Abstracts

Rationale: Certain FIET waveform features are reliable EEG predictors of epilepsy (spikes/sharp waves: SSW), while others, SSW mimickers, often referred to as wicket spikes (WS). Several digital EEG analysis protocols have been tested to objectify polygraphic SSW assessment, and have been applied to FIET other than SSW. Preliminary results suggested significant overlap between SSW and WS. Mixed FIET samples were subjected to 3 analysis protocols to examine nature of overlap and difference between SSW and WS. Methods: FIET collections consisted of 81 representative SSW (base-trough duration less than 150 ms) from single reviewer-EEG cases, and 90 WS from separate EEG cases, chosen for features mimicking SSW. 171 FIET were randomized, subjected to preliminary ranking for polygraphic SSW features, and top 121 FIET were retained, consisting of 72 SSW and 49 WS. Protocol 1 was time-series waveform analysis by expert-viewer. Each FIET was displayed in 2-sec polygraphic serial bipolar and common average derivations. Reviewer, blinded, selected 11 best-formed SSW from 121 FIET, and repeated by selecting 11 best from remaining FIET with 11 left after 10th selection. Protocol 2 was FIET waveform scoring against BPTW (Base-Peak-Trough-Wave) template, derived from 5 best-formed SSW. One representative FIET was chosen in common average derivation. Protocol 3 examined spatiotemporal characteristics by following FIET peak movement, because FIET often consisted of serial peaks. Same representative FIET were displayed in common average derivations at maximal temporal resolution (5 ms). FIET peaks were identified and localized by manually advancing cursor. Results: Protocols 1 produced FIET ranking with definite SSW and WS at opposite ends, separated by gradual transition (p<0.0001). Protocol 2 resulted in BPTW scores ranging from 25 to 4 (2 FIET groups differentiated at p = 0.0008). When examined as percentile rank change from Protocol 1 to 2, WS tended to rise in rank, while SSW, fall (p=0.0004). Protocol 3 detected FIET peak movement in both SSW and WS with exception of 6 FIET, and latency between first and last peaks varied up to 50 and 55 ms, respectively. Conclusions: Well-formed SSW were detected reliably under protocol 1 or 2, and tended to lack of peak movement. Conventional polygraphic montage-based FIET recognition-differentiation performed better in screening out WS mimicking SSW, but reliability of SSW-WS differentiation can be questioned. EEG background as context factor was minimized under Protocol 1, and eliminated in Protocol 2. Additional context factors, either reviewer-dependent or in clinical domain, are more difficult to control. One realistic solution may be to recognize FIET as spectrum, including intermediate categories as well as their co-existence, rather than to maintain SSW-WS dichotomy.