If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Epilepsy Center, Department of Neurology, University Hospital, LMU Munich, Munich, GermanyFaculty of Engineering, University of Porto (FEUP), Porto, Portugal
Ictal vocalization occurs in nearly 40% of focal seizures.
•
It increases the localizing value of automatisms.
•
Vocalization during automatisms identifies temporal seizures with a sensitivity of 92% and a specificity of 70%.
•
An observer independent vocalization intensity analysis was established for the first time.
•
Intensity characteristics help to distinguish frontal from temporal lobe seizures.
Abstract
Purpose
To investigate the frequency, localizing significance, and intensity characteristics of ictal vocalization in different focal epilepsy syndromes.
Methods
Up to four consecutive focal seizures were evaluated in 277 patients with lesional focal epilepsy, excluding isolated auras and subclinical EEG seizure patterns. Vocalization was considered to be present if observed in at least one of the analyzed seizures and not being of speech quality. Intensity features of ictal vocalization were analyzed in a subsample of 17 patients with temporal and 19 with extratemporal epilepsy syndrome.
Results
Ictal vocalization was observed in 37% of the patients (102/277) with similar frequency amongst different focal epilepsy syndromes. Localizing significance was found for its co-occurrence with ictal automatisms, which identified patients with temporal seizure onset with a sensitivity of 92% and specificity of 70%. Quantitative analysis of vocalization intensity allowed to distinguish seizures of frontal from temporal lobe origin based on the intensity range (p = 0.0003), intensity variation (p < 0.0001), as well as the intensity increase rate at the beginning of the vocalization (p = 0.003), which were significantly higher in frontal lobe seizures. No significant difference was found for mean intensity and mean vocalization duration.
Conclusions
Although ictal vocalization is similarly common in different focal epilepsies, it shows localizing significance when taken into account the co-occurring seizure semiology. It especially increases the localizing value of automatisms, predicting a temporal seizure onset with a sensitivity of 92% and specificity of 70%. Quantitative parameters of the intensity dynamic objectively distinguished frontal lobe seizures, establishing an observer independent tool for semiological seizure evaluation.
]. Ictal sound and speech manifestations are also frequent symptoms of focal seizures, but audio analyses have rarely been performed and observer independent audio analysis tools are as yet missing.
Ictal sound or speech manifestations encompass a broad spectrum from intelligible speech to aphasia, dysphasia, dysprosody, or vocalization [
]. Thereby, pure ictal vocalization is defined as an ictal sound not being of speech quality and not accompanying apnea or generalized convulsive seizures [
]. Of note, previous vocalization studies were based on small cohorts of single epilepsy syndromes only and its association with concurrent semiology was not analyzed yet.
Our study thus aimed to (1) investigate the frequency and (2) localizing significance of ictal vocalization in different focal epilepsy syndromes. Further, (3) an observer independent approach was used to perform a quantitative audio analysis and to identify syndrome specific vocalization features.
2. Methods
This study complies with the institutional review board-approved ethical guidelines of the University of Munich and all patients gave written informed consent to the scientific use of their clinically acquired, anonymized data.
2.1 Participants and clinical procedures
Adult patients with drug resistant focal epilepsy syndrome (n = 316) were identified from the database of the Epilepsy Center of the University Hospital Munich. All patients underwent continuous long-term EEG-video-monitoring between 1995 and 2015 with closely spaced surface electrodes using the international 10-10 system of electrode placement. A subsample of 66 patients had additional electrocorticography (ECoG) or stereoencephalography (stereo-EEG) with electrical stimulation mapping in order to precisely distinguish eloquent brain regions. EEG was recorded using XLTEK Neuroworks software (Natus Medical Incorporated, San Carlos, CA, USA) with a sampling rate of 256 Hz or 1000 Hz and 12–16 bit A–D conversion. For intracranial EEG, the amplifier XLTEK EMU128FS was used. Beneath EEG and video recording, all patients had simultaneous electrocardiographic and pulsoxymetric recording. Syndrome classification and localization of the epileptogenic zone were specified in an interdisciplinary patient management conference based on the available EEG data, seizure semiology, neuropsychological test results, as well as functional and structural imaging data. Seizure freedom after focus resection would be the gold standard for seizure focus localization, but not all of the patients have undergone resection of the epileptogenic zone up to the time of our study analysis. As we aimed to investigate the localizing value of ictal vocalization, we thus included only patients with the most clear-cut epilepsy syndromes, i.e. patients with unifocal seizure onset and concordant MRI lesion (n = 294). Only patients who experienced at least one habitual seizure during EEG-video-monitoring were included in the study, those with only auras or subclinical seizures were excluded (n = 17). Relevant clinical data was extracted from the standardized data base of the EEG Video Monitoring Unit.
2.2 Qualitative analysis of ictal vocalization
Up to four consecutive focal seizures were evaluated in the final sample of 277 patients, excluding isolated auras and subclinical seizure patterns. According to previous studies [
], vocalization was considered to be present if observed in at least one of the analyzed seizures, not being of speech quality and not accompanying apnea, generalized convulsive or clonic seizures. Thus, only verbal vocalizations were included in the study. Stereotypic repetitions of intelligible words for example were not classified as ictal vocalizations. Using this definition of a pure ictal vocalization allowed not only a comparison of our results with previous reports, but was also shown to have a high interobserver agreement [
]. The video recordings were reviewed with respect to ictal vocalization and co-occurring seizure semiology by one of the authors blinded to patients’ clinical data. This particular author did not participate in the presurgical evaluation of any of the patients in the study. Unclear files were additionally reviewed by another experienced epileptologist. In the case of disagreement on the presence or abundance of vocalization, the seizure was excluded.
]. The algorithm detects the noise profile of the recording and then uses this profile to filter and smooth the audio signal. Audio files were filtered using the following settings: -30 dB noise reduction, 6.0 sensitivity parameter. The volume of noise reduction can be depicted in Fig. 1. To verify the output quality of the resulting audio signals, three people without previous knowledge of the audio contents were asked to validate the results.
Fig. 1Raw and filtered audio signal. The arrows indicate different segments of noise in the raw audio signal.
]. Analysis of vocalization intensity was performed for up to five vocalizations within the patient´s first seizure included in the study. Ictal audio signals were determined as separate vocalizations if they were at least 200 ms apart. Intensity analysis was not possible in seizures with bad audio quality or too much background noise caused by e.g. ictal testing or ictal movement. Audio analysis was focused on six different features: intensity minimum, maximum, mean and delta (i.e. the intensity range between minimum and maximum), as well as intensity variation (i.e. the standard deviation of the intensity values) and the velocity of the intensity increase rate at the beginning of the vocalization (see Fig. 2). To calculate the latter, the time needed to pass the baseline intensity + 15 dB was determined (intensity increase rate = 15 dB/time [dB/s]). The pitch could not be used for vocalization characterization, as the automatic pitch identification did not work well for many of the vocalizations.
Fig. 2Investigated intensity parameters. The black curve represents the intensity time course of one exemplary ictal vocalization.
Mean and standard deviation were calculated for quantitative parameters. Two-tailed t-test was performed for group comparison of continuous data and two-tailed Fisher´s exact test was applied on categorical data. Statistical significance was considered at p < 0.05. Multiple testing was accounted for by a Bonferroni-Holm review of the resulting p-values. Statistical testing was performed using Microsoft® Excel® for Mac 2011 (version 14.4.0).
3. Results
3.1 Demographics and clinical characteristics
In total, 277 patients with lesional, unifocal epilepsy were included in the study. Mean age at monitoring was 37.11 ± 12.94 years and both genders were equally represented. Most of the patients had a temporal (66.8%) or frontal lobe epilepsy syndrome (22.4%), and only few were diagnosed with pericentral or perieto-occipital epilepsy. Mesial temporal sclerosis was the predominant etiology of temporal lobe epilepsy, while focal cortical dysplasia was the most common etiology in frontal lobe epilepsy. A summary of the demographics and clinical data is given in Table 1.
Table 1Demographics and clinical data of patients with unifocal lesional epilepsy (n = 277).
Ictal vocalization was observed in 36.8% of the patients, i.e. in 25.1% of the analyzed seizures. Although seizures of frontal lobe onset revealed highest rates of ictal vocalization (p = 0.0006), the percentage of patients with ictal vocalization did not differ between the epilepsy syndromes. More precisely, ictal vocalization occurred in 43.6% of the patients with frontal, in 34.6% with temporal, in 31.6% with parieto-occipital and in 45.5% with paracentral epilepsy syndrome. There was neither a hemispheric (p = 0.068) nor a gender (p = 0.675) preponderance.
All 231 seizures with ictal vocalization were also analyzed with regard to the co-occurring seizure semiology (Table 2). Of note, focal onset automotor seizures with concurrent ictal vocalization were significantly more common in temporal than in extratemporal epilepsy syndromes (p < 0.0001) and identified temporal lobe epilepsy patients with a sensitivity of 91.5% and specificity of 70.2%. Frontal lobe seizures with ictal vocalization were most frequently characterized by hyperkinetic motor signs. This semiology was more common in frontal than in temporal (p < 0.0001) or pericentral epilepsy syndromes (p = 0.0058) but did not differ from the frequency found in patients with parieto-occipital seizure focus (p = 1.0). Similarly, frontal lobe epilepsy patients experienced significantly more often vocalizations during focal onset tonic seizures than patients with epileptic foci outside the frontal lobe (p < 0.0001). However, this semiology did not distinguish from seizures of pericentral onset (p = 1.0).
3.3 Quantitative analysis of ictal vocalization
Audio signal analysis was performed on 14 patients with frontal, 17 with temporal, and 5 patients with parieto-occipital epilepsy syndrome (Table 3). Patients had up to 6.29 ± 3.02 vocalizations per seizure, with highest numbers in frontal lobe and lowest in parieto-occipital epilepsy syndromes.
Table 3Quantitative analysis of ictal vocalization intensity.
Vocalizations occurring during frontal lobe seizures revealed a lower intensity minimum (p = 0.0034), a higher intensity maximum (p = 0.0332), and thus also a significantly greater intensity range (p = 0.0003) than vocalizations occurring during temporal lobe seizures. They were further characterized by a greater intensity variation (p < 0.0001), as well as a higher intensity increase rate at the beginning of the vocalizations (p = 0.0003) (Fig. 3). No significant difference was found for mean intensity and mean vocalization duration.
Fig. 3Quantitative vocalization analysis of different intensity parameters. The boxplot diagrams represent the minimum and maximum value, the median, as well as the first and third quartile of the respective parameter. Significant differences are indicated by an asterix. FLE: frontal lobe epilepsy (n = 14 patients); TLE: temporal lobe epilepsy (n = 17 patients); POE: parieto-occipital epilepsy (n = 5 patients).
Vocalizations during seizures of parieto-occipital origin showed significantly higher intensity variations compared to seizures of temporal onset (p < 0.0001). The other intensity parameters were not significantly different from vocalizations in temporal lobe epilepsy patients.
Setting a cut-off at >9.5 dB for the intensity variation allowed to identify seizures of extratemporal origin with a sensitivity of 92.86% and a specificity of 72.73%.
4. Discussion
4.1 Frequency and localizing significance of ictal vocalization
Ictal vocalization is a common feature of focal seizures, occurring in 37% of patients with different focal epilepsy syndromes. Although ictal vocalization is commonly thought to be associated with frontal lobe epilepsy [
], we did not find an association with any epilepsy syndrome in an unselected series of patients evaluated for epilepsy surgery. Localizing significance of ictal vocalization, however, was found when taking into account the concurrent seizure semiology. Its occurrence during automotor seizures predicted a temporal seizure onset with a sensitivity of 91% and a specificity of 70%. Automotor seizures, especially those with predominant oral automatisms like chewing, swallowing or smacking the lips, are also commonly associated with temporal lobe epilepsy, but they have a lower potential to distinguish temporal lobe seizures from frontal lobe seizures (23/58 vs. 9/61, i.e. sensitivity of 74%, specificity of 67%) [
]. The ictal occurrence of vocalization, thus, increases the localizing value of automotor seizures, adding a valuable semiological tool to identify temporal lobe epilepsy patients.
Further, quantitative analysis of vocalization intensity parameters allowed an observer independent differentiation between seizures of frontal and temporal origin. Ictal vocalizations during frontal lobe seizures were characterized by a significantly broader intensity range, greater intensity variation and a higher intensity increase rate than in seizures of temporal onset. These group differences were not due to an overrepresentation of female voices among patients with frontal lobe epilepsy, as both subsamples showed similar gender ratios. The lower intensity minimum, though, might be due to the fact that most frontal lobe seizures occur sleep related, i.e. with less background noise. However, this argument is contradictory to the higher intensity maximum found in this subsample. The latter, however, might attract more attention and thus may account for the commonly reported observation that ictal vocalization occurs more frequently in patients with frontal lobe epilepsy than in other epilepsy syndromes. Besides differentiating between seizures of frontal and temporal lobe onset, the audio analysis also allowed to identify patients with extratemporal epilepsy with a sensitivity of 93%, based on a vocalization intensity variation >9.5 dB.
4.2 Lateralizing significance of ictal vocalization
In our study, no hemispheric preponderance was found for ictal vocalization. In contrast, vocalization studies based on smaller cohorts of preselected epilepsy syndromes reported a non-significant association with the dominant hemisphere [
]. However, final assessment of the lateralizing significance of ictal vocalization requires a reliable determination of speech lateralization for the investigated cohort. This is only given in a study of three temporal lobe epilepsy patients investigated with stereo-EEG [
] and another subgroup of 57 temporal lobe epilepsy patients who had an additional Wada test suggesting a lateralization to the speech dominant hemisphere [
]. The lateralizing significance of our study, though, is limited, as speech lateralization using Wada test, functional MRI or cortical stimulation was only performed in selected patients if clinically required.
4.3 Pathophysiological considerations
Besides Broca´s area, vocalization is thought to originate from the supplementary motor area (SMA), and the medial orbitofrontal cortex including the anterior cingulate gyrus (ACG) [
]. This hypothesis is supported by the fact that the ACG is frequently involved in automotor and hyperkinetic seizures and that an activation of the SMA is associated with bilateral tonic posturing [
], i.e. the semiologies most frequently co-occurring with ictal vocalization. However, it has to be considered that especially early-onset epilepsy can modify the organization of the speech network [
]. Thus, additional cerebral regions might be involved in the generation of ictal vocalization in some of the analyzed patients.
The quantitative analysis of the vocalization intensity revealed syndrome specific differences in the intensity range, increase rate and variation. This suggests that different vocalization networks might be activated, depending on the seizure origin and evolution. So far, little is known about the source of sound differences as it was rarely investigated yet. In monkeys, vocalization characteristics changed after cerebellar hemispherectomy: postsurgically vocalization contained less components in different frequencies and its duration became more variable [
]. This suggests that a broad cerebro-cerebellar-network is involved in the generation and modulation of vocalization. It also substantiates the hypothesis that involvement or exclusion of a specific brain region can account for the sound differences described in our study. Further prospective investigations with stereo-EEG and cortical stimulation may add to a better understanding of ictal vocalization in humans.
4.4 Clinical implication of vocalization analysis
Although most of the software used was freeware and the quantitative analysis of ictal vocalization relatively easy to perform, its use in the clinical setting might be limited by background noise and other disturbances as well as the expenditure of time. Further, our approach was still based on synchronous video recording and preselection of the focal seizures with ictal vocalization. For clinical implementation, more data about ictal vocalization characteristics has to be gathered in order to once establish an automated sound detection and analysis system that can be used for aiding in the diagnosis procedures.
5. Conclusion
Ictal vocalization has localizing significance when taken into account the concurrent seizure semiology. For the first time, an observer independent analysis of vocalization intensity was performed, which allows to objectively differentiate between ictal vocalizations of frontal or temporal seizure origin. Qualitative and quantitative audio analysis thus adds a promising entity to semiological seizure analysis in patients considered for epilepsy surgery. Further, it might be a helpful tool for automated seizure detection, especially in combination with other quantified seizure measures like accelerometry or electrophysiological data.
Conflicts of interest
None of the authors has any conflict of interest to disclose.
Ethical publication statement
We confirm that we have read the Journal´s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Acknowledgment
The authors are indebted to the patients for their participation in this study. Further, we thank the reviewers for their valuable comments which helped to improve the manuscript.
References
Jin B.
Wu H.
Xu J.
Yan J.
Ding Y.
Wang Z.I.
et al.
Analyzing reliability of seizure diagnosis based on semiology.
30372-8&id=gr1.jpg){kind=link}
30372-8&id=gr2.jpg){kind=link}
30372-8&id=gr3.jpg){kind=link}