| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Clinical Radiology, Klinikum Innenstadt, Ludwig Maximilans
University of Munich, Ziemssenstr. 1, 80336 Munich, Germany.
2 Radiologie Starnberger See, Oßwaldstr. 1, 82319 Starnberg,
Germany.
Received May 24, 2001;
accepted after revision February 14, 2002.
Address correspondence to G. Leinsinger.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. HASTE sequences were performed in 80 infants and children. Two radiologists who were unaware of the patients' medical histories independently reviewed the images for the presence of nine findings: defects of the parenchyma, hypoplasia or agenesis of the corpus callosum, edema, signs of increased intracranial pressure, myelination disorders, migration disorders, malformations, tumors, and widening of spaces of the cerebrospinal fluid. A conventional MR imaging examination that served as the reference examination was evaluated by the same two radiologists in a final consensus interpretation. The findings detected on the HASTE images were compared with the findings seen on the conventional MR images. The sensitivity and specificity of HASTE sequences were calculated, and Cohen's kappa statistic was used to determine interobserver agreement.
RESULTS. Both radiologists correctly diagnosed all 20 defects of the parenchyma that were present in the patients. Radiologist 1 correctly identified 20 and radiologist 2 correctly identified 21 of the 22 patients with hypoplasia or agenesis of the corpus callosum. Both radiologists correctly diagnosed edema in eight of the nine patients in whom edema was present, and both correctly identified signs of increased intracranial pressure in eight of the nine children who had this condition. Radiologist 1 correctly diagnosed seven and radiologist 2 correctly identified nine of the 11 cases of myelination disorders. Both radiologists correctly diagnosed six of the 14 cases with migration disorders. All 13 brain malformations present in the patients were correctly identified by both reviewers. Both radiologists correctly identified all 11 patients with tumors, and both correctly identified all 35 patients with widening of spaces of the cerebrospinal fluid.
CONCLUSION. HASTE images are highly sensitive for excluding the presence of brain tumor, hydrocephalus, or malformations of the brain. HASTE images are not reliable for evaluating patients with suspected myelination disorders or migration disorders.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
With the development of a new generation of MR imaging systems equipped with stronger and faster gradients, the single-shot acquisition of long echo trains after one excitation pulse has become possible. The half-Fourier single-shot turbo spin-echo (HASTE) technique exploits this feature and allows fast acquisition of T2-weighted MR images. This method requires performance of only little more than half the phase-encoding steps required in other sequences. A mirror-image copy of the data is created to reconstruct the complete MR image. In HASTE sequences, all data are acquired after a single radiofrequency pulse excitation by interpreting multiple echoes [2,3,4]. HASTE sequences have already been shown to be useful in imaging uncooperative patients [3,4,5,6,7,8,9]. In addition, because of the short acquisition time (0.4 sec) needed for a single slice, the HASTE technique renders images less susceptible to motion artifacts. Thus, HASTE sequences are less susceptible to respiratory and intestinal motion artifacts and are therefore routinely used in imaging the abdomen.
In cooperative patients, MR images of the brain are not marred by motion artifacts. When compared with conventional MR imaging techniques, the limitations of HASTE sequences in imaging the brain are the decreased resolution, contrast, and magnetic susceptibility sensitivity. Therefore, HASTE sequences are not typically used in brain imaging. The purpose of our study was to compare the diagnostic value of HASTE sequences with conventional MR imaging techniques in brain examinations of pediatric patients.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In addition to the previously described MR imaging protocol, HASTE images were acquired in the axial plane (5/62; echo-train length, 64; slice thickness, 4 mm; and interslice gap, 1.2 mm). The appropriate field of view was selected on the basis of the different sizes of the heads of the children and ranged from 138 x 220 mm to 165 x 220 mm. The scanner software was set so as to ensure that angles were uniform from sequence to sequence in sedated children (parameters from the first sequence were used as a reference for later sequences). In unsedated children, such uniformity could not be guaranteed.
Patients
Eighty children (45 boys and 35 girls; age range, 6 days to 13 years 8
months; mean age, 4.4 ± 3.7 years) with known or suspected disease of
the brain were referred to MR imaging diagnostics and included in our study.
Written informed consent of the parents or legal guardians of the children was
obtained before the MR imaging examination. Exclusion criteria were
contraindications to exposure to a high magnetic field or former participation
in this study. Twenty-six children were complaint and, therefore, their MR
imaging examination included HASTE sequences performed without anesthesia or
sedation. Fifty-four children underwent the study protocol under sedation or
general anesthesia.
Data Evaluation
All images were printed as hard copies and stored on optical disks. The
contrast on the films was maximized for each individual sequence. HASTE images
were separated from the images acquired by conventional MR imaging sequences
and arranged randomly. The images were retrospectively reviewed independently
by two radiologists experienced in pediatric MR imaging who were unaware of
the medical histories and diagnoses of the patients. Later, all the images of
only the conventional sequences of the MR imaging examination were reviewed in
a final consensus interpretation by the same two radiologists. Judged were the
absence or presence of the following findings: defects of the parenchyma,
hypoplasia or agenesis of the corpus callosum, edema, signs of increased
intracranial pressure, myelination disorders, migration disorders,
malformations, tumors, and widening of spaces of the cerebrospinal fluid. The
presence of motion artifacts was graded on a 3-point scale: 0, no motion
artifacts were noted; 1, moderate motion artifacts were noted, but the images
were still of diagnostic quality; and 2, markedly noticeable motion artifacts
rendered images to be not of diagnostic quality. The results of the final
consensus interpretation combined with clinical findings served as the
standard of reference.
Statistical Analysis
Examinations that received a motion artifact score of 1 or 2 were excluded
from the statistical analysis. The results of the two reviewers were evaluated
separately. Sensitivity, specificity, and positive and negative predictive
values of HASTE sequences were calculated for all evaluation criteria. Cohen's
kappa statistic was used to determine interobserver agreement.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
We used our previously mentioned standard of reference to determine the diagnostic findings. According to our reference, four of the 20 children with defects of the parenchyma had infarctions; five had unspecified small defects (maximum diameter, < 1 mm); two had unspecified gliosis; and nine patients had lesions attributable to ventricular shunts. Eighteen of the 21 children with disorders of the corpus callosum had hypoplasia of the corpus callosum, two had an agenesis of the corpus callosum, and one had partial thinning of the corpus callosum. In eight children, signs of edema were present, and in eight, signs of increased intracranial pressure were present. In nine of the 11 children who presented with disorders of myelination, delayed myelination was detected; Leigh disease was diagnosed in the other two children. In 14 children, migration disorders were found. Of the 13 children who had diagnosed malformations of the brain, nine had Chiari's malformations, and one had an arteriovenous malformation. Two children showed signs of aplasia of the vermis cerebelli and one, colpocephaly. The 11 tumors detected were classified as three primitive neuro-ectodermal tumors, four gliomas, one astrocytoma, one teratoma, one angioma, and one tumor of unknown nature. Nine of the 35 children with widened spaces of the cerebrospinal fluid presented with hydrocephalus.
On conventional MR images, defects of the parenchyma were seen in 20 children. On HASTE sequence images, all 20 defects of parenchyma were correctly identified by both radiologists (sensitivity, 100%). Radiologist 1 had two false-positive findings (specificity, 90%), and radiologist 2 had three false-positive findings (specificity, 89%). The positive predictive value was 80% for radiologist 1 and 77% for radiologist 2. The negative predictive value was 100% for radiologist 1 and 100% for radiologist 2. The kappa value was calculated to be 0.76.
Disorders of the corpus callosum were seen in 22 children on the conventional MR images. On HASTE images, radiologist 1 correctly diagnosed 20 disorders of the corpus callosum (sensitivity, 91%), and radiologist 2 correctly identified 21 (sensitivity, 96%). Both radiologists had one false-positive finding on the HASTE images (specificity, 98%). The positive predictive value was 95% for radiologist 1 and 96% for radiologist 2. The negative predictive value was 96% for radiologist 1 and 98% for radiologist 2. The kappa value was calculated to be 0.97.
Edema appeared to be present in nine children on the conventional MR images. In evaluating for edema on HASTE images, both radiologists correctly identified eight children with edema (sensitivity, 89%). Radiologist 1 had five false-positive findings (specificity, 92%), and radiologist 2 had two false-positive findings (specificity, 97%). The negative predictive value was 98% for radiologist 1 and 98% for radiologist 2. The positive predictive value was 62% for radiologist 1 and 80% for radiologist 2. The kappa value was 0.74.
Signs of increased intracranial pressure were detected in nine children on conventional MR images. Both radiologists correctly identified signs of increased intracranial pressure on HASTE images in eight children (sensitivity, 89%). Radiologist 1 had five false-positive findings (specificity, 92%), and radiologist 2 had four false-positive findings (specificity, 94%). The positive predictive value was 62% for radiologist 1 and 67% for radiologist 2. The negative predictive value was 98% for radiologist 1 and 98% for radiologist 2. The kappa value was calculated to be 0.95.
Myelination disorders (Fig. 1A,1B) were found in 11 children on conventional MR images. Reviewing HASTE images, radiologist 1 correctly identified seven (sensitivity, 64%), and radiologist 2 correctly diagnosed nine (sensitivity, 82%) of the myelination disorders present. Radiologist 1 had six false-positive findings (specificity, 90%), and radiologist 2 had nine false-positive findings (specificity, 75%). The negative predictive value was 94% for radiologist 1 and 97% for radiologist 2. The positive predictive value was 50% for radiologist 1 and 54% for radiologist 2. The kappa value was calculated to be 0.31.
|
|
Fourteen cases of migration disorders (Fig. 2A,2B,2C) were seen on the conventional MR images. Evaluation of HASTE images showed that both radiologists correctly identified six children with migration disorders (sensitivity, 43%). Radiologist 1 had eight false-positive findings (specificity, 86%), and radiologist 2 had seven false-positive findings (specificity, 88%). The positive predictive value was 43% for radiologist 1 and 46% for radiologist 2. The negative predictive value was 86% for radiologist 1 and 86% for radiologist 2. The kappa value was 0.32.
|
|
|
Thirteen malformations (Fig. 3A,3B) were identified on the conventional MR images. All malformations were correctly diagnosed by both radiologists on the HASTE images. There were neither false-negative nor false-positive findings, resulting in a sensitivity of 100%, a specificity of 100%, a positive predictive value of 100%, and a negative predictive value of 100% for both radiologists, the kappa value being 1.00.
|
|
All 11 tumors (Fig. 4A,4B) identified on the conventional MR images were correctly detected on the HASTE images by both radiologists, resulting in a sensitivity of 100% for both reviewers. Because there were no false-negative results, the negative predictive value was 100% for both radiologists. Radiologist 1 had two false-positive findings (specificity, 97%), and radiologist 2 had three false-positive findings (specificity, 95%). The positive predictive value was 85% for radiologist 1 and 79% for radiologist 2. The kappa value was 0.76.
|
|
Thirty-five children had widened spaces of the cerebrospinal fluid (Fig. 5A,5B) on the conventional MR images. All cases of this disorder were correctly identified on the HASTE images by both radiologists. Neither false-negative nor false-positive findings occurred, resulting in a sensitivity of 100%, a specificity of 100%, a positive predictive value of 100%, and a negative predictive value of 100% for both radiologists. The kappa value was 1.00.
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
On HASTE images, defects of the parenchyma were identified with a high degree of sensitivity, whereas specificity was low because of the high number of false-positive findings. Because cerebrospinal fluid appears strongly hyperintense on HASTE images, small perivascular spaces or partial volume effects of sulci may be overestimated by the observer, which perhaps contributed to the high number of false-positive interpretations. In evaluating edema with HASTE sequences, our reviewing radiologists had two and five false-positives, respectively. Sugahara et al. [4] found HASTE sequences to be more sensitive to blurring artifacts than conventional T2-weighted turbo spin-echo sequences. This tendency might have contributed to the false-positive findings of edema in our study. A limited differentiation of the sulci and gyri may simulate the swelling of the parenchyma seen in the presence of edema.
Sensitivity and specificity were high for both radiologists in evaluating the corpus callosum on HASTE images. The one case of partial thinning of the corpus callosum could be detected only on sagittal images. Therefore, it could not have been identified by the two radiologists from the axial HASTE images.
In evaluating for signs of increased intracranial pressure, there was one false-negative compared with five false-positives (radiologist 2, four false-positives). The false-positive findings might be explained by the presence of blurring artifacts, which makes differentiation of the sulci and gyri difficult and gives the appearance of swelling of the parenchyma.
Sensitivity for the detection of myelination disorders and migration disorders was rather low in our study. In addition, poor interobserver agreement was found, which illustrates the difficulty of evaluating these disorders with HASTE sequences. This difficulty may be due to the low signal-to-noise ratio and low signal-to-contrast ratio because of the reduced data acquisition [12, 13], which are limitations of HASTE sequences. (For example, Patel et al. [3] reported that T2-weighted fast spin-echo sequences are better than HASTE sequences for differentiating gray from white matter.) These problems contributed to the unsatisfying results in the detection of myelination and migration disorders. In assessing brain maturation, T1-weighted images are more useful during the first 6-8 months of a child's life and T2-weighted images are more useful thereafter [14]. T2- and T1-weighted images are complementary in the assessment of the different stages of myelination [15,16,17]. Failure to acquire both sequences and to understand the normal maturation process may result in incorrect interpretation of the MR images [18]. As these facts make clear, T2-weighted HASTE sequences are not suitable for detecting diseases associated with disorders of myelination.
Use of HASTE sequences did not result in any false-negative findings in the detection of tumors. The detection of small hyperintense lesions (<5 mm in diameter [3]) and small hypointense lesions has been shown to be better on conventional T2-weighted sequence images than on HASTE sequence images [2]. In our patients, the smallest tumor of the central nervous system was 13 mm in maximal diameter.
All cerebral malformations were detected on HASTE images by both radiologists, showing that these sequences are a reliable diagnostic tool for the exclusion of these anomalies.
Neither of the radiologists had false-negative nor false-positive findings in the assessment of widening spaces of the cerebrospinal fluid, showing that imaging with HASTE sequences is not only a suitable examination tool for the primary diagnosis of hydrocephalus but also is an ideal follow-up examination for the affected children because no anesthesia or dose of radiation is needed.
It is unreasonable to ask children (and their parents) who need a conventional MR imaging examination requiring sedation or anesthesia to undergo two MR imaging examinations. Therefore, HASTE and conventional sequences in these children were performed at the same time. Thus, clinical applicability and occurrence of motion artifacts were tested only in 26 of the 80 children.
A limitation of performing MR imaging with T2-weighted HASTE sequences is that it is not possible to use the regular T1-shortening contrast agents. The additional use of T1-weighted HASTE sequences [10, 11] would lengthen the examination time but should be considered.
HASTE sequences can successfully be performed even in uncooperative children without use of sedation or anesthesia. T2-weighted HASTE sequences are highly sensitive for the exclusion of the presence of brain tumor, hydrocephalus, and malformations of the brain. HASTE sequences are not reliable to use in the assessment of suspected myelination or migration disorders. Both of these disorders are common in pediatric patients and are important to detect. Therefore, in cases in which the HASTE images show no abnormalities and the clinical presentation of the child remains unclear, myelination and migration abnormalities should be excluded by performing a complete MR imaging examination, including conventional sequences. In the case of acute neurologic symptoms, however, exclusion of the possibility of a brain tumor or hydrocephalus or both is critical for therapy planning. On the basis of our findings, exclusion of these conditions can be performed with HASTE sequences without the use of sedation or general anesthesia. If brain edema is suspected from the findings on HASTE imaging, verification via a complete MR imaging examination is necessary.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  K. D. Singh, N. S. Logan, and B. Gilmartin Three-dimensional modeling of the human eye based on magnetic resonance imaging. Invest. Ophthalmol. Vis. Sci., June 1, 2006; 47(6): 2272 - 2279. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Garel, A.-L. Delezoide, M. Elmaleh-Berges, F. Menez, C. Fallet-Bianco, E. Vuillard, D. Luton, J.-F. Oury, and G. Sebag Contribution of Fetal MR Imaging in the Evaluation of Cerebral Ischemic Lesions AJNR Am. J. Neuroradiol., October 1, 2004; 25(9): 1563 - 1568. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
