Java GPU-based multi-histogram volume rendering framework

Wednesday, October 26, 2011 — Poster Session III
10:00 a.m. – Noon	Natcher Conference Center	CIT	IMAG-8

Authors

• R Cheng
• A Bokinsky
• J Senseney
• M McAuliffe

Abstract

This Java graphical processing unit (GPU) based multi-histogram tool extends the utility of one dimensional (1-D) transfer functions for use in a volume rendering framework. The 1-D transfer functions of medical images map optical properties, such as opacity and color, to pixel intensity. This traditional approach is efficient, but it is difficult to perform this operation on overlaid volumes. It is also difficult to identify trends of neighboring pixels or to isolate neighboring pixels of anatomical interest. For example, small changes in the 1-D transfer function can often result in large and unintuitive changes in volume rendering. Multi-histogram is an effective way to extract materials and boundaries for both scalar and multivariate data, also providing the correct overlaying for multiple volumes. Applications for the multi-histogram tool are presented, including integration with a rendering pipeline and the display of composite three dimensional images. Research into integrating 3-D polarized stereo view with multi-histogram is also presented; showing that spatial depth impression can be conveyed and manipulated by the users. The multi-histogram has been developed using the Medical Image Processing, Analysis, and Visualization framework that is developed by the Biomedical Imaging Research Services Section at the National Institutes of Health.

back to top