DOI: 10.1148/rg.262055009
RadioGraphics 2006;26:497-501
© RSNA, 2006
Interactive Digital MR Atlas of the Pediatric Brain1
Srdjan Jelacic, BSc,
David de Regt and
Edward Weinberger, MD
1 From the Children’s Hospital and Regional Medical Center and the University of Washington School of Medicine, 4800 Sand Point Way NE, Seattle, WA 98105 (S.J., E.W.); and Reed College, Portland, Ore (D.d.R.). Recipient of a Certificate of Merit award for an education exhibit at the 2004 RSNA Annual Meeting. Received January 24, 2005; revision requested April 29 and received June 21; accepted June 22. Supported in part by the Clifford A. Garl Radiology and Medical Imaging Education Fund. All authors have no financial relationships to disclose.
Address correspondence to E.W. (e-mail: vankleek{at}u.washington.edu).
 |
Abstract
|
|---|
The interpretation of magnetic resonance (MR) images of the pediatric brain may require consultation of an atlas to determine if a perceived finding represents an abnormality. However, most hard-copy atlases show only a few levels of the brain at selected points of time in myelination, and therefore a simultaneous comparison of different age groups is difficult with a hard-copy approach. The authors believe that a digital atlas of the normal pediatric brain may be a more efficient way to present this information and that correct interpretation of potential abnormalities may be facilitated by the online atlas they have created (available for free download from http://radiology.seattlechildrens.org/teaching/pediatricbrainatlas). Images for the atlas were derived from Digital Imaging and Communications in Medicine–compliant data sets from brain MR imaging in patients younger than 4 years. The images were interpreted as normal with respect to the appearance of the brain parenchyma. The software program used for viewing the atlas, written in C#, incorporates many features of a picture archiving and communication system viewer, such as linked scrolling and resizing. Simultaneous comparison of cases also is possible. The digital atlas facilitates learning about normal changes in the MR appearance of the pediatric brain, and it may be used during online interpretation of cases on a picture archiving and communication system.
© RSNA, 2006
|
Introduction
|
|---|
Magnetic resonance (MR) evaluation of the pediatric brain is confounded by the changing appearance of normal structures during the first 2 years of life, predominantly because of the progression of myelination. Knowledge of the normal progression of change as it appears on MR images obtained with various pulse sequences is essential for correct interpretation of many conditions, including heterotopia, delay of myelination, and infarction. Although hard-copy atlases of the pediatric brain exist, they generally show only a few levels of the brain, at selected points of time during myelination. The utility of a hard-copy atlas may be limited if it does not include an image at the specific level and the specific point of time in myelination that are necessary for comparison during image interpretation. Furthermore, direct comparison is hindered by the necessity of turning the pages back and forth; the author must decide what type of comparison the atlas users are most likely to need, rather than the user being able to decide what is most useful. To enable more efficient and more comprehensive image reference, we created a digital electronic atlas that includes images of the entire normal brain during many points in time and as observed with different MR pulse sequences. The atlas also facilitates comparisons according to user-selected parameters such as patient age, MR pulse sequence, or image level, an improvement over the limited static comparison that is inherent in a hard-copy atlas.
|
Image Selection
|
|---|
The atlas is composed of brain MR images obtained with a 1.5-T magnet in children aged 0–4 years. The appearance of the brain parenchyma on all images included in the atlas was considered normal. Images were excluded if they were obtained in a subject with developmental delay, or if the subject or the subject’s mother had a significant medical condition that could potentially interfere with the normal progression of myelination. The most common indication for MR evaluation was seizure. Other indications included headache and scalp abnormalities that were considered unlikely to interfere with normal myelination. Patient ages were normalized according to an assumed 40-week term of gestation. The final atlas included images obtained in patients from 7 weeks premature to 48 months. Images included axial T1-weighted, axial T1-weighted inversion recovery (if younger than 1 year), axial T2-weighted, axial fluid-attenuated inversion recovery, axial diffusion-weighted (b = 1000), coronal T2-weighted, and midline sagittal T1-weighted images, as well as axial apparent diffusion coefficient maps. Images were saved in Joint Photographic Experts Group, or JPEG, format.
|
Software Program
|
|---|
The program used to view the atlas was written in C# and uses the .NET Framework tool set. The program runs in any version of Microsoft Windows with the .NET Framework installed. No additional software, including database software, is required. An initialization file allows the customization of many program parameters, including the default viewing setup, which dictates the order in which images are viewed with respect to MR sequences and age of the patients. Instructions for downloading and using the atlas, which is freely available, can be found at http://radiology.seattlechildrens.org/teaching/pediatricbrainatlas.
|
Program Features
|
|---|
The interactive MR digital atlas of the pediatric brain has a simple yet comprehensive interface. When the default settings are used, the user is presented with a set of images from patients that provide the best examples of normal myelination progression at the selected ages, a set that is called the standard list. An expanded list may be selected if a more in-depth evaluation is required. This list includes a larger set of images from a larger group of patients. In addition, a simple check-box function (the "checked-only" list) allows for the inclusion or exclusion of any subset of patients so that specific comparisons of a limited data set can be made.
The atlas, when used in compare mode, allows easy comparison of images obtained with the same MR pulse sequence across different age groups (Fig 1), and, when used in case mode, allows comparison of images obtained with multiple MR pulse sequences in patients of the same age (Fig 2). The so-called sheet mode allows the viewing of an entire image series as if it were displayed on a sheet of film (Fig 3). The user can view all the images for a given sequence or may select a filtered list in which the images are aligned with five preselected levels. These levels include the midbrain, cerebral peduncles, two levels at the basal ganglia, and the centrum semiovale (Fig 4). The user also can modify the number and orientation of image windows and can turn on or off the comment field, which contains annotations about the patient’s medical history and image findings. The comment field, which is image specific, appears as semitransparent text overlying the bottom of the image.
View larger version (90K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Example of the use of a 3 x 2 viewing matrix in the compare mode to display T2-weighted images at similar levels of the brain in six patients of different ages (term, 3 weeks, 3 months, 9 months, 18 months, and 31 months) selected from the standard list.
|
|
View larger version (93K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Example of the use of a 4 x 2 viewing matrix in the case mode to display images obtained with eight different pulse sequences in a selected patient. By default, images in the same plane are linked to scroll together.
|
|
The program is designed to allow easy comparison with an MR imaging case that is being evaluated prospectively on a picture archiving and communication system. The inclusion of MR images obtained with many different pulse sequences for all patients, rather than just standard T1- and T2-weighted images, allows a more in-depth understanding of changes that occur in the pediatric brain. For example, substantial changes in the appearance of the brain on FLAIR images in the first 6 months of life make this sequence very helpful as an adjunct for assessing the degree of myelination. There is also considerable change in the appearance of diffusion-weighted images and apparent diffusion coefficient maps obtained during this time frame. Knowledge of the specific changes in appearance is crucial for the correct interpretation of images obtained with these sequences.
|
Limitations
|
|---|
The program is designed to allow easy comparison of all cases in the provided data set. Since, as a rule, pediatric patients without medical indications do not routinely undergo MR imaging, we had to select patients who we believed were likely to have normal MR images for age. We acknowledge that the inclusion of patients with a clinical history of seizure is problematic. Unfortunately, we would have been unable to adequately populate our atlas had we not included patients with a history of seizure. To minimize the likelihood that the images of patients with a history of seizure may be abnormal, we included in the final standard data set only those patients in whom the myelination pattern tended to agree with the patterns seen in other patients who did not have a history of seizure. In addition, if the comment field is turned on, the user knows which patients had seizures and can exclude those cases by using the "selected case only" feature. The user also should be aware that the progression of myelination may not be uniform among all patients and that there is a range at which changes become noticeable at MR imaging. Another limitation may be our reliance on the patients’ medical records to obtain gestational age; it is possible that some information in these records was incorrect. Thus, the data set included in the atlas should be used only as a guideline. It is probably most helpful to use the compare mode and to select a range of patient ages that includes the age of interest as well as slightly younger and older ages.
|
Conclusions
|
|---|
We created an interactive digital atlas to facilitate learning about normal changes in the MR appearance of the pediatric brain at less than 4 years of age. Because of the rule set we used in designing the interface, a simultaneous comparison of different pulse sequences or different cases is possible. The atlas was designed to be used with ease during the online interpretation of cases on a picture archiving and communication system. The interactive nature of the atlas allows the reader to gain a rapid understanding of the normal changes in the MR appearance of the pediatric brain and facilitates correct interpretation of potential abnormalities. Instructions for installation and use of the atlas software can be found at http://radiology.seattlechildrens.org/teaching/pediatricbrainatlas.
Top
Abstract
Introduction
