Overview: main steps for the creation of 3D models

BioVis3D creates 3D models of objects that you identify working on a sequence. A sequence is a document that represents a stack of aligned slices together with a set of objects delimited within those slices. The creation of a 3D model involves a few steps which are summarized here:

  1. Gathering your images.

  2. Creating a BioVis3D sequence.

  3. Optionally, building a mosaic of every slice.

  4. Aligning the slices of the sequence.

  5. Identifying your objects of interest.

  6. Creating and visualizing 3D models.

Gathering your images

The main input to BioVis3D is a set of digital images of serial sections or slices. In general each slice is comprised of one single image, however, when the acquisition device does not allow to capture the full image of a section in one shot, take several overlapped pictures of the same slice and use BioVis3D to construct a mosaic. 

Tip: Make sure mosaic images significantly overlap to simplify mosaic construction.

It is important to be able to compute the pixel size of every image and the distance between slices, as this information will determine the aspect ratio of 3D models and will make it possible to take precise measures of the objects. In some cases the pixel size of the images can be obtained directly from the acquisition devices, in other cases, it may be necessary to take a picture of a sample of known size and use BioVis3D to compute the pixel size from said picture.  

Tip: If possible, before sectioning include at least two straight traces perpendicular to the sections, along the material of study. This makes it easy to detect points in correspondence between different slices, and simplifies the slice algnment process.

In order to incorporate your images to BioVis3D easily, it is convenient to store image files appropriately. Two cases are distinguished:  

When each slice is comprised of one single image: Store image files in a single folder named in such a way that their alphabetical order coincides with their physical order in the stack of slices. For example name the images with their position number, zero padded to a fix number of digits like 009.jpg, 010.jpg, 011.jpg. Do not use variable length names like 9.jpg, 10.jpg, 11.jpg as in this case 10.jpg comes before 9.jpg in alphabetical order.

Figure 1: Organization of image files of a sequence with one image per slice.

When some slices are comprised of several images: Create a separate folder for the images of each slice and name the folders in such a way that their alphabetical order coincides with their physical order in the stack of slices. For example name the folders with their position number, zero padded to a fix number of digits like 009, 010, 011. Do not use variable length names like 9, 10, 11 as in this case 10 comes before 9 in alphabetical order.  

Figure 2: Organization of image files of a sequence with multiple images per slice.  

Creating a BioVis3D sequence

Once the images have been collected, create a sequence in BioVis3D following the steps of the Sequence Creation Wizard. The wizard will request a name for the sequence and a folder where to create it, the location of your images, the distance between slices, and the pixel size of images. Recall the distance between slices and the pixel size of your images determine the aspect ratio of the 3D model and allows for taking measures, thus, make sure you can get this information.

See Create a Sequence for details.

Optionally, building a mosaic of every slice

When the acquisition device does not allow to capture the full image of a slice in one shot, several overlapped pictures of the same slice can be taken. BioVis3D provides the tools necessary to translate and rotate each picture that comprises a slice to build a single coherent image of the section. 

See Mosaic tool.

Aligning the slices of the sequence

Once a sequence has been created and, possibly, a mosaic of every slice has been built, align the sections along a transversal axis to construct a coherent stack in three dimensions. In Figure 3, the slice on top needs to be rotated and the one at the bottom needs to be translated with respect to the second slice.

Figure 3: Registration process.

In BioVis3D the alignment process is performed in an adjacent couple basis, i.e., the second slice is aligned with respect to the first one, the third one with respect to the second one, and so forth. The process of aligning two adjacent slices is called registration.

See Alignment of slices.

Identifying your objects of interest

In order to create 3D models, identify the objects of interest. Each sequence may have several objects which are organized in categories. In Figure 4, there are two categories: Brain Category and Neurons. The former consists of one single object, Brain, and the latter consists of two objects: Neuron1 and Neuron2. Although categories are not really necessary for using BioVis3D, they provide a valuable tool for organizing sequences with a large number of objects.  

Figure 4: Identification of objects.

BioVis3D provides three types of modeling features, each one designed for a specific structure: 3D surfaces modeling, spheres modeling and paths modeling. A sequence can contain objects built using one type of modeling, or several modeling features combined. Figure 5 shows the three types of modeling features.

Figure 5: Modeling features: 3D surfaces, spheres and paths.

Modeling with 3D surfaces: Objects span one or several slices, the contours that delimit an object in every slice that intersects it, must be specified. In Figure 4, the object Brain has three different components which are bounded by contours, Contour1, Contour2, and Contour3 in Slice 1. In the following slice, shown in Figure 6, the three components were joined in a single connected component that is now bounded by two contours: Inner and Outer. As Countour1, Contour2, and Contour3 in Slice 1 and Inner and Outer in Slice 2 belong to the same object Brain, BioVis3D will correctly build a 3D model where the three components of Slice 1 are joined with the single component of Slice 2.

 

Figure 6: Identification of objects with inner and outer contours.

Notice that in Figure 6, the object Brain has "hole" area bounded by a contour we have conveniently named Inner. Contour names are not relevant for BioVis3D, since it analyzes the inclusion of contours belonging to the same object to automatically detect holes, isles, etc. It is important, however, to assign contours to objects correctly along all slices as this is the way BioVis3D knows which contours must be joined together when creating a 3D model of an object.

Modeling with spheres: In addition to contours, dot sets, which are collections of spheres of the same size and color, can also be defined. Dot sets are especially useful when the user is focusing in the position and size of certain elements of an object, and is not in their precise shape. In Figure 7 the object Neurons_1 has two dot sets named RedDots and BlueDots with different diameters. In a 3D view of an object, its dot sets are visualized as clouds of spheres of the selected diameter and color. In the example of Neurons_1, different kinds of neurons can be easily distinguished by their color in a 3D scene.

Figure 7: Dot sets.  

Modeling with paths: In addition to contours and dot sets,paths can also be defined, which are sets of nodes that can be connected with each other or not. Once rendered in 3D, the path will appear as a tubular structure resembling a tree that traverses all the connected nodes. These nodes can be distributed within the same slice or in different slices, thus allowing the path to span multiple slices along the Z axis. Each node defines its diameter which in turn defines the section of the path at that given point. The color of the path is determined by the color of the object it belongs to.

See Identification of objects.  

Creating and visualizing 3D models

The 3D model creation and visualization process is straight forward after all objects are delimited with contours,  drawn with dot sets and paths in every slice. Figure 8 shows a complete 3D model of a fish brain made semi-transparent with several inner objects in different colors.

 

Figure 8: A 3D model made semi-transparent with inner objects.

See 3D View.