A 10 um sphere contained over half the cleft size in 74.7 and 66.2 percent of selected connections in two mouse maps

We computed the smallest sphere that contains more than half of each connection's observed cleft size. The sphere center could move freely. We used 1,519 Allen V1 Deep Dive connections and 1,472 MICrONS connections, each with at least three contacts. At radius 10 micrometres, 74.72 and 66.24 percent met this loss threshold. Requiring the center to lie on a contact gave 68.40 and 59.71 percent. We compared the result to contact shuffles and checked the geometry with separate formulas and optimization. The method measures worst-case spatial vulnerability under a linear size-weight model. It extends a prior random spatial failure study on the same panels. One animal per map, selected incomplete arbors, size as a proxy, and no measured failure law prevent a physiological or population claim.

connectomics · Allen Institute · MICrONS · spatial robustness · minimum enclosing ball · cleft size · worst-case loss