Authors :
Presenting Author: Cigdem Isitan Alkawadri, MD – University of Pittsburgh Medical Center
Rafeed Alkawadri, MD – Associate Professor of Neurology, HBM, University of Pittsburgh Medical Center; Qi Yan, BS – Yale Unviersity
Rationale: This study aimed to compare the signal characteristics of subdural and depth electrodes to gain an understanding of how contact type and size influence intracranial electroencephalography (icEEG) recordings. This insight is essential for the diagnosis, treatment, and development of devices used for neurological diseases.
Methods:
From a single surgery center database, 54 subject cases between 2012 and 2018 met the inclusion criteria. Subdural electrode contacts were cup-shaped with a 4.00mm diameter, a 2.30mm effective exposure diameter, and a total effective recording area 0.042 cm
2 of 0.13 cm
2 total. Depth electrodes had contacts 2.41mm long, 1.10mm in diameter, and 0.081 cm2 in the effective recording area. Out of 134 electrode contacts within a 5 mm vicinity, 24 (12 pairs) met signal-to-noise ratio and data availability criteria and were included in the analysis. Welch's method with 0.5-second moving windows with at least one sample overlap was used to calculate power spectral densities from a full 1-minute EEG signal (timeless) during NREM sleep on the third evaluation day. Additionally, root-mean squares (RMS) were calculated for one second moving windows for different frequency bands (infraslow IS, delta-theta DT, alpha-beta AB, low gamma LG, and high gamma HG; time-aware).
Results:
There was a median distance of 3.7mm between the included electrode-contact pairs, with an interquartile range (IQR) of 3.1-4.2mm. Power spectral densities between depth/subdural electrodes had a strong correlation with a median correlation coefficient (r) of 0.99 and an IQR of 0.915 to 0.996. In the time-aware analysis, the correlation was most robust within the high gamma (HG) band with a median r of 0.889 and an IQR of 0.831-0.996. Power ratios between subdural/depth electrodes had a median ratio for total PSD normalized by frequency of 1.99 and an IQR of 1.27-2.27. The power ratios ramped down from HG to infraslow waves, with the power ratios having a median of 2.07 and 0.97 and IQRs of 1.22-3.38 and 0.57-1.33, respectively.
Conclusions
The findings demonstrate a significant correlation between the signals recorded by standard subdural and depth electrodes within a 5 mm vicinity. This correlation was most significant in the HG band, essential for assessing neuronal activity. Nevertheless, the correlation decreased with decreasing frequency. The power ratios suggest that electrode size may not be the primary factor, but rather, considerations such as electrode design, placement methodology, the bandwidth of interest, and the characteristics of the underlying source of the measured activity. Subdural electrodes may be more suitable for high frequencies, however, the size effects of the electrodes on recording potentials must be considered in context-specific scenarios.
Funding: NA