Comments on Rod b-wave series

-- DavidKeating - 20 Oct 2008

As the scotopic series is specified using photopic luminance there needs also to be a specification of the spectral composition of the light and of the photopic/scotopic ratio to achieve comparable results between laboratories.

John RObson is currently chairing the Committee of ISCEV TEchnical Guidelines (update of the ISCEV calibration guidelines) and has suggested a photopic: scotopic ratio 2.52 based on a colour temperature of 7000°K. Extended protocols should co-ordinate with the work of John's committee.

-- DaphneMcCulloch - 30 Apr 2009

Is there any reason for a "RodBwaveseries" AND a "RodVlogI" module? I think we can exclude one. My suggestion is to keep the "RodVLogI".

-- AndreMessias - 12 May 2009

Of course there is a scotopic B-wave series and a scotopic a-wave series that are the same as what is often called the "Rod V logI" for the b- and a-waves respectively. I prefer 'scotopic' in the name as the responses are initiated by rods but actually generated by the bipolar and other cells. Also, 'V logI' is not understood by non-experts and can be confusing as it refers as to b-wave amplitude ('V') versus the time-integrated luminance of a flash stimulus ('I'). It also encourages the common error of referring to the extended light flash as 'intensity', the photometric term that applies to point sources.

-- DaphneMcCulloch - 12 Jun 2009

I agree that there is a "b-Wave Series" and a "a-Wave Series", but my question was about the existence of a protocol number 11 (Rod b-wave series) AND a protocol number 12 (Rod b-wave VlogI? ) additionally to the number 13 (a-wave analysis). I also agree with you about the name “VLogI”, so we could have a protocol number 11 called Scotopic b-Wave Series, AND a protocol number 13 called Scotopic a-Wave Series, and remove the protocol number 12.

-- AndreMessias - 18 Jun 2009

The model fitting: (1) this seems to be the sum of two Hill functions with the Hill coefficient set to one, right? It fits rather well to my own data. However, (2) the saturation levels V2 is given incorrectly on the axis: when we accept the definition from the formula, then what is depicted on the axis is (V1+V2)

-- MichaelBach - 15 Jul 2009

-- JohnRobson - 08 Apr 2010

VlogE?

Although I must admit to having suggested that b-wave V-logE (E for energy) data would be better described by the sum of two saturating functions than by the single (hyperbolic, or exponentiated hyperbolic) function that has until now usually been used, I have to say that, on reflection, I am not at all happy that ISCEV should be proposing the use of any function whose theoretical basis is quite as dubious as this is without it being quite clear what is being done.

The fact is that we know quite enough about the dark-adapted ERG to be reasonably certain that the growth of b-wave amplitude (as conventionally measured) is normally dependent upon at least four distinct processes or mechanisms and should therefore properly be modeled as the sum of at least four functions. These process are: 1) at the lowest stimulus energies the ERG is a reflection of the activity of proximal retinal cells which generate both positive and negative signals that are manifested in the ERG as the STR (scotopic threshold response); 2) at somewhat higher energies the b-wave amplitude is mainly determined by the amplitude of the positive-going response of rod-bipolar cells (PII); 3) at higher levels still the increase in b-wave amplitude is limited (or in some species usually reversed) by the increase in the amplitude of negative (PIII) signals from rods while 4) at the highest energy levels the b-wave amplitude increases again as the initial negative “nose” of the direct rod component becomes significant.

The whole question of what mathematical function should be used as a template for parameterising b-wave amplitude data is therefore not a simple one. If one takes a modeling approach then the first thing to do would be to characterise the behaviour of the various process or mechanisms (as listed above) whose signals contribute to the measured amplitude. But this is not simply a question of determining how the (peak) amplitude of the signals of these various mechanisms depends upon stimulus energy because conventional b-wave amplitude is (largely for historical reasons) measured not at fixed times or at the times of the peaks of the component signals but at the times of the peaks in the composite waveform. This means that a model that could satisfactorily describe no more than the amplitude of the ERG peaks would have to be based on adequate descriptions of the more-or-less complete time course of the component signals. I do not believe that we yet have these descriptions.

An alternative approach would be an entirely empirical one in which b-wave V-E data from a representative (and suitably large) sample of both normal and patient populations would be used to determine the best way to parameterise the data from individual data sets. This would require not only that the sample data be available but also that some more precise meaning be ascribed to the word “best”. In a clinical context I think that “best” would most appropriately imply “providing the most useful diagnostic indicator”, but I am far from clear how one would ascertain this in practice. Maybe one should just look for the parameterisation that provided the tightest clustering of the normals or the greatest separation between normals and some or all of the others. But whatever the appropriate interpretation of “best” is, I am sure that it is necessary to start with a clear idea of what it is hoped to achieve. I do not think we really have that at the moment.

At an even more radical level I seriously wonder if we should not be giving some serious thought as to whether the conventional trough–to-peak measure of b-wave amplitude is really the best way to parameterise ERG records in the first place, I have no doubt in my own mind (though based entirely on our experience with experimental animals rather than on any clinical experience) that it would be better to make measurements of the positive b-wave peak at a single fixed time, probably at the time of the peak produced by stimuli of intermediate energy in the range over which the time-to-peak is independent of stimulus strength. At this time the b-wave amplitude (from baseline) can be adequately modeled as the sum of a constant (the saturated amplitude of the STR at that time), PII and an adequate characterized of PIII at that time. Using this approach I would expect it to be possible to obtain more reliable estimates of both the sensitivity and saturation level for PII which is, I presume, the immediate goal of the proposed protocol.

-- JohnRobson - 08 Apr 2010