EEG evaluation of focal interictal epileptiform transients (FIET) can be objectified
Abstract number :
1.116
Submission category :
3. Neurophysiology / 3C. Other Clinical EEG
Year :
2016
Submission ID :
186894
Source :
www.aesnet.org
Presentation date :
12/3/2016 12:00:00 AM
Published date :
Nov 21, 2016, 18:00 PM
Authors :
Fumisuke Matsuo, University of Utah School of Medicine, Salt Lake City, Utah
Rationale: Focal spikes/sharp waves (SSW) are interictal epileptiform transients predictive of diagnosis of partial epilepsy. Their detection has depended on polygraphic waveform analysis by experienced expert viewer, and multiple factors limit their reliability and sensitivity. Aiming at objectifying their detection, randomly chosen SSW have been subjected to several analysis protocols and compared with FIET that mimic SSW, atypical wicket spikes (aWS). It was possible to rank-order mixed FIET samples according to likelihood of clinical diagnosis of epilepsy, when sorted by waveform in 2 steps, first, in conventional, single representative channel display, and second, in polygraphic channel overlay (PGCO) display (2014 AES abstract). FIET segmentation, B, P, T and W (base-peak-trough-wave), was previously defined on single representative channel display of exemplar SSW (2005 and 2012 AES abstracts). In common reference PGCO display, each segment consists of in- and out-of-phase deflections, varying in amplitude and enveloped by dipolar derivations, particularly well-differentiated from background in some SSW associated with relatively deep intracranial generator matrix. Voltage difference between dipolar surface electrode pairs (VDd) can be readily measured at deflection points (Figure), and is indifferent to choice of common reference. PGCO in common average reference derivations provides convenient, close approximation of VDd, and can serve as surrogate measure for 3 SSW segments. This study examined mixed FIET samples, comparing 2 factors that have been suggested to affect reliability of SSW detection, 1) FIET amplitude, and 2) W-segment differentiation. Methods: Original series of SSW and aWS samples were combined, and then reduced to random samples of 121 FIET by quicksort. 23 head-surface EEG recording electrode placements included pairs at zygomatic and mastoid bones. Each digital EEG contributed single representative FIET. SSW with B-T duration longer than 150 ms had been excluded. PGCO was manually prepared from digital image by superimposing polygraphic channels under careful control of baseline and temporal grid. VDd was converted from measurements of P-, T- and W-segment amplitude on PGCO image (see Figure for example). Clinical diagnostic data were reviewed up to 10 years following collection and original review of EEG data Results: Clinical data were sufficient to determine diagnosis for 99 FIET, epilepsy (E: 67) and non-epilepsy (N: 32). P was always differentiated by definition, and FIET with differentiation of W-segment by visual inspection of PGCO against baseline-background (PTW: 54) formed subgroup. VDd for PTW (VDd-PTW) was sum of VDd-P, VDd-T and VDd-W. FIET groups were further divided by VDd to 2 halves, low and high. When compared with base ratio (E/N=67/32), diagnostic discrimination against aWS was excellent for PTW with high VDd (E/N = 24/3), followed by P with high VDd (E/N=38/11), and then PTW with low VDd (E/N=18/9). Of 121 FIET, 72 were originally reported as SSW with median values for VDd 92 uv and 250 uv, for P (n=72) and PTW (n=52), respectively. Conclusions: Combination of high VDd and W-segment differentiation could reliably identify SSW, and can possibly predict E diagnosis independently of clinical data. PGCO reformatting is available in commercial EEG review program. Measurement of VDd is reliable, because PGCO reformatting can be standardized for off-line digital EEG analysis, and VDd is independent of choice of common reference. Method proposed has not been tested for differentiation of SSW from physiological EEG transients. It requires relatively simple pre-screening for FIET, once VDd threshold value has been confirmed for reliable of SSW detection. Funding: None
Neurophysiology