Three-dimensional Reconstruction of the Drosophila Larval and Adult Brain

Three-dimensional Reconstruction of the Drosophila Larval and Adult Brain
- 3D Reconstruction

Help Index Where Am I? Search Page Status

A 3D visualisation of a series of optical sections (cf. Fig. 1) can be based either on the complete volume itself or on polygonal surfaces of the structures of interest. The general techniques are called volume-rendering and surface-rendering, respectively. Volume-rendering utilizes a variety of algorithms to generate three-dimensional views from the optical sections obtained with the confocal microscope. Ray-tracing algorithms for example simulate the emission, absorption and reflectance of light along a large number of light rays. Such algorithms require the complete three-dimensional data set and high computation times per frame. More recent approaches use so called texture graphics hardware to accelerate volume rendering (cf. Fig. 2). Nevertheless to generate views of the data in real time still exceeds the capacity of most normal desktop computer. Hence volume rendered visualisations are mainly distributed as images showing single views or video sequences.
In contrast, surface rendering is based on polygonal models (cf. Fig. 3), which are typically obtained from labeled and therefore drastically reduced datasets. Generating views of three-dimensional surfaces can be performed on a normal desktop computer in real-time, i.e such models can be handled and manipulated interactively with a 3D viewer. The Virtual Reality Modeling Language 2 (VRML2) has been established as an ISO standard for the transfer of such surface models via the Internet. VRML2 is programmable and allows one not only to turn and move 3D objects, but also to remove parts of structures and include the original image data on which the reconstruction is based.

Figure 1

Figure 2

Figure 3

Larval Brain Model

Adult Brain Model

Main | The Flybrain | Materials and Methods | Handling of 3D models | References

Return to Flybrain 3D Models

Flybrain Atlas Dissectable Brain Background

AB00121

A 3D visualisation of a series of optical sections (cf. Fig. 1) can be based either on the complete volume itself or on polygonal surfaces of the structures of interest. The general techniques are called volume-rendering and surface-rendering, respectively. Volume-rendering utilizes a variety of algorithms to generate three-dimensional views from the optical sections obtained with the confocal microscope. Ray-tracing algorithms for example simulate the emission, absorption and reflectance of light along a large number of light rays. Such algorithms require the complete three-dimensional data set and high computation times per frame. More recent approaches use so called texture graphics hardware to accelerate volume rendering (cf. Fig. 2). Nevertheless to generate views of the data in real time still exceeds the capacity of most normal desktop computer. Hence volume rendered visualisations are mainly distributed as images showing single views or video sequences. In contrast, surface rendering is based on polygonal models (cf. Fig. 3), which are typically obtained from labeled and therefore drastically reduced datasets. Generating views of three-dimensional surfaces can be performed on a normal desktop computer in real-time, i.e such models can be handled and manipulated interactively with a 3D viewer. The Virtual Reality Modeling Language 2 (VRML2) has been established as an ISO standard for the transfer of such surface models via the Internet. VRML2 is programmable and allows one not only to turn and move 3D objects, but also to remove parts of structures and include the original image data on which the reconstruction is based.
	*Figure 1*

	*Figure 2*

	*Figure 3*