The human brainstem is an anatomically complex and compact structure, and many neurologic diseases are frequently associated with brainstem dysfunction. Despite its importance in brain functioning and neurodegenerative processes, the brainstem and its functional sub-structures are relatively unexplored in medical image analysis. Here we present a data-driven framework to extract functional sub-regions from the brainstem. We first apply a novel motion correction scheme to the brainstem. A simple network is then derived by examining the correlation of BOLD signals between brainstem voxels, and a network community quality function is optimized to extract the sub-networks within the brainstem. We applied this technique to fMRI data from fifteen healthy participants and found 84 group-level, spatially contiguous sub-regions within the brainstem. Association of these regions with other cortical and subcortical brain regions were investigated to assist in interpreting what underlying anatomical structures were associated with the subregions. Although the proposed method was originally developed for the brainstem, the proposed framework has the potential to be integrated into studies investigating functional sub-regions from other cortical or subcortical brain regions.