Analysis of CT data using parallel GPU architectures

Computed Tomography (CT) scanners have a wide variety of uses, particularly in medical and industrial imaging. However, the data generated by these devices does not have intrinsic visual information and is often large enough to make it time consuming and tedious for an expert to manually inspect. We present techniques which allow for the fast and intuitive semi-automatic visual analysis of large CT data, comprising hundreds of slices. These techniques make use of the vast capabilities of modern GPUs, allowing for large bodies of data to be processed in real time, while generating images where the information obtained via CT scans is classified and represented through a precise control of color and transparency. We provide detailed descriptions of the underlying algorithms and methods involved in processing the CT information, present our results via images generated by our proprietary rendering frameworks, and explain how these images convey a faster and more intuitive means of inspecting the underlying data.