Quantifying spectral peaks in theta and gamma brain oscillations detected from long term Local Field Potential (LFP) recordings in mice

Abstract

Most of the analysis methods for brain oscillations are performed on short (10–20 s) local field potential (LFP) recordings. However, a statistical approach on long duration recordings should be much more efficient. Therefore, we used LFP recordings lasting several weeks in 12/12 hours of light and dark cycles. We used Igor Pro (wavemetrics) software to process the LFP recordings. Over 12 hours periods, we used a 8 s sliding window in steps of 0.5 or 1 s. We performed spectral, spike, empirical mode decomposition (EMD), and many other analyses in each of the 8 s periods. Statistical analysis of the data allowed us to find out intrinsic characteristics deeply coded in the LFP recordings. We obtained characteristic brain oscillations by FIR (Finite Impulse Filtering) filtering the LFP, 5–12 Hz and 30–120 Hz for theta and gamma ranges, respectively. We calculated the Root Mean Square (RMS) values of each filtered wave in every 8 s sliding window. The histograms of these values showed a bimodal distribution, allowing us to detect an objective threshold between the low and high theta and/or gamma activity and to distinguish the correspondent periods accordingly. It has been previously shown that brain oscillation can be described not only by the spectral powers, but also by the separation of the spectral peaks (we termed this behavior “peakyness”) too. We have devised a novel procedure to evaluate the peakyness of the spectral components. We first implemented three different methods to measure the separation. 1. The “first derivative”, uses the maximum, minimum and RMS of the first derivate of the FFT spectrum in the characteristic interval (e. g., 5–12 Hz for theta). 2. The “second derivative or convexity method” analyses the convexity/concavity of the spectrum. 3. The most powerful, the “peak and area” method calculates the local maximum and its abscissa (frequency) and the area under the peak in the spectrum.