensity of the spinal fluid. This procedure was repeated for all lung sections, and the mean value for the L/SF ratio was obtained.
|
The sample sizes were determined on the basis of data from a preliminary study in which we measured the volume and signal intensity of the lung at MR imaging in 20 fetuses between 24 and 32 weeks of gestation with good respiratory outcome. The lung volume (mean ± SD) was 37.8 mL ± 18.6, and the L/SF ratio was 0.795 ± 0.117. The study by Coakley et al (16) demonstrated that the mean of the relative degree of lung volume was 28.6% in fetuses suspected of having pulmonary hypoplasia. Kuwashima et al (21) reported mean lung-to-liver signal intensity ratios of 1.41 in a poor outcome group and 2.79 in a control group. Therefore, our power calculations indicated that approximately 14 and four subjects per group would be needed to detect reductions of at least 71.4% and 49.5% in lung volume and signal intensity, respectively, with 95% power at a two-sided 
of .05. Each of the two groups in our study comprised more than 14 subjects.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
ROC curves were plotted to predict fetal respiratory outcome after birth (Fig 4). The area under the ROC curve for lung volume alone was 0.930 (SEM: 0.027), and that for L/SF ratio alone was 0.955 (SEM: 0.020). The area under the ROC curve for lung volume and L/SF ratio combined was 0.990 (SEM: 0.008), and this value was significantly higher than that for lung volume alone (P < .05).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Fetal volumetric measurements can readily be performed on MR images with the use of planimetry. Lung volume also can be measured on MR images, and findings of a longitudinal study of fetal volume estimations with echo-planar imaging techniques in a healthy fetal population have been published (14,22,23). Furthermore, lung volume in fetuses suspected of having pulmonary hypoplasia has been assessed previously (15–18). The results of these studies suggest that MR volumetry can be used in the evaluation of fetal pulmonary hypoplasia. Our study results are consistent with the conclusions of these previous investigations: MR images are useful to quantitate pulmonary hypoplasia.
It is evident that organ size does not always correlate with function. Rypens et al (17) reported no great difference in lung volume in a fetus with diaphragmatic hernia that died shortly after birth, compared with lung volume in fetuses that survived after birth. Differences in intrapulmonary liquid content and lung histologic characteristics, however, might have contributed to the different outcomes. Therefore, in addition to lung volume estimation, other information provided by fetal MR imaging, such as relative lung signal intensity, should be considered in the assessment of the fetal lung.
Duncan et al (19) reported that MR relaxation time in fetal lungs was a good parameter of measurement, although only in healthy fetuses. They proposed that qualitative changes observed on fetal lung MR images might be good indicators of fetal lung maturation. Moore et al (24) presented the first report of in vivo measurements of diffusion in the fetal lungs and gave the range of apparent diffusion coefficients expected in normal pregnancy. The apparent diffusion coefficient increases with gestational age, mainly as the result of an increase in pulmonary vasculature. Unfortunately, the echo-planar techniques used in these two studies do not provide the spatial resolution offered by other, more widely available fast sequences (25).
In relatively small-scale studies, MR imaging in fetuses with pulmonary hypoplasia showed low lung-to-liver signal intensity ratios (20,26). In our study, there was a close correlation between fetal L/SF ratio and lung size. Although the pathogenesis of pulmonary hypoplasia is poorly understood, fetal lung fluid plays a key role in normal lung development (27,28). Presumably, high signal intensity indicates a substantial amount of fetal lung fluid in the voluminous small airways and alveoli, whereas low intensity signifies the absence of fetal lung fluid (20).
It is well established that the use of measured signal intensity values as absolutes in the quantitative analysis of MR images is difficult and that the stability of the imager cannot be guaranteed. Direct comparison of lung signal intensities without standardization of variables is therefore inappropriate (29). In this study, standardization was achieved by calculating the L/SF ratio. Spinal fluid signal intensity was chosen as the reference for the following reasons: (a) The fluid-filled spinal canal is present in every transverse section of the lung; (b) the spinal fluid signal intensity is high, which reduces random error attributable to noise in the calculation of the final ratio; (c) the spinal fluid signal can be measured close to the fetal lung, which minimizes the possible effects of a lack of homogeneity across the image; and (d) the spinal fluid signal intensity is not expected to change with increasing gestation time.
It is noteworthy that although a significant correlation was observed between lung volume and gestational age, there was wide variability in the lung volume and L/SF ratio at each gestational age. For example, the range of normal lung volumes at approximately 34 weeks was 30–90 mL, with the lowest and highest values diverging by a factor of three (Fig 2). The r value for the regression analysis was relatively low (0.54). Accordingly, only 29% of the planimetric variance can be explained by gestational age alone. This variability demands careful clinical interpretation of data, especially when they are to be used as reference values for normal lung volume.
The present analysis is based on a retrospective comparative study of the most severe cases of respiratory disturbance. The study group (group B) contained cases of multiple underlying diseases. The control fetal lung volume was derived from fetuses suspected of having various abnormalities at US examination, although they had a good respiratory outcome after birth. There is a need for future, prospective studies of the use of MR images to predict severe respiratory disturbance in patient subgroups with specific intrathoracic congenital diseases, such as congenital diaphragmatic hernia and oligoamnios.
Neonatal respiratory problems remain an important cause of mortality and morbidity, and any obstetric plan should account for the possibility that the fetal lungs may be hypoplastic. Simultaneous use of volumetric and signal intensity measurements on MR images may play an important role in the evaluation of fetal pulmonary hypoplasia.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, H.O.; study concepts and design, H.O., K.S., S.S.; literature research, H.O.; clinical studies, K.K., K.M., Y.I.; data acquisition, K.K., K.M., Y.I.; data analysis/interpretation, H.O., K.K., K.S.; statistical analysis, H.O., K.S.; manuscript preparation, H.O., K.K., K.M., Y.I.; manuscript definition of intellectual content, H.O., K.K.; manuscript editing, H.O., K.S.; manuscript revision/review, S.S; manuscript final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. F. A. Peralta, J. F. Kazan-Tannus, V. Bunduki, E. M. Santos, C. C. de Castro, G. G. Cerri, and M. Zugaib Evaluation of the Agreement Between 3-Dimensional Ultrasonography and Magnetic Resonance Imaging for Fetal Lung Volume Measurement J. Ultrasound Med., April 1, 2006; 25(4): 461 - 467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. F. Goncalves, W. Lee, J. Espinoza, and R. Romero Three- and 4-Dimensional Ultrasound in Obstetric Practice: Does It Help? J. Ultrasound Med., December 1, 2005; 24(12): 1599 - 1624. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. A. Obuchowski ROC Analysis Am. J. Roentgenol., February 1, 2005; 184(2): 364 - 372. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), fetuses with good respiratory outcome; and group B (
), fetuses with severe respiratory disturbance after birth. Lung volume generally increased as gestation progressed, but there was considerable individual variation. The lung volumes in group B subjects overall were consistently lower than those in group A subjects, and there was little overlap in distribution of values between the two groups for the entire gestation period studied.
