The Phenotypic PointCloud Analysis is an open source python workflow that allows for the three dimensional registration, statistical analysis, and visualization of scalar data associated with points generated from tetrahedral meshes. This workflow utilizes the coherent point drift algorithm or Myronenko and Song, 2010 (10.1109/tpami.2010.46) to perform a rigid, affine, and deformable registration of a mean shape to individual points clouds. Thereafter the scalar information associated with the individual point clouds is statistically analyzed following the statistical parametric methods described by Worsley et al. 1996 (10.1002/(sici)1097-01934:1<58::aid-hbm4>3.0.co;2-o) and Friston et al. 1994 (10.1002/hbm.460020402). The statistical results may then be visualized in 3D on a tetrahedral mesh of the mean morphology in various freely available visualization platforms (e.g. paraview).
Publications
2020
DeMars, L. J., Stephens, N. B., Saers, J. P., Gordon, A., Stock, J. T., & Ryan, T. M. Using point clouds to investigate the relationship between trabecular bone phenotype and behavior: An example utilizing the human calcaneus. American Journal of Human Biology, e23468.
RDN Segmentation
Currently in development, the RDN Segmentation project is a web browser based application that uses a regularized deep-learning network model trained on images segmented by a set of experts to automate the segmentation of images into 3 classes (air, non-bone, and bone). In addition to segmentation this app also provides researchers with a no-code way to train their own custom models, batch segment CT images in various formats, and produce high quality meshes for morphological analysis.
Publications
2020
Stephens, N. B., Yazdani, A., Cherukuri, V., DeMars, L. J., Monga, V., & Ryan, T. M. Machine Learning in Anthropology: A Regularized Deep Network for Osteological Micro-CT Image Segmentation. American Journal of Physical Anthropology, 171, 274-274.