Endomaker. An algorithm for fast, accurate, fully automatic extraction of endocasts and their volumes from digital models of the skull

Reproducing cranial endocasts is the main goal for researchers interested in vertebrate brain evolution. Recent developments in 3D digital modelling enabled to overcome the complex and somewhat risky manual production of physical casts. Yet, even the most recent, cutting-edge methods still require somewhat long processing times and usually cannot deal with complex-shaped endocranial cavities. These problems limit both the statistical power and phylogenetic scope of brain shape and size evolutionary investigations. To fix these problems, we developed endomaker, a fully automatic tool for extracting endocasts from digital models of the skull and calculate their volume by using an innovative automatic voxel-based method. We compared endomaker to manual segmentation and other endocast producing software and illustrate how to apply the tool on vertebrate skulls widely differing in size and shape. In this communication, we show the versatility of endomaker by applying it on different shaped, different sized primate skulls including a modern human and a human fossil specimen (Petralona, Homo heidelbergensis). Our results show how the surface of the endocast obtained via endomaker fits the endocranial surface of the skull much more precisely than the one obtained with other software. Endocranial volume estimates under endomaker are as precise as with the use of other sophisticated approaches. Crucially, both endocast production and its volume estimation are indefinitely faster under endomaker. Besides being faster and more precise than competing software, endomaker is versatile enough that it works with disparate file formats such as .stl, .ply, or .obj and on all operating systems. This is favourable since digital models come into different forms and must be translated into a suitable format prior to use with any software other than endomaker. With that, endomaker is the first software ever capable to process single-handedly entire libraries of skull meshes and their endocranial volume estimates in a few minutes, with minimal user preparation.endomaker and example data are fully available in the Arothron R package.