Semiologic Extrapolation in DS

The semiologic extrapolation function in the DS software implements a set of predetermined rules on the final report. Its aim is to identify specific semiologic elements which can be associated with a variety of event types, and at present it contains four functions, which identify features implying focal onset, PNES, the presence of Jacksonian march and semiologic risk factors for SUDEP. The final report includes an alert detailing specific semiologic features compatible with each of the enabled semiologic extrapolation functions. The semiologic extrapolation functions generate either a possibly positive statement or inconclusive (“don't know”) statement. They are designed as decision support functions, highlighting key semiologic elements, and always require clinical judgment and EEG considerations to assist diagnostic process. For example, if the video report includes all criteria for PNES and DS raises the possibility of PNES, but the ictal EEG is clearly epileptiform, the interpreter should ignore the DS suggestion and define the episode as an epileptic seizure.

The focality feature checks for the presence of semiologic characteristics of focal, rather than generalized seizure onset (Table 2.a). While there are many such characteristics, herein we chose semiologic descriptors which are highly specific. Thus, when observing motor semiologic features, only forced (rather than mild or non-forced) head and eye deviation are reported by the focality function (6). In addition, unilateral movement, whether tonic, clonic or dystonic, must appear at least 5 s before contralateral movement so that equivocal cases, in which it may be uncertain that the movement truly commenced unilaterally, are not falsely classified as focal. Similarly, unilateral hemiparesis must appear at least 5 s before contralateral involvement. Fencer and figure of four posturing which are highly lateralizing were also included. In its current form the focality function does not have localizing properties but alerts the user if features suggestive of focal rather than generalized onset were reported in the encoded episode.

Criteria implemented by the semiologic extrapolation functions in DS.

PNES, psychogenic non-epileptic seizures; SUDEP, sudden unexpected death in epilepsy; B/GTCS, bilateral/generalized tonic clonic seizures.

The PNES feature checks for characteristics shown to be typical of PNES (Table 2.b). Again, the features selected were such that are considered relatively specific in the literature. Eye closure during the ictal episode has been shown to be highly sensitive and specific for PNES (96.2 and 98.1% respectively) (7, 8). PNES are generally significantly longer then epileptic events (9), and a seizure length of over 5 min makes an event 24 times more likely to be PNES (10). Fully controlled falling and resistance to eye opening are also suggestive of PNES (11). Other semiologic characteristics such as head shaking or pelvic thrust were ultimately omitted from the PNES criteria due to low specificity, particularly when compared with frontal lobe epilepsy (11).

The Jacksonian march function (Table 2.c) identifies propagation of clonic body movements between body parts separated by at least 1 s intervals (the temporal resolution of DS).

The SUDEP risk function (Table 2.d) alerts in case of semiologic characteristics which have been shown to increase the risk of SUDEP. These include, first and foremost, the appearance of GTCS (12). GTCS out of sleep, apnea or cyanosis at the end of the GTCS (13) and GTCS ending in prone position (14) are noted as additional risk factors, though the evidence for their contribution to SUDEP risk is less robust.