Functional magnetic resonance imaging (fMRI) is a critical technique for studying brain activity, though its conventional three-dimensional visualization methods often lose spatial depth information by projecting activations onto superficial brain models. This work proposes and explores two alternative methodologies for the 3D representation of fMRI data: the first one based on point clouds and the second one on voxel-based volumetric representation with adaptive transparency. The main objective is to develop and evaluate 3D visualization techniques that preserve and clearly represent the full spatial extent, including depth of fMRI activations, thereby facilitating a more accurate and complete interpretation of the data. The fundamentals of each method are described, a preliminary implementation for point cloud representation is presented, and the potential of both approaches for neuroscientific research and clinical practice is discussed, along with a framework for an interactive visualization tool.