Skip to main content
. 2019 Feb 22;8:e42148. doi: 10.7554/eLife.42148

Figure 2. Classification of two types of high-frequency oscillations in animals with epilepsy.

A Density plot of high-frequency oscillations (HFOs) recorded from all sessions in control rats (n = 4 rats, n = 4 sessions). All events were characterized by a single cluster of similar frequency and envelope amplitude. (B) Density plot of HFOs from all sessions recorded from epileptic rats (n = 4 rats, n = 12 sessions). In animals with epilepsy, two clusters are present, one characterized by envelope amplitudes and frequencies that resemble control ripples (circled in blue), and the other characterized by higher frequencies and larger envelope amplitudes (circled in red), and see Figure – Supplement1 A-C. (C) Distribution of the oscillation frequencies for control ripples (cyan, n = 1414), and in animals with epilepsy (blue and red). The two distributions for animals with epilepsy were separated by clustering the data presented in (B), and see Figure – Supplement1D for display of individual animals. Note that the blue population overlaps with control ripples and is thus referred to as ‘ripple-like’ (n = 1616). The red population is separate from control ripples and is thus referred to as ‘pathological HFOs (pHFO)’ (n = 923). (D) The separation (or not) of each cluster in epileptic animals from control ripples is confirmed analytically using a binary classifier and plotted in a Receiver Operator Curve. True positive and false positive rates are plotted with the range of oscillation frequencies used as criteria values (see Materials and methods for more detail). (E, F) The same as (C, D), but for envelope amplitude rather than oscillation frequency.

Figure 2.

Figure 2—figure supplement 1. pHFO and ripple-like events are both present in each of the chronically epileptic animals.

Figure 2—figure supplement 1.

A Example ‘ripple’ event recorded from the CA1 cell layer of a control animal. Left: The raw LFP is shown. Middle: To calculate the amplitude of the envelope (the simultaneously occurring voltage deflection, which in control animals is called the ‘sharp wave’), the LFP window around the event was band-pass filtered between 0.2 Hz and 40 Hz, and the maximum absolute amplitude of the filtered data was taken. Right: To determine the ‘peak frequency’ the peak in power spectrum at frequencies higher than 150 Hz was selected. (B) Example ‘ripple-like’ event recorded from the CA1 cell layer of an epileptic animal. (C) Example ‘pHFO’ event recorded from the same recording site in the CA1 cell layer of the same animal as in (B). (B) and (C) confirm that both ‘ripple-like’ and pHFO events can be generated by the same hippocampal network. (D) Envelope amplitude vs. frequency plots for high-frequency oscillations recorded in each animal with chronic epilepsy. For each epileptic rat, events from one recording session recorded from a tetrode located in the CA1 layer. Each event is represented by a dot, color coded based on whether they belong to the ripple-like cluster (blue) or pHFO cluster (red), confirming that both types of events are present in individual animals.