---

---

FMRI analysis is performed on temporal data vectors (time-course) from the imaged brain. The dataset contains empty space and the spatial ROI must first be obtained. From these voxels coordinates, data vectors will be extracted for analysis. Usually the entire brain is used and a simple image segmentation method, such as binary thresholding, can be used to separate the brain image from the background. Figure 29 shows the new map with the Binary Thresh Cube module, found in the IBD Analysis / Image Processing package, and connected to the Mean Cube Image module.

Notice in the Output Slots Connections window, that the output of the Mean Cube Image, the slot named mean_cube_image is connected to the input slot image_data in the Image Cube module and the slot image_data in the Binary Thresh Cube module. Scopira does not make copies for each slot, the data provided to the modules is the same, and the engine ensures that the modules do not change the input slot data. That is, input slots are read only. Refer to the User's Guide for further information on Scopira's data sharing mechanism.

Figure 30 shows the Input Slot Properties for the Binary Thresh Cube. The image intensity of the brain voxels are much higher than the background, so a binary image/mask of the brain can be obtained by displaying only the voxels that have intensities above a certain pre-determined threshold value. The mean fMRI volume, with improved signal-to-noise ratio, is used as the image source, and the input slot image_data is made sticky. This way, once the dataset is loaded, and the mean image volume calculated, the user can interactively try different threshold values without having to run the map every time the Binary Thresh Cube needs to run.

Double clicking the Load Image Quad module loads the dataset and propagates data to the new Binary Thresh Cube module. Figure 31 shows the proponent window for the Binary Thresh Cube module. It displays the binary image of the brain. All the white voxels will be used for analysis. The user can interactively change the threshold value, and threshold type. Refer to the User's Guide for further details on this particular module. Entering a lower value in the Hi: entry field lowers the threshold value and more voxels will be displayed. However, until Apply and Run is pressed, the module will not produce an output. Only the display is updated until Apply and Run is pressed, then data (the binary image, as a double-cube data type) will be propagated via the output slot connections to other modules that use the binary image.

From the binary multi-slice image volume, the (x,y,z) coordinates of the ROI can be calculated by the Binary Cube ROI module, located in the IBD Analysis / Image Processing package. Figure 32 shows the new map with the Binary Thresh Cube output connected to the Binary Cube ROI module's image_data input slot. The ROI module produces a list of slice points ROI, that is, for a given z slice, it keeps track of the (x,y) coordinates for every voxel with a non-zero intensity value. The data type of the output slot, image_roi, is ibdkit-points-roi-vec, refer to the User's Guide for documentation on this data type.

Given a multi-slice dataset, and slice-based ROIs, the module ROI Data Vecs Quad, found in the IBD Analysis / Data Generation package, will extract a set of data vectors that can be used in further analysis. Figure 33 shows the new map with the ROI Data Vecs Quad module added. Note that it accepts three inputs. The volumetric temporal dataset, from where data vectors will be extracted, is required, as well as the list of slice-based ROI (x,y,z) points. See the User's Guide for full documentation on the Roi Data Vecs Quad module.

When connecting to a slot, Scopira will ensure that only slots with compatible data types are coupled. As such, even though ROI Data Vecs Quad has three input slots, when connecting the output of the Scale Quad module, a double-quad data array, only the image_data(double-quad)(req:1) slot will be available as shown in Figure 34. The output double-quad of Scale Quad was selected for connection (by right-clicking and selecting the output slot named scaled_data), then the eligible input slots of ROI Data Vecs Quad were displayed by right-clicking on the target module.

Once the connections are complete, the Input Slots Connection window is updated as shown in Figure 35. The ROI Data Vecs Quad module also needs the ROI coordinates. Figure 35 shows the updated map with the output slot of the Binary Cube ROI, the list of slice-based ROI coordinates named image_roi, connected to the img_roi(ibdkit-points-roi-vec) input slot for ROI Data Vecs Quad.

The vec_pts(int-vec) slot is left unconnected. It is not a required slot. If no data is provided, the module will still run as long as the required slots have data pending. This particular slot is used to specify which time instances to omit from the fMRI analysis. In fMRI data, sometimes the first few time instances (image acquisitions) are not used, as they could be saturation scans, meaning the image intensity is artificially high, as the scanner stabilizes at the start of an image acquisition sequence. If no data is provided, the module will use all the time instances to generate data vectors for analysis.