You are here: Processing Images > Stitching 3D Images

Stitching 3D Images

The Stitch 3D Images feature lets you combine, or stitch, multiple volumetric datasets acquired from different modalities or from the same modality with an overlap into a single 3D dataset. The example below shows the Stitch 3D Images box that defines the volume of the new dataset that will be created when the top dataset (shown in cyan) is stitched to the bottom dataset (shown in yellow).

2D view of 3D images prepared for stitching

Acknowledgments: Datasets acquired and processed at the Los Alamos National Library by members of the E-6 Group. Acquisition of data done on a SKYSCAN 1272 from Bruker.

Video Presentations and Tutorials

The video presentations and tutorials in this section provide in-depth instructions for stitching 3D images into a single volumetric dataset.

Settings for Stitching 3D Images

The settings for stitching 3D images are available in the Stitch 3D Images dialog. Select the images you need to stitch in the Data Properties and Settings panel and then choose Stitch 3D Images in the pop-up menu to open the dialog which is shown below.

Stitch 3D Images dialog

A. Datasets list B. Settings

Settings

The following settings are available for stitching volumetric datasets.

Output parameters
	Description
Reset Box	Lets you reset the reference box that determines the dimensions of the output dataset. The initial size and orientation of the reference box, which appears automatically as the Stitch 3D Images Box in the Data Properties and Settings panel, is modeled on the selected datasets to provide the best initial fit.
Out-of-bounds value	Lets you assign a data value to voxels that are out-of-bounds. Out-of-bounds voxels are voxels that are included in the Stitch 3D Images box, but do not intersect with any of the voxels of the images selected for merging.
Blending	Lets you choose a blending method for overlapped areas as follows: Max… The maximum intensity values of all datasets selected for the stitch operation will be applied to the voxels of the output image. Mean… The mean intensity values of all datasets selected for the stitch operation will be applied to the voxels of the output image. Min… The minimum intensity values of all datasets selected for the stitch operation will applied to the voxels of the output image. No blending… In this case, voxel values will be assigned based on the order in which the datasets appear in the Stitch 3D Images dialog. This means that the values of the dataset at the top of the list will be applied directly to all corresponding voxels in the Stitch 3D Images box. If there are not intersecting voxels in the dataset at the top of the list, then the values of the dataset below will be applied. If there are no intersecting voxels in any dataset, then the out-of-bounds value will be applied. If required, you can re-order the datasets in the list by dragging items up or down, as shown below. Weighted… Intensity values will be smoothly adjusted between all datasets selected for stitching by applying a weighted average of the closest 2x2 neighborhood. This is the recommended Blending option for most cases.
Interpolation	Lets you choose the type of interpolation applied to stitch the selected datasets as follows: Nearest… This basic interpolation scheme requires the least processing time because it only considers one pixel — the one closest to the interpolated point. Linear… Considers the closest 2x2 neighborhood and then takes a weighted average of these four pixels to arrive at its final interpolated value. Tricubic… This interpolation scheme goes one step beyond Linear by considering the closest 4x4x4 neighborhood of known pixels. Will result in smoother looking images than Linear.
Resolution	Lets you choose the resolution of the output dataset as follows: Most precise… Uses the smallest voxel size of all datasets selected for the stitch operation. User defined voxel size… Lets you choose the required values for spacing in the X, Y, and Z axes, which determines the voxel size. You should note that increasing spacing along an axis will reduce the number of pixels along that axis by downsampling. Conversely, decreasing the spacing along an axis will increase the number of pixels along that axis by upsampling. Note User-defined voxel size selections are limited to a factor of 10 of the original image data. User defined matrix… Lets you choose the resolution of the output dataset as the number of pixels in the X, Y, and Z axes. You should note that increasing the number of pixels along any axis will decrease the corresponding spacing value, while decreasing the number of pixels along an axis will increase the corresponding spacing value. Note In all cases in which the resolution of the original image data is changed, the geometry of the Stitch 3D Datasets box — which is calculated as the [(number of pixels)*spacing] — will be maintained.

How to Stitch 3D Images

Load all of the datasets that you plan to stitch.
Preview the datasets in the same scene and in multiple orientations, recommended (see Multi-Image Visualizations).
If required, adjust the position of one or more of the datasets that you plan to stitch so that any overlapped area is properly aligned.
You can use the tools on the Translate panel to manually translate or rotate a dataset (see Registering Images Manually) or you can automatically register datasets with Dragonfly's feature-based registration workflows (see Registering Images Automatically).
Select all of the datasets you plan to stitch in the Data Properties and Settings panel, right click, and then choose Stitch 3D Images in the pop-up menu.
The Stitch 3D Images dialog appears. In addition, a Stitch 3D Image box, which is created large enough to encapsulate all of the selected channels, automatically appears in the current scene views.
Note If required, you can also drag a dataset from the Data Properties and Settings panel to the Stitch 3D Images dialog.
Adjust the size of the Stitch 3D Images box, if required.
You can edit the size of the box and its position in any 2D view or in the 3D view by simply dragging the available control points (see Editing Shapes).
Click the Reset Box button to reset the box to its original size, which should encapsulate all of the datasets you selected for stitching.
Choose the settings that you want to apply to the output dataset, as described in Settings.
Out-of-bounds-value… If required, select the data value that will be assigned to voxels that do not intersect with the datasets selected for merging, but that are included within the Stitch 3D Image box.
Blending… Choose a blending method for overlapped areas — Max, Mean, Min, No blending, or Weighted — as required. In most cases, Weighted can provide the best results.
Interpolation… Choose an interpolation method — Nearest, Linear, or Tricubic — as required.
Resolution… Choose a resolution — Most precise, User defined voxel size, or User defined matrix — as required.
Click the Stitch button.
A new stitched dataset appears in the Data Properties and Settings panel.