ASL Plugin Usage

Data preparation

Before executing the ASL plugins, the raw ASL (control/tag) and, if available, the calibration (M0) data are to be imported, and preprocessed, as standard 2D (FMR) BrainVoyager projects. During the import of ASL data sets, please keep in mind that different scanners may arrange the control and tag slices and volumes in different ways, causing that these may not be handled correctly by the standard project wizard. Therefore, you may need to use the standard project creator tool and manually enter the correct values for the total number of slices and volumes, the number of skipped volumes, the MOSAIC options and all other options (if needed), etc.. Special attention is required to the number of skipped volumes at the beginning of the series ("Skip first N volumes"), since this may affect the (expected) order of control and tag volumes in the resulting time-series.

NOTE: For FMR projects with ASL data it is possible to invoke the standard FMR preprocessing dialog. In this case, however, only the motion correction option is to be checked whereas all other steps are to be avoided. Specifically, slice scan time correction and temporal filtering are strictly prohibited at this stage because these operations would alter the tagging effect in the data! When needed, both spatial and temporal filtering can be always done in a later stage (i. e. after reconstruction of the perfusion-weighted volume time-series and CBF maps).

Slice mode: ASL "Slice" Plugin operation and settings

The ASL "Slice" plugin can be executed from the Plugins menu. By invoking this plugin, a GUI will pop out (see figure below), where the input FMR with ASL time-series data can be specified. If available, the FMR with ASL calibration (M0) time-series data can be similarly specified. This plugin is only intended for image time-series in the native slice space (FMR/STC) and only generates (and displays) slice maps (MAP), compatible with the input FMR will be generated.

If no calibration FMR is provided, no absolute quantification (aCBF) will be performed. In this case, the only crucial relevant setting is the correspondence between control and tagged volumes in the input ASL series with odd and even time points. Choosing the wrong correspondence will produce an inverted rCBF map, as shown in the examples below where the rCBF map is computed with either the correct (Odd/Even) or wrong (Even/Odd) Control/Tag correspondence.

  

When providing calibration data for quantification, before creating the new perfusion data, it is important to check all settings for calibration in tab 2 (Calibration, see next figure below). First, the ASL sequence type should be checked (PASL or CASL/PCASL), together with the correct TR and TE in ms. For PASL type sequence the typical Q2TIPS paramters TI1 and TI2 should be checked. For CASL/PCASL type sequence, the tagging pulse width and the post-tagging delay should be checked and adjusted if needed. Other important calibration parameters are relaxometric parameters (T1 and T2*) for grey matter and arterial blood tissues (white matter values are not considered to be reliable and so will be the resulting CBF values), and the calibration preferences, which are related to sequence features.

NOTE:  if you do not know the values of these parameters, ask your local MR physics expert, because different or wrong calibration parameters will likely produce unexpected or unreliable absolute CBF maps. Once all these parameters are set, the values can be exported and saved to a file, e. g. "ASL.settings" (default) name. In fact, when running the ASL plugin, the plugin automatically checks for a settings file in the current folder and if it finds such a file the settings are automatically adjusted to these values. Thereby, transferring the settings from one data set to another data set (e. g. another subject), can be obtained by copying the ASL settings file in the current folder of the new data set, but this must be done before running the ALS plugins and anyway verfied (the first time a plugin is called, the current path might be different from what one would expect). When the "ASL.settings" file is not found (or the current folder is not the right one), the calibration parameters are not udpated and must be checked (imported and exported) again if necessary. However, when the settings are found in the current folder, all the calibration values are also written to the log pane, to facilitate a quick verification of their correct value.

 

Volume mode: ASL "Volume" Plugin operation and settings

The ASL "Volume" plugin can be executed from the Plugins menu. By invoking this plugin, a GUI will pop out (see figures below), where (again) the input FMR with ASL time-series data can be specified browsing from the current path and, if available, the FMR with ASL calibration (M0) time-series data can be similarly specified. Please note: it is not necessary that you call the ASL "Slice" plugin if you plan to use the "Volume" plugin next. However, the ASL "Volume" plugin strictly requires that a VMR is loaded in advance to executing it. The current primary VMR will be automatically taken as the "current" VMR for the plugin calculations.

This plugin is only intended for generating perfusion and BOLD image time-series and maps in the volume space, thereby, a standard 2D/3D (FMR/VMR) alignment procedure should be performed between the ASL FMR and the VMR where the initial and final (IA/FA) alignment transformation file (TRF) are generated and saved. If AC/PC and TAL transformations have been also performed, these can aslo be used in the ASL "Volume" plugin to generate volume time-course (VTC) and map (VMP) data in the AC/PC and TAL normalized spaces.

In the first tab of the ASL "Volume" Plugin, besides the input data files, it is possible to define the geometry of the volume space (native, ACPC or TAL) and the voxel resolution of the output data as well as change the interpolation type between linear and cubic spline (default). In the second tab, it is possible to change the file names for the output data. If no calibration FMR is provided, no absolute quantification (aCBF) will be performed. In this case, the only relevant settings are those controlling the relative CBF (rCBF) map and T2*/dM time-series, i. e. the correspondence between control and tagged volumes in the input ASL series with odd and even time points (tab 1), the FMR offset to add to output dM time-series (FMR bias) and the optional output rCBF map scaling to % values (tab 2):

In "Volume" mode, two separate VTC files will be created at the desired isotropic resolution for the perfusion ("dM") and BOLD ("T2s") time-series and two VMP files will be created for the rCBF and aCBF maps in the same reference space (see examples below). The aCBF map will be only created if the calibration FMR (M0) is provided. In this case, the resulting perfusion-weighted VTC ("dM") will have the bias of each voxel time-course set to the corresponding aCBF value and not to the (arbitrary) FMR bias value set in tab 2.

In contrast to the "Slice" mode, it is possible to change the names for the output files (tab 2: output, see figure below) and choose the final resolution in terms of the isotropic voxel size relative to VMR voxels (double: 2x2x2 or triple: 3x3x3). Like for "Slice" mode, if no changes are made to the output file name, the same imput ASL file name will be used with a "suffix" reflecting the type of data. In "Volume" mode, a VTC mask (MSK) can be provided but this should be compatible with chosen output data space and resolution. The calibration settings (tab 3: Calibration) are handled in the same way as for the "Slice" mode (see notes above).

 

 

NOTE-1: The fact that the "calibrated" perfusion-weighted VTC ("dM") has the bias of each voxel time-course set to the corresponding aCBF value and not to an (arbitrary) FMR bias value can be quite important and is specifically illustrated in the figure below. The importance of this assignment stems from the fact that the variability of the CBF on a single time point basis cannot be easily ascribed to an absolute variation of the CBF (due to low SNR) but the mean signal level of the time-course does have a quantitive physical meaning as it represents the mean aCBF estimated over the entire time-series. As a consequence, this quantitive baseline assigns an intrisic physical meaning to what would be the "intrinsic" confound of any (single-study) GLM to which the same VTC would be submitted (the so called baseline confound). In multi-study GLM analyses performed on several runs or subjects "quantitive" perfusion VTCs, by enabling the "confounds" in the GLM overlays, one would thus compare the confounds (baselines) between separate (long) runs and separate subjects in fixed-effects and random-effects statisitcal analyses similar to what is typically done in PET studies using standard GLM contrasts.

NOTE-2: The output VTC files for the T2s (BOLD) and dM (Perfusion) time-series, will normally contain 2 or 3 time points less than the original time-series, depending on whether the input ASL FMR contains an even or odd number of time points. This is because: (i) the starting number of volumes should be adjusted to an even number in order to have an equal number of control and tag volumes before interpolation, and (ii) both the control and tag image time-series must be temporally interpolated and therefore the first and the last volume of the even-numbered ASL time-series must be discarded. Moreover, to avoid interpolation-related border effects in the resulting time-series, the first and the last time point of the output time-series are respectively set to the second and the penultimate time point. As a consequence, especially when designing a new experimental paradigm for an ASL functional study, it is very important to always make sure that the real stimuli do not start or change at the very begin or the very end of the series, e. g. by inserting some "rest" scans at the first few and the last few scans of the series. Of couese, these rest scans must be included in the FMR project as well as taken into account during the preparation of the protocol.