Volume rendering of 3D medical images generates 2D perspective views by assigning colors and transparency to voxels, allowing users to visualize internal structures through outer tissue layers. Real-time volume rendering requires constant and intensive image transformations, which present significant computational challenges, particularly for image rotations. Although existing algorithms have been optimized, they remain constrained by extensive coordinate relocation and interpolation within the voxel grid, leading to high computational complexity. Meanwhile, the computational burden intensifies rapidly with increasing resolution and data size. To overcome this limitation, we propose a novel 3D Gaussian representation for efficient medical image transformations. This approach models the image as a combination of Gaussian functions, simplifying the voxel grid into a compact representation defined by a limited set of Gaussian parameters. Crucially, geometric transformations of images are reformulated as matrix operations on these Gaussian parameters, significantly reducing computational complexity and enhancing efficiency. Experiments demonstrate that the proposed method achieves exceptional computational efficiency in image transformations while maintaining reconstruction quality, highlighting its potential for medical image volume rendering.