In this paper, we propose a novel framework for reconstructing 3D Confocal Laser Scanning Microscopy (CLSM) vasculature volumes from single 2D slices using synthetic and real CLSM data. CLSM is an imaging technique commonly used for in-vivo 3D imaging. However, the utility of 3D CLSM imaging is limited due to long acquisition times and phototoxicity caused by prolonged light exposure. We propose a novel solution to reconstruct a 3D volume from a 2D slice by off-focal plane signal using deep learning (DL). However, the limited availability of real 3D data has to be overcome to train such a 3D reconstruction model. Therefore, we propose a novel framework to effectively address the lack of 3D data for training a robust and accurate 3D reconstruction model by simulating and synthesizing CLSM volumes. Our framework consists of (1) a 3D CLSM volume simulation and synthesizing for training data and (2) a 3D CLSM volume reconstruction part. Experiments show that we can effectively reconstruct 3D CLSM vasculature volumes from single slices, even only using synthetic data. Furthermore, we will also release our CLSM Image of Vascular Organoid for REConstruction (CLSMI-VO-REC) dataset and code at https://github.com/MoriLabNU/3D_CLSM_Volume_Reconstruction.