New graphical user interface addresses MR reconstruction

Susceptibility-weighted MR imaging provides high-resolution, distortion-free blood oxygen level-dependent data for assessment of cerebral veins, blood products, and brain lesions. Currently, however, reconstruction of susceptibility-weighted imaging data is not implemented on all MR systems.

A new German paper introduces a graphical user interface for image reconstruction and data analysis in susceptibility-weighted (SW) MR imaging (RadioGraphics 2008;28:639-651).

"We have designed an imaging system-independent cross-platform stand-alone (offline) application suite called GUIBOLD for SW imaging data reconstruction," said Andreas Deistung of the Institute for Diagnostic and Interventional Radiology at Friedrich-Schiller University Hospital in Jena.

The main job of GUIBOLD is to streamline postprocessing steps involved in SWI data reconstruction and visualization, such as computation of combinations of magnitude and phase images, Deistung said. GUIBOLD also provides image projections over different stacks of adjacent slices to obtain so-called MR venograms.

"Although some manufacturers of MR systems provide SWI reconstruction, their postprocessing is not freely available or they tend to limit user control," Deistung said.

GUIBOLD exploits a free runtime utility (IDL Virtual Machine) to make it platform-independent.

The system reconstructs either raw SWI data or DICOM data. Due to different and rapidly changing raw data formats of various MRI vendors, however, the researchers focused primarily on DICOM data.

"The GUI provides different methods to eliminate unwanted phase wraps on phase images, combines magnitude and phase information by employing different phase masks, and improves the visibility of venous structures by using a median filtering approach," Deistung said.

GUIBOLD visualizes the reconstructed image data (magnitude, phase, and susceptibility-weighted image) in different specifiable orientations (axial, sagittal, and coronal) and projections (maximum, minimum, and average) with various rendering techniques, such as isosurface representation.

Region-of-interest analysis tools are also integrated. The user interface allows exporting of the processed data in DICOM and other formats, such as JPEG, PNG, or BMP.

"With this flexibility, we feel that GUIBOLD is able to fill a gap between researchers experienced in SWI and clinical users who so far have had to rely on rather limited MR imaging software," Deistung said. "With GUIBOLD, we have the opportunity to react quickly to new SWI developments and to share these postprocessing methods with the user community."

GUIBOLD and associated test data are available free on the researchers'

website behind the Downloads button. GUIBOLD execution requires IDL Virtual Machine, also

available free.