A comparison of spatial analysis methods for the construction of topographic maps of retinal cell density

Eduardo Garza-Gisholt, Jan M. Hemmi, Nathan S. Hart, Shaun P. Collin

Research output: Contribution to journalArticlepeer-review

41 Citations (Scopus)
102 Downloads (Pure)


Topographic maps that illustrate variations in the density of different neuronal sub-types across the retina are valuable tools for understanding the adaptive significance of retinal specialisations in different species of vertebrates. To date, such maps have been created from raw count data that have been subjected to only limited analysis (linear interpolation) and, in many cases, have been presented as iso-density contour maps with contour lines that have been smoothed 'by eye'. With the use of stereological approach to count neuronal distribution, a more rigorous approach to analysing the count data is warranted and potentially provides a more accurate representation of the neuron distribution pattern. Moreover, a formal spatial analysis of retinal topography permits a more robust comparison of topographic maps within and between species. In this paper, we present a new R-script for analysing the topography of retinal neurons and compare methods of interpolating and smoothing count data for the construction of topographic maps. We compare four methods for spatial analysis of cell count data: Akima interpolation, thin plate spline interpolation, thin plate spline smoothing and Gaussian kernel smoothing. The use of interpolation 'respects' the observed data and simply calculates the intermediate values required to create iso-density contour maps. Interpolation preserves more of the data but, consequently includes outliers, sampling errors and/or other experimental artefacts. In contrast, smoothing the data reduces the 'noise' caused by artefacts and permits a clearer representation of the dominant, 'real' distribution. This is particularly useful where cell density gradients are shallow and small variations in local density may dramatically influence the perceived spatial pattern of neuronal topography. The thin plate spline and the Gaussian kernel methods both produce similar retinal topography maps but the smoothing parameters used may affect the outcome.

Original languageEnglish
Article numbere93485
Pages (from-to)1-15
Number of pages15
JournalPLoS ONE
Issue number4
Publication statusPublished - 18 Apr 2014
Externally publishedYes

Bibliographical note

Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.


Dive into the research topics of 'A comparison of spatial analysis methods for the construction of topographic maps of retinal cell density'. Together they form a unique fingerprint.

Cite this