Abstracts

Rationale: Prompt administration of the radiotracer used in Ictal SPECT studies following the electrographic onset of a seizure may enhance the localization of epileptogenic foci. Ictal SPECT protocols that rely on a caregiver recognizing the clinical onset of a seizure result in variable and delayed injection of the radiotracer; these delays contribute to poor localization of the seizure focus. To achieve consistent and prompt injection of radiotracers we developed and prospectively evaluated at Children’s Hospital Boston a computerized system that automatically initiates radiotracer injection following fully automated real-time detection of the electrographic onset of a seizure.Methods: In our prospective evaluation a clinical team carries out Ictal SPECT studies in accordance with an existing protocol at Children’s Hospital Boston; this protocol includes capturing the scalp EEG of the Ictal SPECT study participant. In parallel and unobserved by the clinical team, our computerized system analyzes the collected EEG in real-time using a patient-specific seizure detection algorithm (Shoeb A et al Epilepsy Behav 2004; 5:483-98). Following the onset of a seizure, the clinical team infuses the Ictal SPECT study participant with the appropriate radiotracer dose. Detection of the electrographic onset of the same seizure by the computerized system triggers the infusion of a surrogate radiotracer (water) into a plastic receptacle using a computer controlled drug infusion pump (Harvard Clinical Harvard 2 Pump). At the conclusion of the Ictal SPECT study, we compare the time elapsed between electrographic seizure onset and radiotracer infusion for both the clinical team and the computerized system.Results: We evaluated the performance of the computerized system on eight Ictal SPECT studies. Figure 1 shows the time elapsed between electrographic seizure onset and dispensing of the radiotracer dose for both the computerized system (column 2) and the Ictal SPECT clinical team (column 3); also shown in Figure 1 is the delay with which the computerized system first recognizes the electrographic onset of a seizure (column 1). Our computerized system recognized the electrographic onset of a seizure within 5.1 +/- 2.8 seconds and completed dispensing the appropriate radiotracer volume within 19.3 +/- 2.3 seconds. The computerized system incorrectly dispensed the surrogate radiotracer on 4 occasions over the course of 33 hours (approximately 4 hours of monitoring for each of the eight Ictal SPECT studies). The Ictal SPECT clinical team completed infusion of the radiotracer within 27.7 +/- 8.5 seconds. For patient number 7 the first seizure went unnoticed by the clinical team (absent column 3), but was detected by the computerized system. Missing the first seizure of patient 7 prompted the need for a second Ictal SPECT Study, for which the clinical team and system detected the seizure.Conclusions: Neither system dominated the other. The automated system was slightly more sensitive and had slightly better latency, but was less specific. This suggests a hybrid system in which automated detection is used to alert a clinical team.