Week #18

Advanced, Experimental VFX Animation and Techniques
Blog WK #18 (June 1), Dendritic Spines.

The Andrew Tran Tutorial, Part 2

As per part one of the Tran 3-part tutorial (2014), the general skeleton for the neuron has been completed. Specifically, two neurons have been completed (plz review Blog wk#17). The first was processed entirely by means of polygon modelling, and the second aiming at NURBS modelling, was initially successful with the NURBS effort, and then finished by polygon modelling. Beyond however, "merely" the backbone or skeleton of the neuron, the surface of the neuron requires several dendritic spines, as demonstrated in Tran's 2nd tutorial. His approach (as visually summarized in the GIF on the left), involves use of mushroom images from the Maya Content Browser. Because Tran had no intent to show closeup shots of the dendritic spines, this approach is justified. In other words, it is perhaps further notable that while the shape of a dendritic spine is approximated by that of a mushroom, upon closer inspection (reaching a resolution beyond what may be necessary), a dendritic spine does not truly have the dome-like head typical of a mushroom.

Dendritic Spine Anatomy: Potential Need for Closeup

While Tran's tutorial was very effective for his endpoint, the obvious downside of strictly adhering to it is the issue of not having the option to zoom in for any sort of closeup that might highlight a dendritic spine. Thus, to allow myself some freedom to do so, I opted to have a subset of the dendrites show more realistic spines. As below, papers by Benavides-Piccione and coworkers (2013) and Luengo-Sanchez and colleagues (2018) were used as a reference for more realistic closeup examples of pyramidal cell dendritic spines.

Modelling of a Single Dendritic Spine

A Dendritic Spine "Forest"

On a pyramidal cell, the dendritic spine density is quite high (Benavides-Piccione et al, 2013). To model each spine and cover all the dendrites of a full neuron with unique spines would probably consume a month of time. Thus, for the sake of developing a showreel this term, many of the spines will merely be duplicates, but with enough variability across the spines for the dendrite to still appear natural. For the sake of simply this term (i.e. this effort will be revisited over our extended summer break), I have modelled 10 unique dendrites, and duplicated patches of them across one component of a single dendrite. The clip for this term's showreel will plan for a strategic camera movement that avoids highlighting redundancy of the spines and further avoids showing dendrites which do not even include spines.

The following GIF shows a closeup of 2 unique spines and a subsequent slide portraying 6 and then 10 variants. For the fully arborized dendrite, this set of 10 spines was duplicated, but modified to portray different sizes, angles and cluster patterns so as to appear natural. At this point, the planned camera movement for the final clip will start w passing from a closeup of part of one of the dendrites, but with full neurons in the background, and a complete hippocampal model in the distant background. The didactic nature of the scene will become apparent as the scene progresses from the dendritic closeup to the full hippocampus in the background. The following GIF shows the initial construction of the dendrite to be used in closeup.

Bibliography

Benavides-Piccione R, Isabel Fernaud-Espinosa, Robles V, Yuste R, DeFelipe J. (2013). Age-Based Comparison of Human Dendritic Spine Structure Using Complete Three-Dimensional Reconstructions, in Cerebral Cortex. 23(8): 1798–1810. Available at doi: 10.1093/cercor/bhs154

Luengo-Sanchez S, Fernaud-Espinosa I, Bielza C, Benavides-Piccione R, Larranaga P, Deflipe J. (2018). 3D morphology based clustering and simulation of human pyramidal cell densritic spines. PLOS Computational Biology.
https://doi.org/10.1371/journal.pcbi.1006221

Tran A (2014). Modeling a neuron in Maya 2014 https://www.youtube.com/watch?v=vhHlelGB1qE.