Arborization pattern was studied in pial vascular networks by treating them
as fractals. Rather than applying elaborate taxonomy assembled from measur
es from individual vessel segments and bifurcations arranged in their branc
hing order, the authors' approach captured the structural details at once i
n high-resolution digital images processed for the skeleton of the networks
. The pial networks appear random and at the same time having structural el
ements similar to each other when viewed at different scales-a property kno
wn as selfsimilarity revealed by the geometry of fractals. Fractal (capacit
y) dimension, D-cap, was calculated to evaluate the networks spatial comple
xity by the box counting method (BCM) and its variant. the extended countin
g method (XCM). Box counting method and XCM were subject to numerical testi
ng on ideal fractals of known D. The authors found that precision of these
Fractal methods depends on the fractal character (branching, nonbranching)
of the structure they evaluate. D(cap)s (group mean +/- SD) for the arteria
l and venous pial networks in the cat (n = 6) are 1.37 +/- 0.04. 1.37 +/- 0
.02 by XCM, and 1.30 +/- 0.04, 1.31 +/- 0.03 by BCM. respectively. The arte
rial and venous systems thus appear to be developed according to the same f
ractal generation rule in the cat.