This paper presents a method for accelerating volume rendering and tomographic reconstruction using texture mapping hardware. Volume rendering and reconstruction are based on solving two related integral equations: the volume rendering integral (a generalized Radon transform) and the filtered back projection integral (the inverse Radon transform). Both equations have the same mathematical form and can be approximated using Riemann sums over a series of resampled images. When viewed as a form of texture mapping and frame buffer accumulation, significant hardware-enabled performance acceleration is possible. Volume visualization involves both viewing and constructing volumetric data from projection data. Most volumes used in rendering are derived from sensor data, such as CAT x-ray data. Tomographic reconstruction is the process of converting projection data back into a volume, which can then be visualized using volume rendering techniques. The paper shows that these two operations can be viewed as having the same mathematical and algorithmic form. Traditional volume rendering techniques can be reformulated using hardware texture mapping and summing buffer. Similarly, the Filtered Back Projection CT algorithm can be reformulated using texture mapping and accumulation buffer. The mathematical and algorithmic similarity of these operations, when reformulated in terms of texture mapping and accumulation, is significant. It means that existing high performance computer graphics and imaging computers can be used to both re-derive and reconstruct volumes at rates of 100 to 1000 times faster than CPU-based techniques. The paper discusses the Radon and inverse Radon transforms, which are fundamental to volume rendering and reconstruction. The Radon transform maps the physical object space to its projection space. The inverse Radon transform is used to reconstruct the original object from its projections. These transforms can be represented in discrete form using texture mapping and accumulation. The paper also discusses fan beam reconstruction and cone beam reconstruction, which are generalizations of the Radon transform. These methods are used to reconstruct three-dimensional volumes from projection data. The paper shows that these methods can be implemented using texture mapping hardware, which allows for significant performance improvements. The paper presents a texture map-based reconstruction algorithm that uses texture mapping and frame buffer accumulation to accelerate volume rendering and reconstruction. This algorithm is implemented on existing state-of-the-art graphics systems and is significantly faster than previous algorithms. The cone beam algorithm is a modification of the fan beam algorithm and is used to reconstruct three-dimensional volumes from projection data. The paper also discusses texture mapped volume rendering, which is a hybrid approach combining backwards projecting and forwards projecting techniques. This approach is implemented using texture mapping hardware and is significantly faster than previous algorithms. The paper presents performance results showing that hardware-accelerated algorithms for volume rendering and reconstruction result in significant performance improvements. These results are demonstrated using a RealityEngine system, which is a high-performance graphics system capable of rendering large volumes of data quickly. The paper concludes that the algorithmic unification of volume rendering and reconstruction in the spatial domain means that a single hardware-accelerated solution is possible. Future directions include handling curvThis paper presents a method for accelerating volume rendering and tomographic reconstruction using texture mapping hardware. Volume rendering and reconstruction are based on solving two related integral equations: the volume rendering integral (a generalized Radon transform) and the filtered back projection integral (the inverse Radon transform). Both equations have the same mathematical form and can be approximated using Riemann sums over a series of resampled images. When viewed as a form of texture mapping and frame buffer accumulation, significant hardware-enabled performance acceleration is possible. Volume visualization involves both viewing and constructing volumetric data from projection data. Most volumes used in rendering are derived from sensor data, such as CAT x-ray data. Tomographic reconstruction is the process of converting projection data back into a volume, which can then be visualized using volume rendering techniques. The paper shows that these two operations can be viewed as having the same mathematical and algorithmic form. Traditional volume rendering techniques can be reformulated using hardware texture mapping and summing buffer. Similarly, the Filtered Back Projection CT algorithm can be reformulated using texture mapping and accumulation buffer. The mathematical and algorithmic similarity of these operations, when reformulated in terms of texture mapping and accumulation, is significant. It means that existing high performance computer graphics and imaging computers can be used to both re-derive and reconstruct volumes at rates of 100 to 1000 times faster than CPU-based techniques. The paper discusses the Radon and inverse Radon transforms, which are fundamental to volume rendering and reconstruction. The Radon transform maps the physical object space to its projection space. The inverse Radon transform is used to reconstruct the original object from its projections. These transforms can be represented in discrete form using texture mapping and accumulation. The paper also discusses fan beam reconstruction and cone beam reconstruction, which are generalizations of the Radon transform. These methods are used to reconstruct three-dimensional volumes from projection data. The paper shows that these methods can be implemented using texture mapping hardware, which allows for significant performance improvements. The paper presents a texture map-based reconstruction algorithm that uses texture mapping and frame buffer accumulation to accelerate volume rendering and reconstruction. This algorithm is implemented on existing state-of-the-art graphics systems and is significantly faster than previous algorithms. The cone beam algorithm is a modification of the fan beam algorithm and is used to reconstruct three-dimensional volumes from projection data. The paper also discusses texture mapped volume rendering, which is a hybrid approach combining backwards projecting and forwards projecting techniques. This approach is implemented using texture mapping hardware and is significantly faster than previous algorithms. The paper presents performance results showing that hardware-accelerated algorithms for volume rendering and reconstruction result in significant performance improvements. These results are demonstrated using a RealityEngine system, which is a high-performance graphics system capable of rendering large volumes of data quickly. The paper concludes that the algorithmic unification of volume rendering and reconstruction in the spatial domain means that a single hardware-accelerated solution is possible. Future directions include handling curv