In many areas of biology there is an abundance of noisy and imprecise data, but often a lack of high precision quantitative data. It is therefore important to design, and use, data analysis methods that are insensitive to small errors in measurements. We present such a robust data analysis as we investigate the three dimensional spatial structure of the locus of afferent neuron terminals in crickets Acheta domesticus. Each afferent neuron innervates a filiform hair positioned on a cercus: a protruding appendage at the rear of the animal. The hairs transduce air motion to the neuron signal that is used by a cricket to respond to the environment. We stratify the hairs (and the corresponding afferent terminals) into four classes, which depend on hair length as well as the position of each hair. Our analysis uncovers significant structure in the relative position of these terminal classes and suggests the functional relevance of this structure. Our method is very robust to the presence of significant experimental and developmental noise. It uses ideas from persistent homology [1-3]. This method is applicable to any point cloud data set that exhibits an underlying topological structure corrupted by noise.
We show that each individual terminal class has similarly positioned significant tunnels and/or voids. However, while each class' point cloud is of similar shape and structure, the space that each class occupies is slightly offset, resulting in no significant tunnels and/or voids in the set of terminals of all hairs. This result is not surprising from an efficiency standpoint: the space in the terminal ganglion is limited and the terminals of afferent neurons are filling the entire available space. While, we emphasize that there are still visible tunnels and/or voids in the entire point cloud, our analysis provides us with the conclusion that these remaining tunnels are not significant, and very likely contain the dendrites of the interneurons in the terminal ganglion. These are downstream from the afferents, and make synaptic contact with the afferent terminal cloud.
[1]
Arne Storjohann,et al.
Near optimal algorithms for computing Smith normal forms of integer matrices
,
1996,
ISSAC '96.
[2]
Graham I Cummins,et al.
The cricket cercal system implements delay-line processing.
,
2010,
Journal of neurophysiology.
[3]
R. Pollack,et al.
Surveys on discrete and computational geometry : twenty years later : AMS-IMS-SIAM Joint Summer Research Conference, June 18-22, 2006, Snowbird, Utah
,
2008
.
[4]
B LeeAnn,et al.
The Nonlinear Statistics of High-Contrast Patches in Natural Images
,
2003
.
[5]
Gunnar E. Carlsson,et al.
Topology and data
,
2009
.
[6]
Bernard W. Silverman,et al.
Density Estimation for Statistics and Data Analysis
,
1987
.
[7]
F. Theunissen,et al.
Functional organization of a neural map in the cricket cercal sensory system
,
1996,
The Journal of neuroscience : the official journal of the Society for Neuroscience.
[8]
坂上 貴之.
書評 Computational Homology
,
2005
.
[9]
Kim Steenstrup Pedersen,et al.
The Nonlinear Statistics of High-Contrast Patches in Natural Images
,
2003,
International Journal of Computer Vision.
[10]
Herbert Edelsbrunner,et al.
Topological Persistence and Simplification
,
2000,
Proceedings 41st Annual Symposium on Foundations of Computer Science.
[11]
Marian Mrozek,et al.
Coreduction homology algorithm for inclusions and persistent homology
,
2010,
Comput. Math. Appl..
[12]
Marian Mrozek,et al.
Coreduction Homology Algorithm
,
2009,
Discret. Comput. Geom..
[13]
James R. Munkres,et al.
Elements of algebraic topology
,
1984
.
[14]
G. Jacobs,et al.
Neural Mapping of Direction and Frequency in the Cricket Cercal Sensory System
,
1999,
The Journal of Neuroscience.
[15]
F. Theunissen,et al.
Extraction of Sensory Parameters from a Neural Map by Primary Sensory Interneurons
,
2000,
The Journal of Neuroscience.