How Radiens Calculates Signal Metrics (KPIs)

How Allego and Videre measure noise, RMS, spike events and SNR, and how values are combined online in Allego and offline in Videre.

Documentation Videos Downloads Glossary

Reference

v1.0

6 min read

Updated October 22, 2026

16px

Radiens reports per-channel signal-quality metrics (KPIs) such as noise, RMS, event rate and SNR. This reference explains how each metric is measured, how Allego combines measurements online as the signal streams in, how Videre combines them exactly offline from a recorded file, and when a metric reports no value.

The metrics appear in the Signal Metrics module of Allego and Videre. You choose a lookback, the seconds of recent signal the metrics describe, from 0.5 to 10 s (the default is 1 s). Allego estimates the metrics over the incoming signal while it streams, whether or not it is recording. Videre computes them exactly from a recorded file (how the two differ). Both apps process each channel in short consecutive blocks of samples.

Metric reference#

In the formulas, the window holds $N$ samples $x_i$ and $K$ detected events, $w_k$ is the waveform of event $k$ over the spike waveform window, and $T$ is the window duration.

All amplitude values are in microvolts (µV) and are measured on the filtered signal. Event Rate is in events per second (Hz), and Signal To Noise has no unit. Which metrics are offered depends on your Radiens plan: some plans offer only Noise and RMS, or leave out the event mean and peak-to-peak metrics.

Metric (as shown in the app)	Formula	What it measures
Mean	$\frac{1}{N}\sum_i x_i$	Average signal level. A mean that is large compared with the standard deviation of the samples indicates a DC offset.
Min, Max	$\min_i x_i$ , $\max_i x_i$	Lowest and highest sample values.
Max (abs)	$\max(\lvert\text{Min}\rvert, \lvert\text{Max}\rvert)$	Largest excursion in either direction.
Peak-to-Peak	$\text{Max} - \text{Min}$	Full range of the signal.
RMS	$\sqrt{\frac{1}{N}\sum_i x_i^2}$	Overall signal level, spikes and any DC offset included.
Noise (µV)	$1.4826 \cdot \operatorname{median}_i \lvert x_i - \operatorname{median}(x) \rvert$ , per block	Background noise floor. Equals the standard deviation for Gaussian noise, but spikes and brief artifacts barely move it.
Event Rate	$K / T$	Detected events per second.
Event Max, Event Min	$\max_k \max w_k$ , $\min_k \min w_k$	Most extreme event sample in each direction.
Event Max (abs)	$\max(\lvert\text{Event Max}\rvert, \lvert\text{Event Min}\rvert)$	Largest event excursion.
Event Max Peak-to-Peak	$\max_k (\max w_k - \min w_k)$	Peak-to-peak amplitude of the largest single event.
Event Mean Max (abs)	$\frac{1}{K}\sum_k \max \lvert w_k \rvert$	Typical event size.
Event Mean Peak-to-Peak	$\frac{1}{K}\sum_k (\max w_k - \min w_k)$	Typical event peak-to-peak amplitude.
Signal To Noise (SNR)	$\text{Event Mean Max (abs)} / \text{Noise}$	Typical event size relative to the noise floor.

Signal To Noise uses the mean event size rather than the largest event, because the largest event grows with the window length and the mean does not.

What changes the metrics#

These settings determine the metrics. Changing any of them makes Videre recalculate the open file and restarts Allego's running estimates.

Visualization (stage 2) filters. Metrics are measured on the filtered signal, so the filter band changes the numbers: noise and SNR measured with a 300 Hz–6 kHz band differ from the same channel measured broadband. Compare channels and sessions only under the same filters. Hiding the filters on a plot does not affect the metrics. In Allego, the hardware and Data (stage 1) filters are already in the samples, and a recording saves them that way.
Spike thresholds. An event is a crossing of a channel's negative or positive threshold, set in microvolts or as a multiple of noise (for example −4 × noise). A multiple of noise uses the channel's noise level averaged over about 2 seconds. Every event metric depends on the thresholds.
Artifact rejection. When it is on, an event whose waveform exceeds the artifact thresholds is discarded.
Spike waveform window. Event amplitudes are measured over this window, and detection skips one window after each event so one spike is not counted twice.

Allego detects events only on channels selected for display, so other channels report an Event Rate of 0. Videre detects on every channel. Detection stops at 550 events per second per channel.

How Allego and Videre compute over the lookback#

Videre combines every block in the window, which gives the same result as measuring the whole window in one pass. Allego adds each new block to a running estimate.

Metric	Videre: exact over the window	Allego: running estimate
Mean, RMS	True values over every sample in the window	Exponential moving average
Min, Max	Minimum or maximum across blocks	Peak envelope: jumps to a new extreme, otherwise moves toward the block's own extreme at the same rate as the moving averages
Event Max, Event Min, Event Max Peak-to-Peak	Extremes across blocks	The same envelope, updated only on blocks with events. Holds its value between events.
Event Rate	Total events ÷ total duration	Moving average. Blocks without events pull it toward 0.
Event Mean Max (abs), Event Mean Peak-to-Peak	Mean over every event in the window	Moving average, updated only on blocks with events
Noise (µV)	Average of the block estimates, weighted by sample count	Moving average
Signal To Noise	Combined event mean ÷ combined noise	The same ratio, from the running values

In Videre, the window is the lookback ending at the playback position. Through the Radiens API, the same exact computation is available over any time range.

In Allego, the lookback is how long the running estimate takes to reflect 95% of a sudden change. Allego does not save the metrics, even while recording, so open the recording in Videre for exact values.

When a metric shows no value#

A blank means the value was not measured:

No data yet, for example when Allego has just started or restarted streaming.
No events in the window (Videre): SNR and the event amplitude metrics show no value. Event Rate shows 0, because no events is a measured rate.
No events yet (Allego): SNR and the event amplitude metrics show no value until the channel's first event. After that they keep their last values when events stop, while Event Rate decays toward 0.
Large DC offset: when |Mean| ÷ the standard deviation of the samples exceeds 0.5, the channel's baseline is offset far enough that threshold detection is unreliable. All event metrics (Event Rate, the event amplitude metrics, and SNR) are withheld for that channel. Signal and noise metrics are still reported. An ordinary electrode offset is enough to trigger this, so apply a high-pass Visualization filter when you want event metrics.