The field of optical microscopy has grown rapidly during the last several years, reflecting the development of genetic methods for expressing fluorescent proteins and the increasing availability of confocal and multiphoton microscopes capable of collecting high-resolution three-dimensional (3D) images. We previously developed and distributed volume rendering software for 3D microscopy (Voxx), which used low-cost high-performance graphics processing units (GPUs) to perform real-time volume rendering on personal computers. However, the low-precision pixel arithmetic used in early versions of the GPUs, combined with limited programmability, made it difficult or impossible to perform many 3D image processing operations during real-time volume rendering. The latest generation of GPUs features improved programmability and higher-precision pixel arithmetic.

Here we describe a new version of this software that uses these recent advances in GPUs to provide significant new real-time capabilities for the volumetric visualization of microscopy data, including multi-channel crosstalk correction, depth-based intensity correction, and orientation-based lighting effects. Voxx can also render volume data collected in a time series, which permits one to visualize 3D motion and changes in shape. We use Voxx and a GeForce 6800 GPU to demonstrate how real-time image processing during volume rendering can be employed in 3D and 4D studies performed on confocal and two-photon microscopes.

J.L. Clendenon, J.M. Byars, K.W. Dunn, "Real-time 3D Image Processing during Volume Rendering", Molecular Biology of the Cell, Vol. 15 supplement (November, 2004), pp. 102a. poster at the 44th Annual Meeting of the American Society for Cell Biology