Several methods have been proposed for finding the fiber orientation within the brain, using dMRI data. However, few studies have been done on finding the fiber orientation within voxels in a way that encourages and subsequently reveals the neural connectome. In this study, we explore the MCMC based approach to estimate the fiber orientation within a set of voxels using the ball and stick model as the likelihood and the Watson distribution as the directional prior distribution. The Bayesian hyperparameter is being calculated from the surrounding voxel’s possible fiber orientations. The choice of suitable hyperparameter is determined by the most probable orientation of fibers in the surrounding voxels such that it facilitates a plausible fiber network. We evaluated the accuracy of our technique for different simulated nerve networks.