We explore heterogeneity in osteosarcoma using imaging “habitats”, which identify different physiological subregions by MRI. We propose a method to establish the relationship between the microenvironmental of a habitat by relating histology to MRI. Computational image analysis was used to cluster tumor habitats with a 3D printing approach to co-register MR images with histology and immunohistochemistry. Compared with H&E and the CA-9, we found that cellular morphology and density were in concordance with the clustered habitats, although there are subtle differences between histology and MRI slices. Thus, identifying tumor habitats in osteosarcoma using multiparametric MRI is feasible and promising.