Retina

Some particular zones of the retina

the Fovéa
the Macula
the optical Papille
the Point plugs

Physiology of the retina

Morphology of the retina: light with the impulse

As the remainder of SNC, the retina consists of Neuron S accompanied by cells by support and is largely vascularized by blood-vessels. According to the observations of Santiago Ramón Cajal, whose cuts were often prepared vertically on the surface, one knows there that the retina consists of 6 successive layers of differentiated neurons, since the layer of the Neuron S sensitive to the light, the photoreceiver S (PHR) to its former layer, that of the ganglionic cells (CG), whose gathering of the Axone S form the optical nerve.

Advance of the visual information in the retina

axial Organization simplified of the retina (modified since a drawing of Santiago Ramón there Cajal).

The Lumière focused by the eye crosses (here from left to right) the retinal layers to activate the photoreceiver S (which are presented in the form of cones or of Bâtonnet S, layer at the right end). Those activate ahead (axially, here towards the left) the bipolar cells and laterally the horizontal cells (in the layer represented above out of orange), so that the signal is sensitive to space contrasts. The signal is propagated ahead towards the ganglionic cells (whose cellular bodies form the first layer, here on the left), which one sees appearing two sizes corresponding to the ways Magno and Parvo. This signal is filtered laterally by the cells amacrines (they touch the ganglionic cells and are in the direction opposed on the drawing), which they are mainly sensitive to temporal contrasts.

Nature of the retinal signals

The luminous image which arrives at the level of the retina is reversed compared to the external image because the Cristallin of the eye is a biconvex lens. This image, reversed and distorted because of rotundity of the retina, is transformed analogically in the retina into an image complexes corresponding to the activity of the neurons of the various layers. Indeed, since the synaptic signal emitted by the photoreceivers (PhRs), information is coded by an analogical signal being propagated by synaptic contacts and chemical diffusion. The bipolar cells are sensitive to space contrasts whereas the cells amacrines are more particularly sensitive to temporal variations. They thus will temporally transform the image of luminous contrasts which will be propagated ahead towards the layer of the ganglionic cells (CGs). However, these behaviors more complex and are varied and correspond to a whole family of space-time transforms which are not all yet today known (for a review, one will be able to consult Meister 99). Finally, it is only with the layer of the ganglionic cells (thus most external) that luminous contrasts can produce neuronal impulses or potential of actions (Pa) which will then constitute the signal borrowed by the optical nerve, completely impulse signal. The retina completes on this level the transformation of luminous information in the train of impulses.

Space organization of the neurons of the retina: luminous image with the retinal image

This radial approach of the retina was refined by techniques instrumental modern which showed the importance of the space organization on the surface of the retina. As Cajal had already observed, concentration and the nature of the various neurons vary according to the position on the retina. Indeed, it is observed first of all that the concentration in cones increases strongly when one approaches the axis of vision of the eye, the fovea for almost cancelling itself apart from mackled which corresponds to a disc of a degree since the fovea). In complement, the concentration in sticks is almost null in mackled but reached a maximum towards 3°. This observation explains why we do not have sensitivity to the color out of the fovea, and a symmetrical way that to see a weak star, it is advised to fix its glance slightly at side for that the light of star strikes the sticks rather than the cones.

One then defines for the ganglionic cells their receiving field (Imbert, 1983) like the whole of the PHR which take part in its excitation. One observes whereas those have appreciably circular contours of which the ray grows on average proportionally with their eccentricity. Since the fovea - where it is equivalent to a PHR (cone) by CG - the receiving fields can to reach 10deg field of vision with the periphery of the retina. One thus can to define a retinotopic chart which will correspond to the space transformation image by this space arrangement of the grid of the neurons. It is often approached by a log-polar transform of the centered spherical image on the eye. This chart is incomplete due to the covering of areas by axons of CGs (the scotome) and in particular with the convergence of the axons towards optical nerve, which constitutes an insensitive zone, the blind point.

Finally, CGs are sensitive to contrasts of lights to several sizes, and qualitatively certain answering CG maximalement when it signal corresponds to a spot of light surrounded by a circumference of the size of sound receiving field (cell centers it) or its reverse (dark center on bottom clearly, the cells center-OFf). Also, even if the answers of the cells are very varied and in spite of the complexity of the retinal network, Rodieck (1965) showed that the answer until summoned of CGs could be modelled in a linear way compared to the answers of the photoreceivers. This simplification can theoretically to allow to determine in an exact way the transform of a CG (which is then a linear space-time filter) thanks to its answer impulse with a spot of light. However, of many phenomena non-linear, like perception color or the adaptation to the total intensity, is introduced into the retinal answers.

Many scientists of the field of the computational Neurosciences tried to model the retina in order to create prostheses but also for better including/understanding its operation. Among them, one can quote work of David Marr on the perception of brightness in the retina of the primates (Marr, 1974).

Transformation multichannel: the impulse image

It was seen that they are only CGs which emits the Steps which will be transmitted with the remainder of the SNC, which shows that since the 10⁸ photoreceivers (PhR) via approximately 10⁹ intermediate cells and to the optical nerve consisted axons of the million ganglionic cells (that is to say a compression of about 100 of many cells), the retinal transform is one transformation of an luminous intensity varying in time with a signal spatiotemporel impulse of Pa. It is also noticed that the number relatively not very high fibers at exit show that the size of the signal must be compressed so that it is transmitted effectively to the remainder of the SNC. A method used by the retina is then to transform the visual information into signal multichannel which tends to separate the sources which produced the feeling luminous, thus reducing the dimension of the signal to be transmitted.

In particular, Atick showed that the answer of the cells ganglionic at various space frequencies coincided with a reduction space correlations between close hirings, showing thus that ecological principles can guide the comprehension of the retinal functions. An aspect of retinal coding is thus to underline nonredundant parts and which is thus relatively projecting. Within our framework, this sensitivity goes to allow to propagate more quickly the parts of the image more projecting, leading moreover to one temporal transformation of information space. In a similar way, it is observed that the ganglionic cells transform luminous information into relatively independent signals. Thus the information of color, the chrominance is separated from information of luminous intensity, leading to a multiplexing of information luminous. One observes thus that of CGs morphologiquement and functionally different (cells has, B and G) will carry different channels. It decoupling will be also temporal since the information of luminous intensity is more quickly activated than the color, thus creating ways with several latencies for retinal information.

Finally, if one rather quickly presents an image on a subject to avoid any ocular jerk, this one will be projected at the retina in an image distorted and reversed, to activate the photoreceivers then all the retinal network for finally being transformed into multiple channels by the cells ganglionic. In a synthetic way, each one of these cells can then be characterized by a maximum sensitivity to a particular channel and by one temporal answer, but the sensitivities can overlap with those other CGs and are interdependent (Salinas01). The image which we perceive then is entirely coded in the train of impulses in approximately 20-40ms. Then that the wave of activity joined now the optical nerve, the decoding of this transformation in the remainder of the visual system then seems to hold of the miracle.

See too

Random links: