Odor Processing with an experimental model of Olfactory epithelium and bulb

• Artificial olfaction  was introduced  as a model tool  to investigateolfaction  properties  [1]. Nonetheless,   the  only  analogy  between the natural and the artificial system lies just in the selectivity proper- ties of the receptors. The implementation of more sophisticated fea- tures such as the large number of receptors and the glomerular layer have been hampered  by technical difficulties related to the manage- ment of large numbers  of simultaneous  signals.As demonstrated in the past, optical imaging is a read-out  tech- nique for sensors development that can provide large sensor arrays [2]. On that basis, we recently introduced  an artificial olfaction sys- tem based on the imaging of a continuous layer of chemical indi- cators [3]. In this situation an image sensor provides a segmentation of the whole sensing layer in a number  of elementary  units corre- sponding to the pixels of the image. Eventually, since it is possible to evaluate the optical properties of every single pixel, each pixel of the image may correspond to an individual sensor. In this regard, even low-resolution  images may easily result in thousands of independ- ent sensing units.In our system a collection of arbitrarily shaped regions of color indicators  is illuminated  by a controlled  light source;  the optical characteristics  of each pixel of the image are measured by a camera yielding the light intensities in the three channels  red, green, and blue.   The  combination  of  illumination   sequence  and   cameraread-out  results  in  a  fingerprint  encoding  the  optical  properties of the sensing layer portioned in image pixels. Even a simple clas- sification of these fingerprints assigns each pixel to a class, and each class contains pixels carrying the same color indicator.  This behav- ior resembles the association between ORNs carrying the same chemical receptors into the same glomerulus [4]. On the basis of this analogy it is straightforward to describe the layer of indicators as an artificial epithelium, pixels of the image as artificial olfactory  neu- rons, and the classes provided  by the classifier as an abstract  rep- resentation of artificial glomeruli.This system thus allows the generation of a complex model of olfaction,  including  glomerular  compartmentalization [5], which is then applied to data generated by the exposure to pure and mixed gases. Results show that such a model enhances the discrimination of pure and mixed odors. Eventually,  such a platform,  apart  from evidencing the similarities between natural and artificial olfactory systems, is also proposed as a practical tool to test olfactory models.1. K. Persaud  and G. Dodds,  Nature  299 (1982) 3522. Dickinson  et al., Nature  382 (1996) 6973. C. Di Natale  et al., PLoS  ONE 3 (2008) 31394. P. Mombaerts, Annu Rev Neurosci 22 (1999) 4875. D. Schild and H. Riedel, Biophysical Journal,  61 (1992) 704

Nyckelord

TECHNOLOGY

TEKNIKVETENSKAP

Publikations- och innehållstyp

vet (ämneskategori)

art (ämneskategori)