Normalization of Voltage-Sensitive Dye Signal with Functional Activity Measures

Abstract

In general, signal amplitude in optical imaging is normalized using the well-established ΔF/F method, where functional activity is divided by the total fluorescent light flux. This measure is used both directly, as a measure of population activity, and indirectly, to quantify spatial and spatiotemporal activity patterns. Despite its ubiquitous use, the stability and accuracy of this measure has not been validated for voltage-sensitive dye imaging of mammalian neocortex in vivo. In this report, we find that this normalization can introduce dynamic biases. In particular, the ΔF/F is influenced by dye staining quality, and the ratio is also unstable over the course of experiments. As methods to record and analyze optical imaging signals become more precise, such biases can have an increasingly pernicious impact on the accuracy of findings, especially in the comparison of cytoarchitechtonic areas, in area-of-activation measurements, and in plasticity or developmental experiments. These dynamic biases of the ΔF/F method may, to an extent, be mitigated by a novel method of normalization, ΔF/ΔFepileptiform. This normalization uses as a reference the measured activity of epileptiform spikes elicited by global disinhibition with bicuculline methiodide. Since this normalization is based on a functional measure, i.e. the signal amplitude of “hypersynchronized” bursts of activity in the cortical network, it is less influenced by staining of non-functional elements. We demonstrate that such a functional measure can better represent the amplitude of population mass action, and discuss alternative functional normalizations based on the amplitude of synchronized spontaneous sleep-like activity. These findings demonstrate that the traditional ΔF/F normalization of voltage-sensitive dye signals can introduce pernicious inaccuracies in the quantification of neural population activity. They further suggest that normalization-independent metrics such as waveform propagation patterns, oscillations in single detectors, and phase relationships between detector pairs may better capture the biological information which is obtained by high-sensitivity imaging.