51
extrapolation down to the level of fluorescence of the non-fluorescent beads spans more that
one decade of fluorescence. It is unlikely that most log amplifiers are "linear" in the lowest
decade. Thus very small changes in the position or resolution of the non-fluorescent beads can
lead to large differences in the estimation of sensitivity.
The results are much more encouraging for determination of linearity, which many would
argue is really more important. In the published survey, all the tested instruments proved to
have r2 values of greater than 0.99 and there was very little variation between instruments or
within instrument models. Thus the basic tenant of flow cytometry, that the amplitude of the
fluorescent signal is proportional to the amount of the fluorochrome, seems to be a good one
in practice. Again this is limited by the caveat that the instruments are being tested primarily
in the range of greater than 10,000 fluorochrome equivalents and linearity almost certainly
does not extend down to the limits of fluorescence detection, at least using log amplifiers.
Another, more economical approach to the question of linearity can be made by using
particles, usually beads, that are partially aggregated. When this is done, one obtains a series
of peaks which should be spaced uniformly corresponding to singlets, doublets, triplets, etc.
If the coefficient of variation of the beads is sufficiently small, these peaks can be resolved
cleanly and can be used to determine linearity over a limited range.
For the purposes of determining both sensitivity and linearity, it is not necessary for the beads
to be calibrated in terms of any absolute fluorescence standard. Thus the fact that the
commercial beads provide a result in terms of "molecules of equivalent soluble fluorochrome"
is not really relevant to our current discussion. The special problems introduced by this type
of calibration will be discussed below.
COMPENSATION
Another instrument variable that must be checked in the process of obtaining quantitative data
from a flow cytometer is that of compensation. By this is meant the correction of the
measured fluorescence signals for signals spilling from another fluorochrome. This problem
arises because we generally identify each fluorochrome by measuring its emission through a
band pass filter. Some fluorochromes have wide emission spectra and may "spill" their
emission into the band pass of a filter designed to measure another fluorochrome. The
correction for this is usually done in hardware by subtracting a fraction of the measured signal
in one channel from the measured signal in another. The fraction subtracted is set empirically
by the operator so that each fluorochrome, when analyzed by itself, gives a signal only in its
extrapolation down to the level of fluorescence of the non-fluorescent beads spans more that
one decade of fluorescence. It is unlikely that most log amplifiers are "linear" in the lowest
decade. Thus very small changes in the position or resolution of the non-fluorescent beads can
lead to large differences in the estimation of sensitivity.
The results are much more encouraging for determination of linearity, which many would
argue is really more important. In the published survey, all the tested instruments proved to
have r2 values of greater than 0.99 and there was very little variation between instruments or
within instrument models. Thus the basic tenant of flow cytometry, that the amplitude of the
fluorescent signal is proportional to the amount of the fluorochrome, seems to be a good one
in practice. Again this is limited by the caveat that the instruments are being tested primarily
in the range of greater than 10,000 fluorochrome equivalents and linearity almost certainly
does not extend down to the limits of fluorescence detection, at least using log amplifiers.
Another, more economical approach to the question of linearity can be made by using
particles, usually beads, that are partially aggregated. When this is done, one obtains a series
of peaks which should be spaced uniformly corresponding to singlets, doublets, triplets, etc.
If the coefficient of variation of the beads is sufficiently small, these peaks can be resolved
cleanly and can be used to determine linearity over a limited range.
For the purposes of determining both sensitivity and linearity, it is not necessary for the beads
to be calibrated in terms of any absolute fluorescence standard. Thus the fact that the
commercial beads provide a result in terms of "molecules of equivalent soluble fluorochrome"
is not really relevant to our current discussion. The special problems introduced by this type
of calibration will be discussed below.
COMPENSATION
Another instrument variable that must be checked in the process of obtaining quantitative data
from a flow cytometer is that of compensation. By this is meant the correction of the
measured fluorescence signals for signals spilling from another fluorochrome. This problem
arises because we generally identify each fluorochrome by measuring its emission through a
band pass filter. Some fluorochromes have wide emission spectra and may "spill" their
emission into the band pass of a filter designed to measure another fluorochrome. The
correction for this is usually done in hardware by subtracting a fraction of the measured signal
in one channel from the measured signal in another. The fraction subtracted is set empirically
by the operator so that each fluorochrome, when analyzed by itself, gives a signal only in its