Gabriella Colicchio

Real-Time Epileptic Seizure Prediction Using AR Models and Support Vector Machines

This paper addresses the prediction of epileptic seizures from the online analysis of EEG data. This problem is of paramount importance for the realization of monitoring/control units to be implanted on drug-resistant epileptic patients. The proposed solution relies in a novel way on autoregressive modeling of the EEG time series and combines a least-squares parameter estimator for EEG feature extraction along with a support vector machine (SVM) for binary classification between preictal/ictal and interictal states. This choice is characterized by low computational requirements compatible with a real-time implementation of the overall system. Moreover, experimental results on the Freiburg dataset exhibited correct prediction of all seizures (100 % sensitivity) and, due to a novel regularization of the SVM classifier based on the Kalman filter, also a low false alarm rate.

Interictal epileptic activity during sleep: a stereo-EEG study in patients with partial epilepsy

Cerebral electrical activity was recorded through chronic stereotactically implanted electrodes in 19 epileptic patients suffering from different types of severe and medically refractory partial seizures and who were considered for surgical treatment. 213 brain sites, in all cerebral lobes, in neocortical as well as in archicortical structures, were explored. The behaviour of the interictal spiking across wakefulness and nocturnal physiological sleep was analysed, using automatic elaboration. (i) Spike rate is affected by the occurrence of sleep and by the passage from one sleep phase to another. The degree and direction of the phenomenon differ remarkably in the various patients and, in the same patient, in the different cerebral sites explored. Generally, interictal spiking increases at the beginning of sleep, reaches its maximum during the deep non-REM phases and returns to a level slightly lower than that in wakefulness during REM. (ii) The nocturnal spike rate is hardly influenced by spike location. In most cases, however, the variations recorded during sleep are more significant in the frontal regions than elsewhere. (iii) Spike rate across wakefulness and sleep is affected by the local level of epileptogenicity: spiking variations are less in the most epileptogenic cerebral zone (identified by the origin of the seizure discharges and by the disappearance of seizures following its surgical removal) than elsewhere. The physio-pathological meaning and the diagnostic value of these findings, and particularly of the peculiar stability or autonomy of the electrical epileptic activity of the most epileptogenic cerebral zone, is discussed.

The long-term effect of vagus nerve stimulation on quality of life in patients with pharmacoresistant focal epilepsy: The PuLsE (Open Prospective Randomized Long-term Effectiveness) trial

To evaluate whether vagus nerve stimulation (VNS) as adjunct to best medical practice (VNS + BMP) is superior to BMP alone in improving long‐term health‐related quality of life (HRQoL).

Callosotomy for severe epilepsies with generalized seizures: Outcome and prognostic factors

SummaryThe purpose of the present study was to verify the effect of callosotomy on generalized seizures, to check the effect on other seizure types and to search for possible prognostic factors.Twenty patients with a minimum follow-up of one year (mean 3.5 years) were available for our analysis. In six of them the callosotomy was performed in two stages (total: 26 surgical procedures). Age ranged from 14 to 40 years (mean 23 years). Different aetiologies were known in 15 patients. Duration of epilepsy ranged from 6 to 23 years (mean 15 years). The frequency of seizures ranged between 19 and 750 per month.The most significant effect of surgery was the complete suppression of the generalized seizures associated with falling in 9/19 and their reduction of more than 80% in 7/19 patients (total “good results”: 16/19). The generalized tonic-clonic seizures were less affected. The surgical effect on the partial seizures was very variable, the partial simple seizures being the most affected. A positive statistical association with the outcome of the generalized seizures with fall was found for a presurgical seizure frequency below 90 per month, a prevalent bilateral EEG epileptic activity and, to a less extent, the absence of cerebral structural lesions. The role of age, aetiology, duration of the disease, single or more seizure types, mental impairment and extent of callosotomy remains uncertain. Disconnection syndrome does not appear if the splenium is spared.The present findings confirm that the main indication for callosotomy is the occurrence of generalized seizures with fall. Surgery can be initially limited to the anterior 2/3 of the corpus callosum; further posterior section of the corpus, excluding the splenium, should be regarded as a second step, when necessary.

Different neuronal contribution to N20 somatosensory evoked potential and to CO2 laser evoked potentials: an intracerebral recording study.

OBJECTIVE To investigate the possible contribution of the primary somatosensory area (SI) to pain sensation. METHODS Depth recordings of CO2 laser evoked potentials (LEPs) and somatosensory evoked potentials (SEPs) were performed in an epileptic patient with a stereotactically implanted electrode (Talairach coordinates y=-23, z=40) that passed about 10 mm below the hand representation in her left SI area, as assessed by the source of the N20 SEP component. RESULTS The intracerebral electrode was able to record the N20 SEP component after non-painful electrical stimulation of her right median nerve. The N20 potential showed a phase reversal in the bipolar montage (at about 31 mm from the midline), which confirms that the electrode was located near its generator in area 3b. In contrast, no reliable response was recorded from the SI electrode after painful CO2 laser stimulation of the right hand. An N2-P2 response was evoked at the vertex electrode (Cz), thus demonstrating the effectiveness of the delivered CO2 laser stimuli. CONCLUSIONS Since the N20 SEP component originates from the anterior bank of the post-central gyrus (area 3b), our result suggests that this part of SI does not participate in LEP generation. In fact, the previously published LEP sources in the SI area estimated from scalp recordings are about 10-17 mm posterior of the electrode in our patient, suggesting that they are more likely located in area 1, 2 or posterior parietal cortex.

A comment on impaired peri-nidal cerebrovascular reserve in seizure patients with brain arteriovenous malformations

Sir, We read with great interest the article by Fierstra et al . (2011), recently published in your journal. This article reopens the debate regarding the unsolved problem of the origin of seizures in patients with brain arteriovenous malformations, the pathophysiology of which remains poorly understood. Although angiographic studies have shown some associations between different brain arteriovenous malformation angioarchitectural features and epilepsy, they have not elucidated the underlying pathophysiology (Turjman et al ., 1995 a , b ; Hoh et al ., 2002). Similarly, MRI studies have often shown gliosis and chronic ischaemic changes in peri-nidal tissue as well as in remote regions, but any significant correlation has been demonstrated between extent of gliosis and clinical symptoms (Essig et al ., 2000). The main pathophysiological hypotheses to date include: (i) focal cerebral ischaemia attributable to a steal phenomenon as described by Spetzler et al . (1992) and Taylor et al . (2002); (ii) gliosis, demyelination and haemosiderin deposits in peri-nidal tissue; and (iii) secondary epileptogenesis in ipsilateral temporomesial cortex (Yeh et al ., 1990), or at a distant site, attributable to a ‘kindling’ phenomenon, in which epileptic discharges are enhanced by excitatory synaptic connections from brain arteriovenous malformations (Hoh et al ., 2002). Bleeding was deemed responsible for triggering seizure activity by Turjman et al . (1995 a ) and Stein and Wolpert (1980); in contrast, in the study by Hoh et al . (2002), intracranial haemorrhage was not a statistically significant factor associated with epilepsy. Fierstra et al . (2011) studied in vivo the possible haemodynamic effect …

Correlation between provoked ictal SPECT and depth recordings in adult drug-resistant epilepsy patients.

Summary:  Purpose: To correlate ictal hyperperfusion single‐photon emission computed tomography (SPECT) area during provoked seizures to the epileptogenic zone (EZ), as defined by depth recordings in adult drug‐resistant patients.

The human supplementary motor area-proper does not receive direct somatosensory inputs from the periphery: data from stereotactic depth somatosensory evoked potential recordings

Somatosensory evoked potentials to median nerve (MN) stimulation were recorded by stereotactically implanted electrodes in the frontal lobe of two epileptic patients in order to evaluate whether short-latency cortical responses could be generated in the supplementary motor area (SMA)-proper. In both patients two potentials, called P20 and N30 responses, showed a decreasing amplitude from the most superficial to the deepest contacts and were recorded in the white as well in the grey matter of the frontal lobe. Furthermore, no phase reversal of both P20 and N30 potentials was identifiable along electrode trajectories. Our results suggest that short-latency somatosensory evoked responses recorded in the SMA-proper contralateral to MN stimulation are volume conducted from remote cortical generators.

Factors of surgical outcome in tumoural epilepsy.

Summary Objectives. The purposes of the study were the assessment of the role of surgery in the suppression of epilepsy due to low-grade primitive cerebral tumours and the search for factors relevant to the surgical outcome. Patients and Methods. Forty-eight patients with epilepsy due to low-grade supratentorial cerebral tumours were considered. They presented drug-resistant daily to monthly seizures since for least one year (mean 7 yrs). Twenty-four patients underwent a combined tumour and epileptogenic zone resection (“epilepsy surgery”) and 24 tumour resection alone (“lesionectomy”). The surgical outcome was evaluated two years after surgery. Several variables related to the characteristics of the epilepsy, the tumour and surgery, were considered for a possible association with the outcome. Statistical analyses were performed. Results. Seizure freedom, including aura, was obtained in 35 patients (72.9%). Mild permanent complications occurred in 6 cases. Seizure suppression was significantly associated with complete tumour resection (post-surgical CT or MRI) and relatively low presurgical seizure frequency; it was also related, though not significantly, to small tumour size and histological grade I. The surgical outcome was only slightly better following “epilepsy surgery” than “lesionectomy”. However: i) the extent of tumour resection was not relevant regarding the “epilepsy surgery” outcome, while significantly influencing the outcome after “lesionectomy”; ii) the presurgical frequency of seizures and, to a less extent, the tumour size, had a higher influence on the outcome after “lesionectomy”. Conclusion. Long-lasting and drug-resistant epilepsy due to cerebral tumours can be suppressed surgically in the majority of cases. The extent of tumour resection and the frequency of the seizures are the most relevant prognostic factors. Both “epilepsy surgery” and “lesionectomy” can provide good results. However, the two approaches should not be regarded as interchangeable: a choice of the approach based on the characteristics of seizures and of the tumour appears relevant to improve the surgical prognosis.

Effect of Vagal Nerve Stimulation on Systemic Inflammation and Cardiac Autonomic Function in Patients with Refractory Epilepsy

Objective: Recent data suggest that vagus nerve stimulation (VNS) can inhibit cytokine release by inflammatory cells. Accordingly, an association between impaired cardiac parasympathetic function, as assessed by heart rate variability (HRV), and increased markers of inflammation has recently been reported. In this study we assessed the effect of direct left VNS on inflammatory markers and HRV in patients with refractory epilepsy. Methods: A 24-hour electrocardiogram Holter recording was performed both at baseline and after 3 months of left VNS in 8 patients (age 32 ± 24 years, 2 men) who underwent implantation of a VNS device because of refractory epilepsy. Tumor necrosis factor-α, interleukin-6 and C-reactive protein serum levels were measured, as markers of inflammation, at the same times. Results: No significant changes were found after 3 months of left VNS, compared to baseline, both for HRV variables and inflammatory markers. Also, no consistent correlation could be demonstrated between HRV parameters and inflammatory markers in these patients. Conclusions: Our data in epileptic patients without cardiovascular disease failed to show a significant effect of left VNS on cardiac autonomic function and on systemic inflammation at short-term follow-up.