<conference poster>
Statistical analysis for clustering of areas on the olfactory bulb and estimation of the physico-chemical properties detected by glomeruli in each area

Creator
Language
Publisher
Date
Source Title
Vol
First Page
Publication Type
Access Rights
Related DOI
Related URI
Relation
Abstract The odorants are received by the olfactory receptors on the olfactory receptor cells and activate their cells. The activations of the cells are conveyed to glomeruli in the olfactory bulb, and activat...e neurons in the glomeruli. An olfactory receptor receives a partial structure of an odorants, and all olfactory receptors connected to a glomerulus receive the same partial structure of an odorants. Furthermore, the location of every glomerulus in the olfactory bulb does not have individual difference. Therefore, a glomerulus in a particular location of the glomerular layer in the olfactory bulb detects particular physico-chemical properties. Some researchers have clarified this correspondence between physico-chemical properties and locations in the glomerular layer. However, properties or areas in the glomerular layer they clarified were coarse.
We clarified this correspondence by statistically analyzing the activation pattern images of the glomerular layer in the rat olfactory bulb for various odorants, which are put on the site http:
gara.bio.uci.edu/index.jsp. First, we constructed the statistical model to generate these images. This model has parameters that express the probabilities of generating each type from each pixel or the probabilities of generating each brilliance from each pair of image and type. The type of the pixel will correspond to the physico-chemical properties that the glomerulus in the pixel detect. Next, we estimated the parameters using the activation pattern images by Gibbs sampling. We found clusters of pixels (that is to say, the type for each pixel) and found probability that the pixel with the type t is brilliant for each image and each type t. Finally, we estimated the physico-chemical properties corresponding to each type based on the correlation between the probability that the pixels with type t is brilliant and whether the odorant has property p for each p and each t.
show more

Hide fulltext details.

pdf ISOT2016-poster pdf 900 KB 64 ISOT 2016 poster

Details

Record ID
Peer-Reviewed
Notes
Created Date 2017.09.14
Modified Date 2021.12.13

People who viewed this item also viewed