Ординатура / Офтальмология / Английские материалы / Electrophysiology of Vision_Lam_2005

inner retina. A proposed model of the cellular origin of the multifocal ERG components is presented in Fig. 2.4. (22). Like the a-wave of the photopic full-field flash response, N1 receives contribution from the hyperpolarization of cone photoreceptors and OFF-bipolar cells. Similar to the photopic b-wave, P1 is influenced mostly by the depolarization of the ON-bipolar cells.

The first-order kernel also contains contributions from ganglion cells and the optic nerve head but these waveforms are small and not readily recognizable with typical clinical recording protocols. The clinical value of these components still warrants further investigation, and they are mentioned under the section on ‘‘Specialized Multifocal ERG Techniques and Waveforms.’’

DISPLAYING MULTIFOCAL ERG RESULTS

The trace array is the true representation of the calculated topographical ERG responses and is the most useful display of multifocal ERG results (Fig. 2.4). Amplitudes and implicit times of each response are readily available. Software can also generate a combined response from a chosen group of hexagons. The size of each hexagon is scaled to produce approximately equal ERG responses. The amplitudes of the N1 and P1 components of each hexagon are similar but falls off somewhat with eccentricity. When response amplitudes are calculated per retinal area, response density is highest in the fovea corresponding to the highest density of cone photoreceptors. Implicit times of P1 are high at the blind spot, the upper and lower borders of the stimulus, and the macula (23). Low P1 implicit times are found in the area encircling the macula.

The three-dimensional (3D) plot is the most colorful eyecatching display of multifocal ERG results but may be misleading. The 3D plot is helpful in visualizing the blind spot and scotomas. However, the 3D plot should always be accompanied by the trace array and should never be the sole display of multifocal ERG data. The 3D plot is obtained by dividing

the response amplitude by the area of the hexagon (response density, nV=deg2) (Fig. 2.7). Because the hexagonal area is the smallest at the center, the amplitude-to-area ratio may be falsely elevated at the center resulting in a small artificial center peak when all responses are markedly reduced. An artificial center peak may also occur with noisy recordings since the 3D plots typically combine the amplitudes of both positive and negative components and cannot distinguish components from electrical noise. Combining amplitudes of positive and negative components causes loss of information. Actual spatial resolution of the responses is also compromised when responses are interpolated to give the appearance of a continuous 3D surface. In addition, the appearance of the 3D plot is dependent on what reference template was used to scale the plot, called the scaler template. A reference template from averaged data of normal control subjects is better than a template from other patients.

Another way of displaying multifocal data involves response density expressed as amplitude per degree area (nV=deg2) from the center and grouped responses from concentric rings (Fig. 2.7). These response density plots are helpful to evaluate responses with respect to eccentricity from fixation but should be used with caution. Although the highest response density is found in the fovea due to high cone photoreceptor density, the summed response is actually lowest in the fovea and increases with eccentric rings as responses from progressively larger retinal area are summed. This type of data representation obscures isolated scotomas unless the scotoma is in the center. The representation also does not account for asymmetric responses due to disease or the physiologic naso-temporal asymmetry that may occur in normal recordings.

PHYSIOLOGIC BLIND SPOT IN

MULTIFOCAL ERG

The physiologic blind spot is not sharply delineated by the multifocal ERG. The optic nerve head reflects more light than

the retina, and this scattered light produces multifocal ERG responses from adjacent retinal areas that are calculated as small and delayed responses attributed to the region of the blind spot (Fig. 2.4) (24). In addition, the optic nerve head does not completely cover adjacent hexagonal stimulus areas and this could also contribute to an indistinct blind spot. The second-order responses are absent at the blind spot because no short-term adaptation response to a preceding flash in this non-retinal region is possible (Fig. 2.8).

SECOND-ORDER ‘‘RESPONSE’’ OF THE

MULTIFOCAL ERG

Like all multifocal waveforms, the second-order response, also known as second-order kernel (K2), is not a true recorded response but a mathematical derivation. Although the second-order response is perhaps more closely related to the pattern ERG, not all aspects of the physiologic contribution to this mathematical extraction are understood (25). The clinical utility of the second-order response continues to be developed, and whether it is more sensitive and specific than other clinical measures remains to be determined.

While the first-order response is calculated to determine the response of a flash, that is, a hexagonal white frame, the second-order response calculates the effect of successive flashes. The first slice of the second-order kernel calculates the effect of a flash in the immediately preceding frame while the second slice calculates the effect of a flash in the frame before the immediately preceding frame. As most discussions about the second-order kernel deal with the first-slice, the ‘‘second-order response’’ is usually used to refer to be the first slice of the second-order kernel unless specified otherwise.

The second-order response measures the short-term adaptation from a preceding flash. For example, the effect of a flash in the immediately preceding frame causes the normal multifocal ERG flash response to be slightly smaller and faster. The second-order kernel is calculated by adding all of the calculated records following a frame change from either

white-to-black or black-to-white and subtracting all calculated records following no change (Fig. 2.8). The second-order response is smaller than the first-order response and, similar to the first-order response, lasts much longer than the duration of one frame (13.3 msec at 75 Hz). In fact, the first-order response overlaps with its own induced second-order response and actually contains an inverted copy of the second-order kernel referred to as the ‘‘induced component’’ (26).

The second-order response is related to the activity of the inner retina and ganglion cells. In monkeys, the second-order response is reduced but not eliminated by tetrodotoxin (TTX) which abolishes retinal action potentials generated by ganglion cells and perhaps some amacrine cells. The second-order response is reduced in ganglion cell disorders, and further investigation of its clinical utility is required.

SPECIALIZED MULTIFOCAL ERG TECHNIQUES

AND WAVEFORMS

Several other multifocal ERG waveforms are recognized. Specialized recordings and algorithms for extracting signals are required to maximize these waveforms as they are not readily available with typical clinical recording protocols.

Figure 2.8 (Facing page) The second-order response (secondorder kernel, K2) of the multifocal ERG. Top: Schematic representation of how the second-order response (second-order kernel, K2) of the multifocal ERG is calculated for a specific hexagon. The first slice of the second-order kernel calculates the effect of a flash in the immediately preceding frame. The second-order kernel is calculated by adding all of the calculated records following a frame change from either white-to-black or black-to-white and subtracting all calculated records following no change. Similar to the first-order response, the second-order response lasts much longer than the duration of one frame (13.3 msec at 75 Hz). Bottom: Normal trace array of the second-order response is shown. The calculated sec- ond-order response is smaller than the first-order response, and in contrast to the first-order response, the second-order response is not present at the blind spot.

Optic Nerve Head Component

The first-order kernel also contains contributions from ganglion cells and the optic nerve head. The strength of these contributions is related to the distance of the retinal region to the optic nerve head, and generally, the components have higher amplitudes and shorter implicit times in the nasal region of the retina as compared to the temporal region. When retinal action potentials are abolished by TTX which blocks voltage-gated sodium channels in monkeys, the physiologic naso-temporal asymmetry of the first-order responses resolves suggesting that action potentials generated by ganglion cells and perhaps some amacrine cells are likely responsible (27).

The optic nerve head component (ONHC) of the multifocal ERG is thought to originate from the beginning of axonal myelination near the optic nerve head. This component can be better isolated with an interleaved global flash protocol that inserts all-white and all-black frames into the multifocal stimulus presentations (Fig. 2.9) (26). The delay in ONHC is related to the length of the unmyelinated nerve fibers that action potentials must travel between the focal stimulation and the nerve head. The ONHC is reduced in optic neuropathies such as glaucoma, and its clinical utility requires further investigation (28).

The s-Wave

The s-wave (‘‘s’’ for small) is a positive wavelet on the descending limb of P1 of the first-order response that appears when the stimulus frequency is decreased from 75 to 18 Hz or lower by inserting ‘‘blank’’ frames into the multifocal stimulus presentations. The ‘‘blank’’ frames are all-gray frames that would tend not to elicit a multifocal ERG response (Fig. 2.10) (29). The s-wave is likely to originate from the neural activity of the ganglion cells.

Other Specialized Techniques

The multifocal ERG can provide topographical oscillatory potentials which are best detected by a decreased stimulus

frequency and a higher low-end bandpass to remove other ERG components (30). Both first-order and second-order analysis of the oscillatory potentials are possible, and the topographic distribution of the second-order oscillatory potentials shows combined features of both rod and cone distributions. The oscillatory potentials are generally higher in the temporal retina than corresponding nasal locations (31).

The distribution of ONand OFF-responses can be studied with the multifocal ERG by a low stimulus frequency of 5 Hz or lower to simulate long duration stimulus (32). A study of a-, b-, and d-wave components shows different spatial distribution across the retina (33). However, with the CRT monitors used for multifocal ERG, the luminance of each white hexagon is optimally maintained only during the first few milliseconds after the frame change and then decreases subsequently; this may hamper the simulation of a long duration stimulus.

Dark-adapted multifocal rod ERG recordings are possible but require the use of blue flashes and modified conditions to allow rod recovery and to reduce the effect of stray light (34). The signal-to-noise ratios of rod ERG recordings are worse than cone recordings.

Figure 2.10 (Facing page) The s-wave. The summated waves (alltrace waves) of the individual retinal multifocal ERG first-order kernel for different stimulus frequencies from a normal eye are shown. The s-wave (black arrows) is a positive wavelet on the descending limb of P1 detectable with stimulus frequencies equal to or less than 18 Hz (typical clinical stimulus frequency ¼ 75 Hz). The s-wave is likely related to activities of the ganglion cells. The amplitude of the s-wave (white arrows) may be measured as the height of a vertical line from the peak of the wavelet to where it intersects a line connecting the troughs of successive negative waves on either side of the wavelet. (From Ref. 29, reproduced with permission of Investigative Ophthalmology & Visual Science via Copyright Clearance Center.)