6/30/2006 DGM/HBChapter 8: Data Analysis: Surface-Based

CHAPTER 8: POPULATION PROPORTION MAPS

The statistical maps used for ROI definition derive from F-statistics applied in volume space. Voxels that pass threshold are converted to 1’s and al others are set to 0’s. These files are labeled as *bin* files. There is one file per task. The object of the following steps is to summate the participant bin files in standard space, which produces a map wherein each node shows how many participants had significant activity at that location.

A.Getting started.

  1. All subject data must be projected to the PALs standard mesh before starting.
  2. Create Visualize_LH and Visualize_RH directories in the study_atlas_registereddirectory. The latter should have been created when the participant surface mesh files were deformed to the standard PALS atlas.
  3. Copy the files in caret/Visualize_LH and caret/Visualize_RH to the LH and RH hemispheres, respectively.
  4. Rename the .spec file in each directory to be in the form: Human.PALS_B12.B1-12.hemisphere.study.73730.spec
  5. Open caret>FileSet the caret current directory to study/study_atlas_registered/selected hemisphereOpen
  6. File>Open Spec File>select

Human.PALS_B12.B1-12.hemisphere.study.73730.spec>Open

  1. In the Specification File:…dialog window
  2. Remove unnecessary files by clicking on the ‘x’
  3. Select All > Load>New Spec Only
  4. File > Open Data File>in Open Data Filedialog window…
  5. Change file type to Border Projection Files (*.borderproj) and select *OP_MPM*
  6. Change file type to Border Color Files (*.bordercolor) and select *OP_MPM*file[DGM1]
  7. Change file type to Paint Files (*.paint) and select each of them
  8. Change file type to Area Color File (*.areacolor) and select each of them
  9. Change file type to allow you to load any other specific files relevant to your study that have not been loaded.
  10. On the main Caret window select D/C (Display Control)Page Selection>Border>Nametab>All Off button
  11. Change Model to FLAT…

B.File>Open Data File>in Open Data Filedialog window change File type: Metric Files

  1. Navigate to the study_atlas_registered directory and select one of the deformed metric files that are the binary coded Fzstat maps (e.g., deformed_stimtest18L.Fzstat_bin.metric);make certain the selection is hemisphere specific.
  2. For Open Options box:
  3. Check‘Append to Current File’
  4. Check ‘Add to Specification File’
  5. After the selected file appears in the File name area, click Open
  6. in Copy Filedialog window: “Would you like to copy the file to the current directory? > No
  7. in Choose Columns to Loaddialog windowFile Comment window> select Append>OK
  8. Repeat steps listed above for each participant’s deformed F-statistic maps until all cases have been appended to the spec file.
  9. Save data file study_Fzstat_bin.metric

C.Creating a metric file that contains the summed contributions to each node from all participants for each task.

  1. If study_Fzstat_bin.metric is not loaded, load it now.
  2. Attributes>Metric>Mathematical Operations
  3. inMetric File Mathematical Operationsdialog window
  4. Metric Columns set Column A to the first participant’s Fzstat_bin data for a particular task (e.g., deformed stimtest27_N_Fzstat_bin.4dfp.ifh)
  5. Set Column B to the next participant’s data for the same task (e.g., deformed stimtest28_N_Fzstat_bin.4dfp.ifh)
  6. Set Result window to Create New Column > set New Column Name to: study_task_ppm

This column will show the number of participants who had significant activity per standard mesh surface nodes.

  1. in Operations box select A+B
  2. select Apply button
  1. Repeat as follows:
  2. Set column A to next data set for another participant
  3. Set column B to study_task_ppm
  4. Set Result to study_task_ppmfrom drop-down list.

(a)Otherwise a new column gets created each time. BAD!!!!

  1. select Apply button
  1. Continue until all participant data is summed in the study_task_ppmcolumn.

D.Save ametric file when all participants’ data has been summed for each task. This avoids losing previous column calculations for finished tasks.

  1. File>Save Data File>File type: Metric Files> File name: study_ppm.metric.

E.Repeat steps C and D to create a *ppm column for each task.

  1. When all *ppm.metric columns have been assembled and saved,select CloseMetric File Mathematical Operations dialog window.

F.Converting values in each study_task_ppm column to 1’s and 0’s.

  1. The study_ppm.metric file must be loaded into caret.
  2. Display ppm metric column of interest (e.g., stimtest_N_ppm) to follow selection of nodes with the following steps so that it is seen on the main window display.
  3. Surface>Regions of Interest Operations.This opens the Surface Region of Interestdialog window with several tabs, each with windows that have drop-down menu items. Start with tab Query. Ignore other tabs.
  4. Set Selection MethodtoNodes with Metric
  5. Choosing the latter option opens Threshold Low and High windows[DGM2]

(a)set Threshold Low the desired threshold for the study (e.g., 6 for 50% of 12 participants)

(b)setHigh setting to be greater than the number of participants

(c)select All Nodes Within Threshold (this is the default selection)

  1. In the drop-down menu below the Select Nodes button, select Normal Selection
  2. click Select Nodes
  3. Operate on Selected Nodes box

(a)set Operation to Assign Metric Column for Selected Nodes

(b)set Metric Column to Create New Column

(c)enter name to the right of the Metric Column dropdown:study_task_ppm_bin

(d)set New Value to1

(e)click Assign Metric button

  1. File>Save Data File>File type: Metric Files> File name:study_ppm_bin.metric
  2. Close Surface Region of Interestdialog window
  1. Repeat steps F.2-F.5 for each task specific column

G.Creating a composite map of all task data.

  1. Attributes>Metric>Mathematical Operations
  2. inMetric File Mathematical Operationsdialog window
  3. Metric Columns >set Column A to column for first task (e.g., stimtest_N_ppm_bin)
  4. Set Column B to column for the next task ( e.g., stimtest_P_ppm_bin)
  5. Set Result window to Create New Column > set New Column Name to: study_sum

This column will be the summation of the binary coded task columns.

  1. in Operations select A+B
  2. click Apply button
  1. Repeat as follows:
  2. Set Column A to the column for the next task (e.g., stimtest_SD_ppm_bin)
  3. Set Column B to study_sum
  4. Set Result to study_sumfrom drop-down list.

(a)Otherwise a new column gets created each time. BAD!!!!

  1. clickApply button
  1. Continue until all binary coded task columns are summed in the study_sum column.
  2. Select A/B in the Operations box
  3. Set Column A to study_sum column from the drop-down menu
  4. Set Column B to the same study_sum column from the drop-down menu
  5. Set Result to Create New Column> study_logical_OR
  6. click Apply button to divide the column data by itself to create a logical OR where every node has a value of 1 where any task contributed to that nodes activation.
  7. click Close button to exit Metric File Mathematical Operationsdialog window
  8. File>Save Data File>File type: Metric Files> File name: study_logical_OR

CHAPTER 8: DEFINING ROI

PALS has an array of bounded regions based on cytoarchitecture and functional subdivisions. A bounded region is defined by the nodes within the border. However, the standard Caret borders do not necessarily reflect functional data collected ina study. The first step, therefore, is to draw borders around the data obtained in a study. Next is to limit the nodes within these borders to boundaries associated with defined cytoarchitecture, other functional subdivisions from prior studies or anatomy. There are four circumstances associated with assigning functional data to a set of surfaces nodes that will constitute a region of interest (ROI). (1) All nodes for a ROI fall within a single, previously defined Caret border. (2) Different clusters of nodes associated with subparts of a Caret defined cytoarchitectonic area, e.g., multiple ROI within an established border. (3) Multiple borders intersect the same nodes and these nodes must be assigned to only one of the borders.(4) A combination of conditions 2 and 3.

A.The steps outlined below are executed after displaying in CARET the model:Flat Human.colin.hemisphere.RegToPALS_B12.Flat.CartSTD.clen.73730.coord.

  1. Once this model is loaded in the main CARET window, use the Display Control (e.g., button D/C) option and Page Selection menu item: Border.
  2. In the dialog window for the latter choose Name tab>All Off>toggle on named borders for your study (e.g., selected Brodmann areas and any previously identified functional borders (e.g., V1v).
  3. Return to the Page Selection menu and select Overlay/Underlay – Surface.
  4. In the menu: “Surface Coloration Applies To” select All Surfaces.
  5. In the menu: “Surface Attributes” set the options in the named columns as follows:

(a)Primary Overlay Metric

(b)Underlay > Shape

(c)Pull-down menu associated with Metric to the metric file that contains your statistical parameter map

(d)Pull-down menu associated with Shape to AVERAGE B1-B12 HEMISPHERE DEPTH

(e)Leave the remaining items at the default settings.

  1. Return to the Page Selection menu and select Metric.
  2. Adjust the Settings options to display your statistical parameter map on the Flat surface.
  3. Move the Display Control dialog window aside

B.Drawing borders foryour study results is best done after obtaining a “population proportion map.” However, you can use any summary statistical map (e.g., average z-scores) that has been rendered on the standard PALS surface.

  1. On the main Caret screen select from the tool bar: Layers>Borders>DrawBorders
  2. In the Draw Bordersdialog window enter the following:
  3. In the “Attributes” box and edit window labeled Name, enter a border name.
  4. In the “Type” box select Closed (Boundary).
  5. In the “Closed Border Assignment” box toggle “Assign Paint Identifiers to Nodes Within Border”
  6. Leave the remaining options at the default settings and click the Apply button.
  7. The previous step opens a Create Border Color dialog window
  8. Slide the color scales to find a color and click OK.
  9. An alternative is to use the Advanced Color Selection >toggle one of the boxes for Basic colors >click OK > click OK in the Create Border Color dialog window.
  10. Move Draw Bordersdialog window aside.
  11. Draw your border in main window by holding the left mouse button.
  12. Shift+left mouse click when finished with a border
  13. Repeat steps 2-4for each new border.

C.Projecting the borders to all surfaces.

  1. On the main Caret screen select from the tool bar: Layers>Borders>Project Borders.
  2. In Border Projection dialog window selectNearest Tile>OK
  3. Review borders on fiducial or inflatedmodels. However, borders can only be edited on the flat map. See Caret Tutorials for options.

D.Identifying the nodes with non-zero metric values in established borders (e.g., PALS or uniquely drawn borders). All of the following steps are performed within the Surface Region of Interestdialog window, which is opened from the main Caret screen tool bar: Surface>Regions of Interest Operations…

  1. Before any ROI can be defined, all of the activated nodes defined in a particular metric file must be selected.
  2. In the “Region of Interest Node Selection” box obtained from the Query tab:

(a)Drop-down menu labeled ‘Selection Method’> select ‘Nodes With Metric’

(b)Edit-window labeled ‘Threshold Low’set to 1

(c)Edit-window labeled ‘Threshold High’ set to 50000

(d)Drop-down menu located below button labeled‘Select Nodes’ changed to ‘Normal Selection’

(e)Click Select Nodes (Inspect the main window to note that all nodes with a value of 1 have been selected.)

  1. All nodes for a ROI fall within a single, previously defined Caret border.Selecting nodes identified uniquely within a single border (e.g., one just drawn or a Caret defined cytoarchitectonic area). The following steps are executed after all metric file nodes are selected (see D.1 above). Note that steps D.1.i.(a)-(d) must be performed before specifying the nodes for a particular ROI.
  2. Drop-down menu labeled ‘Selection Method’ >select ‘Nodes Within Border’

(a)Click the button: ‘Select Name From List…’ (This opens a Border Names Selection dialog window) > highlight a border name that exclusively contains the nodes of interest >OK. Sometimes it is faster to ‘Select Border with Mouse…’, which allows one to point to any edge of a visualized border on the Flat model.

(b)The next step is a logical ‘AND’ that confines the specified nodes in the entire metric file to those within the selected border. In the drop-down menu located below button labeled ‘Select Nodes’ select ‘And Selection (Intersection)’

(c)Click Select Nodes button (inspect the main window to note that all nodes with a non-zero value were selected within the named border)

(d)Skip to step E below to create the ROI.

  1. Different clusters of nodes associated with subparts of a Caret defined cytoarchitectonic area, e.g., multiple ROI within an established border. The following steps are executed after all metric file nodes are selected (see D.1 above). Note that steps D.1.i.(a)-(d) must be performed before specifying the nodes for a particular ROI.
  2. Selecting the first subdivision.

(a)Drop-down menu labeled ‘Selection Method’ > select ‘Nodes Within Border’

(b)Click button labeled ‘Select Name From List…’ (This opens a Border Names Selection dialog window) > highlight the border that contains the two subdivisions>OK

(c)Drop-down menu located below button labeled ‘Select Nodes’ > select ‘And Selection (Intersection)’

(d)Click button labeled ‘Select Name From List…’ > highlight the named border that is one of the subdivisions >OK

(e)Click ‘Select Nodes’ button (inspect the main window to note that all nodes within named subdivision border were selected with a non-zero value)

(f)Skip to step E below to create the ROI.

  1. Selecting the second subdivision ROI.

(a)Repeat steps C.3.i.(a) and (b).

(b)Click button labeled ‘Select Name From List…’ > highlight the ROI subdivision selected first in step C.3.i. >OK

(c)Drop-down menu located below button labeled ‘Select Nodes’ > select ‘And-Not Selection’

(d)Click ‘Select Nodes’ button (inspect the main window to note that all nodes within named subdivision border were excluded and that the remaining non-zero value nodes are now in the second subdivision.)

(e)Skip to step E below to create the ROI.

  1. Multiple borders intersect the same nodes and these nodes must be assigned to only one of the borders.
  2. The selection process is a series of logical steps needed to intersect and exclude selected nodes as specified by the Selection Method (e.g., Nodes With Metric, Nodes Within Border, Nodes with Paint) and the appropriate application of the Select Nodes options (e.g., Normal Selection, And Selection (Intersection), And-Not-Selection, etc.). Proceed from the larger, more inclusive node specifications to progressive smaller regions by appropriate combinations. Some experimentation and trial-and-error is needed to get the right combinations.

E.Creatingan identified ROI using drop-down menus located in box labeled ‘Operate on Selected Nodes,’which is at the bottom of the Surface Region of Interestdialog window. The following steps must be done immediately after specifying the non-zero nodes for each ROI.

  1. Drop-down menu labeled ‘Operation’ > select ‘Assign Paint Attributes to Selected Nodes’
  2. Drop-down menu labeled ‘Paint Column’> select ‘Create New Column’

NOTE: Select ‘Create New Column’ option only for the first ROI column saved for study_ROI_hemisphere file (e.g., stimtest_ROI_RH). Subsequent ROI column entries start with the ‘Paint Column’ value set to the original study_ROI name.

  1. In the Edit-window located immediately to the right, enter study_ROI_hemisphere
  2. In the Edit-window labeled ‘Paint Name’enter your named ROI (e.g., BA3_fingers)
  3. Click button labeled ‘Assign Paint’
  4. In Create Area Color dialog window slide the color scales to find a color ( or select ‘Advanced Color Selection’)and click OK
  1. Inspect the assigned paint colors on the Main display window after setting Primary Overlay to Paint and column to your study_ROI_hemisphere file
  2. When repeating the above for the next ROI, select study_ROI_hemisphere for the Paint Column from the drop-down menu.
  3. Insert instructions to remove other Paint Columns[DGM3].
  4. File>Save Data File>File type: Paint File> File name: study_ROI_hemisphere
  5. File>Save Data File>File type: Area Color File> File name: study_ROI_hemisphere

CHAPTER 8: EXTRACTING FUNCTIONAL DATA ASSOCIATED WITH DEFINED ROI

The Paint file is a text listing of all 73730 nodes and the ROI assigned to those nodes.The Metric files contain timecourse data columns for each node. Caret stores these as binary coded files that need to be translatedinto ascii (text format).

In Windows Explorer: make folders in the CARETstudy_atlas_registered directory: timecourse_LH and timecourse_RH. Copy all the deformed metric files (e.g., those in PALS space) containing the timecourse data to their respective hemisphere timecourse folder.

Open a DOS command prompt. (Start Menu > Run > cmd > enter)

Navigate to the caret/bin directory and type the following:

caret_file_convert –text the relative path to the directory containing the timecourse data\*

caret_file_convert –text ..\study\study_atlas_registered\timecourse_LH\*

Once the text files have been created, you need to use a Matlab script created by Donald McLaren.

Open Matlab

Double click ‘transform_data.m’

Modify line 3, which is:

X= textscan(fid, ‘%f %f %f %f %f %f …..’, 73730, ‘headerlines’, X)

A %f must be present for every column of data + 1.

If all participants in a study are the same, then a generic %f entry can be used. Otherwise each use of this command line has to be self generated.

X must be replaced by the line number of the last row of before data begins. Line contains ‘tag-BEGIN-DATA’.

A simple way to determine this number is to open the Metric.file into Excel.

CHAPTER 8: EXTRACTED FUNCTIONAL DATA IN EXCEL

CHAPTER 8: CROSS-CORRELATION ANALYSES USING DEFINED TIME COURSE TEMPLATES

CHAPTER 8: MANOVA ANALYSES OF TIME COURSES

1

[DGM1]Master location???

[DGM2]Word choice

[DGM3]Missing text.