Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: 75x75

MIDAS eNewsletter

 

FEBRUARY, 2013

New and Improved

MIDAS v2: Version 2 is now officially released and available for download. This has gone through several months of testing and becomes final with the completion of some changes in the FITT2 program. Some of the changes were described in the announcement of the beta release in the March 2011 newsletter, and additional major changes in this release are as follows:

TOOLBAR: The MIDAS toolbar has been changed to initially present only the most-used functions, i.e. those dealing with importing the data, running the batch processing, and displaying the results:

If you click on the “MIDAS Fn” or the “MIDAS Functions” button on this toolbar, then another section with the individual processing functions appears:

FITT2: This is a multi-core version of our spectral fitting and thanks to Brian Soher for the challenging development. This provides significantly faster processing relative to the previous version on modern PCs. For example, a fit of the volumetric EPSI data (64x64x32 points with 7358 fitted voxels) for 3 metabolites takes ~30 min for 6 iterations on a 12-core PC. There are also several modifications in the guts of the program, including simpler use of the Viewer mode, and details can be found in the FITT2_Help document.

The FITT2 program uses a different method to compute the values seen in the linewidth map that was implemented in the previous version (FITT17); therefore, with this release we have also changed the linewidth calculation in FITT17 to make these results comparable.

METAFIT: This is a variant of FITT2 that aims to fit low-amplitude signals with additional spatial smoothing applied. It also helps if you are getting single-voxel drop-outs in your fitting result; however, most users will get satisfactory results with FITT2. A description is provided in a separate document.

QUID: Or “Quick Image Display”. This is a neat program to display multi-slice images of any image type, such as this one for NAA:

This is started from the new icon on the toolbar: . Selection of the data must be made at the frame or volume level and operation using the widget is very straightforward. Data can also be read from an Analyze format file. The resultant images can be saved to a JPEG or copied and pasted directly into your document.

BRAIN ATLAS: The lobar-level brain atlas and associated reference image that is supplied with MIDAS has been changed from 90-points (in Y) to 91-points, which was done to make this compatible with atlases that are available from other sources. To perform a group analysis in the PRANA program it is necessary to have all studies processed using the same version of the atlas and registration program. We also have a modified version of the AAL atlas that works with the PRANA analysis, which is available on request.

SID: Options have been added to save spectra in a format that can be read by the VESPA and LCModel spectral analysis programs. These programs require time data, so the inverse FT is also applied. For copying spectra into programs like Word or PowerPoint the spectra can now also be saved in the WMF format, which is more convenient and maintains excellent quality of the plot.

GRAPPA Reconstruction: The FDFT now supports reduced k-space acquisition with spectral GRAPPA reconstruction. There are some restrictions in terms of varying the number of central lines and acceleration factor since these parameters are not available in the raw data taken off GE and Philips instruments. A report describing how this affects data quality is available online in JMRI (see below).

IDL VM License: The IDL Virtual Machine is a free IDL license and this is now supported by the MIDAS software. There are limitations to this license with the main one being that the new FITT2 program cannot be run in multicore mode, so the run-times will be similar to the previous version of FITT (i.e. a few hours). The MIDAS installation includes an option to create a desktop icon to use the Virtual License. Remember the installation also creates an icon for the MIDAS Viewer, which is written in Java, and that also can be run on any computer without the need for IDL.

Multiple SI Series: The MIDAS data organization now supports acquisition of several SI acquisitions in the same imaging session. On import the user will have to assign a unique label to each series, with the first being the standard “SI” and “SI_Ref” labels and the subsequent ones being “SI2” and “SI2_Ref” and so on with increasing numbers for additional acquisitions. To process multiple SI series in the BATCH program the pipeline file would have to be edited to repeat the processing steps for each SI series, with the series identifiers specifically listed as SI2, SI3, etc. Processing using the GUI mode of the individual would be done in the usual manner, just requiring that the specific SI series is selected.

Other Notes:

Image Scaling in v2: The metabolite image quality obtained with v2 is comparable to that of the previous version but there are differences in the scaling and signal normalization, so any analysis of multiple study data should ensure that all studies have been processed with the same software version and of course the same processing parameters. Remember that it’s easy to reprocess multiple studies. Just select all studies in the BATCH program, check the “Delete results of previous processing”, and let the computer do the work.

Processing Files for v2: As always the processing files continue to change and latest versions should be downloaded.

 

MIDAS Tips, Questions, and Answers

Including Other Image Types: Although processing for image acquisitions such as perfusion imaging and DTI is not supported in the MIDAS pipeline these images can still be incorporated and viewed in the display and analysis programs. The MIDAS Viewer provides an excellent method for viewing such combinations of multiparametric image series (example shown below) and the image analysis functions in PRANA can also be applied to these other image types. These other image types can be imported using Analyze-format files (under the “Tools” menu), but first a “parent” image must be imported from the DICOM files to get the necessary parameter information and to assign an appropriate Series label. This parent image must be of suitable quality for the registration programs to work, so for DTI data the B0 image is generally used and the Series is labeled “MRI_DTI”. When the MSREG program is run these additional image types will also be registered to the T1 MRI. The program derives the transformation parameters from the parent image and then applies the transformation to all of the maps in that Series node.

Inter-Series Registration: The standard MIDAS processing pipelines includes an optional processing step, the MSREG program, which can also be run from the toolbar using the  icon. This has three functions:

1.      The position of the SI and SI_Ref with the FOV is modified to match the T1 MRI. The reason for this is that the EPSI acquisition is always positioned at iso-center, and cannot be moved up or down. In the Siemens/Trio this sometimes means that the resultant image is positioned at the top of the FOV, so this shift is corrected for.

2.      A rigid registration is applied to all image series to ensure they are aligned with the T1 MRI. This will correct for between-series motion.

3.      A rigid registration is applied to all image series from a repeat study to align all images with the T1 MRI of the first study.

If additional image series, such as perfusion weighted images or DTI results are imported into MIDAS, then these image series will similarly be aligned by the MSREG program.

TIP: For studies on phantoms the MSREG program will not work for objects such as a sphere or cylinder if there is not enough structure to determine the registration. For these studies the MSREG processing step should be skipped. While this doesn’t affect the processing there can be issues with image display for data acquired on Philips instruments, since the required SI spatial information is not available from the raw data. A better solution is to put additional objects in the FOV that will show in the MRI and water-reference SI and enable the registration algorithm to work.

MIDAS Viewer Layouts: The MIDAS Viewer now provides options to save and restore image layouts. This is available under the “File / Window Layout” menu option. For example the image below shows a layout that includes MRIs, metabolite, perfusion, and diffusion images, and this layout was saved as the “Default for Current Project”. Following this, for any subject opened from this project the initial window layout will be the same. You can also save layouts under a specific filename and reuse them across different projects.

Another new feature in the Viewer is the “Image / Zoom” option, which brings up a small widget (visible at the top in the image below) that you can use to zoom all images, centered at the current position of voxel selection, as in the example below:

Thanks go to James “JC” Norman for the continued development of the Viewer.

Rearranging Data: MIDAS uses a simple scheme to keep track of the various “Projects” and “Subjects” within each Project, which means that the user does not need to keep track of where the data is stored when using the MIDAS Browser. Occasionally, however, data may need to be moved to a different location and then the information in the MIDAS data organization needs to be changed. If a whole project is moved then it is first necessary to remove the pointer to the old location, which is done in the Importer. To remove the old project, make this the Working Project and then select “File / Remove Project”. The Importer always checks if the projects are valid on start up, so another way to do it is to first move the project before starting Importer, and the program will prompt the user to ask if the old entry should be removed. To include the new project location go to “File / Import Existing Project” and browse for the XML file containing the project information, which is named something like “xxx_Project.xml”, where xxx is the name you gave it when it was initially created.

If you want to just remove a single Subject, then this is done in the MIDAS Browser by right clicking on the Subject and selecting the “Remove subject from project” option (see Figure). This does not delete the data and that need to be done by the user.

For more ideas on rearranging projects also check out the section on Virtual Projects in the March 2011 newsletter.

EPSI

Reduced K-space: We now have short-TE versions of the EPSI acquisition running on MR instruments from GE, Philips, and Siemens. These all include reduced-k-space acquisition with GRAPPA reconstruction. The spectral quality for a 13-minute acquisition, TE=17.6 ms and 8-channel detection (Siemens/Trio) is excellent (single voxel spectrum shown on the right). Our initial experience with patient studies suggests that the GRAPPA reconstruction may be more sensitive to motion and optimization of the method continues.

Information on the GE and Philips EPSI implementations can be obtained from Dan Spielman and Peter Barker respectively.

Publications

and

Example

Data

 

Many thanks to our collaborators for the following reports:

·        B. E. Levin, H. L. Katzen, A.A. Maudsley, J. Post, C. Myerson, V. Govindaraju, F. Nahab, B. Scanlon, and A. Mittel. Whole-brain proton MR spectroscopic imaging in Parkinson’s Disease. J. Neuroimaging, Published Online Dec. 10 (2012).

·        C.J. Stagg, S. Knight, K. Talbot, M. Jenkinson, A.A. Maudsley, M.R. Turner. Whole-brain MRI spectroscopy measures are related to disability in ALS. Neurology, Jan 16. [Epub ahead of print] (2013).

·        G. Verma, S. Chawla, S. Wang, S. Sheriff, J.H. Woo, L.B. Elman, L.F. McCluskey, M. Grossman, E.R. Melhem, A.A. Maudsley, H. Poptani. Whole brain analysis of Amyotrophic Lateral Sclerosis using echo planar spectroscopic imaging. Radiology, Published online January 29, (2013).

·        B. Roy, R.K. Gupta, A.A. Maudsley, R. Awasthi, S. Sheriff, M. Gu, N. Hussain, S. Mohakud, S. Behari, C.M. Pandey, D. Spielman, J.R. Alger. Utility of multiparametric 3T MRI for glioma characterization. Neuroradiology. In Press (2013).

·        M. Sabati, J. Zhan, V. Govind, K.L. Arheart, A.A. Maudsley. Volumetric MR spectroscopic imaging with reduced k-space acquisition: Variability and pathologic detectability. In Press, JMRI, (2013).

An example result for a comprehensive multiparametric MR protocol of brain cancer, obtained from Dr. Gupta and colleagues, is shown below. This is for a Glioblastoma Multiforme, obtained at TE=70 ms. Observations include the extensive MRS coverage of the brain, including the cortical surface in the region of the tumor, and mapping of Glycine+myoInositol (Gly) and lipid+lactate (Lac). For spin-echo detection at TE=70 the signal at 3.5 ppm is heavily weighted to Glycine and this is fit as a singlet.

 

Developer’s Corner

Adding Toolbar Utilities: The MIDAS toolbar contains a Utilities section, which is opened from the  icon. If you have some custom-made tools or favorite programs you can add these to the list. This is done by editing the file:

“C:\MIDAS\bin\MIDASTools_Misc_Tools_List.txt”  (Or the drive/directory used for your MIDAS installation).

This text file contains a simple comma separated list that defines: a) An IDL batch procedure to run, and b) a description of that program to put in the list, e.g.:

@runmakesidata, SI Simulation

In this case, the IDL command @runmakesidata will be issued and the line “SI Simulation” appears in the list shown on the widget. If you have a non-IDL program then the IDL procedure that is called can start up the other program, e.g. using the SPAWN command.

Documenting BATCH Pipeline Files: The new version of the BATCH program allows for some text lines to be added to include some documentation to the pipeline file. From the initial widget, go to the “Edit” mode and you will see a File Information section in the upper right. An example is shown below:

 

 

Meeting Announcement

Stefan Posse is organizing an ancillary meeting at ISMRM 2013 to discuss recent advances in high-speed MRSI methodology for clinical research. The meeting will be held on Monday April 22 at 6:30pm, after the end of the ISMRM sessions. Room assignment is to be announced.

 

People

Jiping Zhan, who has kept our software development organized since 2010, has left us to join a company in Kansas. We wish him success in all future endeavors.

 

More Midas

Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: Description: http://upload.wikimedia.org/wikipedia/commons/thumb/0/01/Cochranella_Midas.jpg/220px-Cochranella_Midas.jpg Cochranella midas (Image from Wikipedia)

 

Andrew Maudsley, Feb. 2013

This newsletter is aimed at providing information to the developers and users of the MIDAS software package. If you would prefer not to receive notification of these reports just let me know (amaudsley@med.miami.edu).