(Wednesday, 8/1/2018) 10:00 AM - 10:30 AM
Room: Exhibit Hall | Forum 8
Purpose: A single 3D MR sequence can potentially replace multiple 2D sequences acquired in multiple planes for a reduction in overall scan time. However, to view at multiple orientations, 3D sequences must be acquired at high isotropic resolution, and often suffers from low SNR and/or long acquisition times. Here, we present a protocol for acquiring, reconstructing, and reformatting 3D data of the spine for diagnostic viewing in arbitrary planes.
Methods: Sagittal 3D T1 Cube sequences were acquired in a clinical spine protocol. All sequences were acquired with 1mm isotropic resolution, but reconstructed at approximately 0.5mm isotropic voxels to minimize image degradation from reformatting. All reconstructions and reformats were performed offline using MATLAB. An image set of thicker, 2.5mm slices was created by reformatting the isotropic high resolution magnitude images. A third set of images was created by reconstructing a complex image volume for each channel, then reformatting to 2.5mm slices by averaging slices of complex data before sum of squares channel combination. Difference images were calculated, and all images were analyzed for noise and image quality.
Results: Native isotropic resolution images were of diagnostic quality but appeared relatively noisy compared to thicker slice 2D images. Magnitude reformatted images improved overall SNR of the image, but exhibited noticeable background noise and haze that reduced contrast of hypointense signal regions. Images reformatted with complex averaging presented similar SNR in high signal regions but reduced noise in air and low signal regions. This was confirmed with difference images showing the greatest reduction of noise in the background.
Conclusion: Reformatting isotropic 3D data into thicker slices by averaging complex data can improve visualization of 3D MR data sets. While all clinical reformatting software will perform magnitude averaging on magnitude DICOM images, complex averaging must be performed offline with raw data exported from the scanner.
Keywords
Taxonomy
IM- MRI : Image Reconstruction
Contact Email
