Enzymind Labs

Human brain surface model comparisons

    In the process of making a reasonably accurate bronze cast of a human brain, I came up against the limitations of polygonal surface models. In particular it is difficult to generate a highly detailed and topologically correct polygonal mesh from high resolution (~1mm)  MRI volume data.

    This is just a visual comparison between renderings of brains, using both volume rendering and polygonal mesh surface rendering techniques.

Rough unshaded polygonal mesh rendering, derived from MRI data. The surface extraction was designed to "reach down" into sulcii (dark fissures), accentuating the gyrii ridges but not representing the cortical surfaces accurately. Topologically validated vertex mesh: spherical topology, no self-intersection, no intersection with the other surfaces. Derived from a T1-weighted anatomical magnetic resonance image of a healthy subject. Segmentation with a topology-preserving level set evolution followed by a topology-consistent marching cubes algorithm. 653852 facets File, meta, and mesh data

Shaded and artificially colored volume rendering of MRI data of a different brain, from a different view angle. The cerrebellum (lower right) is included. Not using surface data: rendered based on line-of-site image gradient inflection points; Derived from a an average of 27 T1-weighted anatomical magnetic resonance image of a healthy subject.