This paper presents MGSO (Monocular Gaussian Splatting Optimization), a novel real-time SLAM system that integrates photometric SLAM with efficient 3D Gaussian Splatting for dense 3D reconstruction. The proposed approach leverages photometric SLAM to generate dense structured point clouds that accelerate 3D Gaussian initialization and optimization. By producing more efficient maps with fewer Gaussians while maintaining reconstruction quality, the system achieves an excellent balance between quality, memory efficiency, and speed. Experiments demonstrate that MGSO outperforms state-of-the-art 3DGS-based SLAM systems, making it particularly suitable for real-time dense mapping on resource-limited devices.