An open-source, ready-to-use and validated ripple detector plugin for the Open Ephys GUI

Abstract

Sharp wave-ripples (SWRs, 100-250 Hz) are oscillatory events extracellularly recorded in the CA1 subfield of the hippocampus during sleep and quiet wakefulness. SWRs are thought to be involved in the dialogue between the hippocampus and cortical regions to promote memory consolidation during sleep and memory-guided decision making. Many studies employed closed-loop strategies to either detect and abolish SWRs within the hippocampus or manipulate other relevant areas upon ripple detection. However, the code and schematics necessary to replicate the detection system are not always available, which hinders the reproducibility of experiments among different research groups. Furthermore, information about performance is not usually reported. Here, we present the development and validation of an open-source, real-time ripple detection plugin integrated into the Open Ephys GUI. It contains a built-in movement detector based on accelerometer or electromyogram data that prevents false ripple events (due to chewing, grooming, or moving, for instance) from triggering the stimulation/manipulation device. To determine the accuracy of the detection algorithm, we first carried out simulations in Matlab with synthetic and real ripple recordings. Using a specific combination of detection parameters (amplitude threshold of 5 standard deviations above the mean, time threshold of 10 ms, and RMS block size of 7 samples), we obtained a 97% true positive rate and 2.48 false positives per minute on the real data. Next, an Open Ephys plugin based on the same detection algorithm was developed, and a closed-loop system was set up to evaluate the round trip (ripple onset-to-stimulation) latency over synthetic data. The lowest latency obtained was 34.5 ± 0.5 ms. Besides contributing to increased reproducibility, we anticipate that the developed ripple detector plugin will be helpful for many closed-loop applications in the field of systems neuroscience.

A consensus statement on detection of hippocampal sharp wave ripples and differentiation from other fast oscillations

“…With filter-based methods only approximately half of the larger amplitude ripples are detected 47,89 . Thus, novel strategies that improve detection quality are needed 40,83,91 . For some applications, detecting neuronal sequences, as opposed to LFP features, may be more reliable 92,93 , although online detection also faces the problem of short fragments.…”

“…For some applications, detecting neuronal sequences, as opposed to LFP features, may be more reliable 92,93 , although online detection also faces the problem of short fragments. The development of a publicly available ripple detection algorithm, tested by the community (perhaps dedicated platforms for rodents and humans, depending on constraints), is highly desirable 40,89 .…”

“…Methods vary from paper to paper (Supplementary Table S1 and see ref. 40), which likely drives much of the variance across laboratories. To discuss these challenges, we gathered more than 30 neuroscientists actively studying hippocampal SPW-Rs in rodents, non-human primates, and humans.…”