Brain Stimulation xxx (2015) 1e12
Contents lists available at ScienceDirect
Brain Stimulation journal homepage: www.brainstimjrnl.com
Original Research
Directional Recording of Subthalamic Spectral Power Densities in Parkinson’s Disease and the Effect of Steering Deep Brain Stimulation L.J. Bour a, *, M.A.J. Lourens a, R. Verhagen a, R.M.A. de Bie a, P. van den Munckhof b, P.R. Schuurman b, M.F. Contarino a, c, ** a
Department of Neurology and Clinical Neurophysiology, Academic Medical Center, University of Amsterdam, the Netherlands Department of Neurosurgery, Academic Medical Center, University of Amsterdam, the Netherlands c Department of Neurology, Haga Teaching Hospital, the Hague, the Netherlands b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 September 2014 Received in revised form 3 February 2015 Accepted 6 February 2015 Available online xxx
Background: A new 32-contacts deep brain stimulation (DBS) lead, capable of directionally steering stimulation, was tested intraoperatively. Objective: The aim of this pilot study was to perform recordings from the multidirectional contacts and to investigate the effect of directional current steering on the local field potentials (LFPs). Methods: In eight patients with Parkinson’s disease, after standard microelectrode recording and clinical testing, the new lead was temporarily implanted. The 32-channel LFP recordings were measured simultaneously at different depths and directions before and after directional stimulation. Results: The spatial distribution of LFPs power spectral densities across the contact array at baseline marked the borders of the subthalamic nucleus (STN) with a significant increase in beta power and with a mean accuracy of approximately 0.6 mm in four patients.The power in the 18.5e30 Hz frequency band varied across different directions in all patients. In the three cases that showed improvement of rigidity, this was higher when current was steered toward the direction with the highest LFP power in the beta band. Subthalamic LFPs in six patients showed a differential frequency-dependent suppression/ enhancement of the oscillatory activity in the 10e45 Hz frequency band after four different ‘steering’ modes as compared to ring mode, suggesting a higher specificity. Conclusions: Through a new 32-contact DBS lead it is possible to record simultaneous subthalamic LFPs at different depths and directions, providing confirmation of adequate lead placement and multidirectional spatial-temporal information potentially related to pathological subthalamic electrical activity and to the effect of stimulation. Although further research is needed, this may improve the efficiency of steering stimulation. Ó 2015 Elsevier Inc. All rights reserved.
Keywords: New 32-contact DBS lead Subthalamic nucleus Local field potentials Spectral analysis
Introduction The subthalamic nucleus (STN) is the predominantly chosen target for deep brain stimulation (DBS) in patients with Parkinson’s
Abbreviations: STN, subthalamic nucleus; GPi, Globus Pallidus interna; DBS, deep brain stimulation; LFP, local field potential; MER, micro electrode recording. * Corresponding author. Department of Neurology and Clinical Neurophysiology D2-108, Academic Medical Centre, P.O. Box 22660, 1100 DD Amsterdam, the Netherlands. Tel.: þ31 (0)20 5663515; fax: þ31 (0)20 5668197. ** Corresponding author. Department of Neurology and Clinical Neurophysiology D2-124.1, Academic Medical Centre, P.O. Box 22660, 1100 DD Amsterdam, The Netherlands. Tel.: þ31 (0)20 5668117; fax: þ31 (0)20 5668197. E-mail addresses: [email protected] (L.J. Bour), [email protected] (M. F. Contarino). http://dx.doi.org/10.1016/j.brs.2015.02.002 1935-861X/Ó 2015 Elsevier Inc. All rights reserved.
disease [1e8]. Abnormal oscillatory electrical activity of ensembles of neurons in the STN, observable in the amplitude modulation of the firing patterns of these neurons in a frequency range from approximately 10e35 Hz, is related to the severity of the parkinsonian state [9e12]. The amount of spectral power is related to the clinical status of the patient and the level of anti-parkinsonian medication [1,10,13e22]. Furthermore, STN DBS is capable of modulating this oscillatory activity [22e25]. Oscillatory amplitude modulation is not only observed in the spiking patterns of STN neurons, but also can be recorded from the local field potentials (LFP), which are the low frequency (1 TU) of the amplifiers makes them rather insensitive to impedance variations of the source (Porti7-32et, TMS International BV; http://www.tmsi.com/ products/item/porti). The 32-contact lead was then extracted and the Medtronic 3389 DBS lead was implanted for chronic stimulation. All patients underwent a post-operative CT scan. Signal analysis LFP signals were recorded against a common reference, i.e., the average potential of all active contact points with a sample frequency of 2048 Hz. Analog data were filtered from DC to 500 Hz. For initial inspection of the data, they were digitally band-pass filtered between 3 Hz and 80 Hz with a non-causal 4th order zero phase Butterworth filter (forward and reverse IIR filtering) and segments of interest were selected. To avoid circular convolution artefacts, at the start and the end of the recording an extra second of data was added (data padding), which afterward was removed. For subsequent detailed analysis of the spatiotemporal distribution of the LFPs inside the STN and the effect of DBS on the LFPs, segments of
4
L.J. Bour et al. / Brain Stimulation xxx (2015) 1e12
Figure 3. In four patients the average depth power of n 2-s epochs for two frequency bands, i.e., 3e10 Hz (blue dashed line) and 11e40 Hz (green line), is illustrated as a function of depth. The number of epochs was 77, 9, 7, and 50 for patients 1, 2, 3, and 4 respectively. For convenience the dorsal border of the STN, as detected by MER, is positioned at þ0.00 mm. The upper and lower black dashed lines indicate the dorsal and ventral border of the STN as determined by MER, respectively. The red error bars around each mean are the comparison intervals determined by the multi-comparisons Tukey Kramer method (significant level is a ¼ 0.001). For balanced one-way ANOVA, significant difference between two mean values means no overlap of their comparison interval. For all patients ANOVA showed a significant variability across depth (P < 0.005 for both frequency bands). Post-hoc contrast showed a significantly higher power in the 11e40 Hz measured inside than outside the STN for all patients (P < 0.0001). For the 3e10 Hz band power was significantly higher inside than outside only in patients 3 and 4 (P < 0.001). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
interest were selected from the original data and were band-pass filtered between 8 and 80 Hz with the same filter as described above. In the LFP signals sometimes artefacts were observed in the forms of short ‘jumps’ in the data, which are probably hardware related. To account for artefacts, an automatic artefact detection algorithm was applied [45] and visually inspected afterward. In short, after data padding and subtraction of the mean the segments were band-pass filtered and the amplitude envelop of the filtered segments were calculated using the Hilbert transformation. If the average of the z-score of the amplitude envelope of all channels exceeded a predefined value, it was supposed there was a common artefact present in all channels. The artefact detection procedure results in the start and end indices of epochs, which can be used to extract artefact-free data epochs from the original data. Only artefact-free data epochs longer than 2 s were selected for further analysis. Subsequently, the selected artefact-free data epochs were filtered in the same way as the segments in the artefact detection method, however, now after filtering the epoch means were subtracted, i.e., ignoring data padding. Furthermore, a 50 Hz discrete
Fourier transform filter was applied to remove powerline artefacts [45] and the added data parts again were removed. Finally data were down-sampled to 256 Hz. The Welch method with 50% overlap and a spectral resolution of 0.5 Hz was used for the estimation of the power spectral densities. If there was more data than one epoch (length of at least 2 s), power spectra of all epochs were averaged. Calculations of the power spectra were performed on monopolar derivations where each contact was referenced to the average of all active contacts (common reference). Based on the PSD measured in the patients of this study, which partially reflected also subdivisions used in the literature [14], the spectrum between 3 and 45 Hz was subdivided into four bands; 1) from 3 to 9.5 Hz 2) from 10 to 18 Hz 3) from 18.5 to 30 Hz 4) from 30.5 to 45 Hz. To evaluate the effect of stimulation, power spectral densities were calculated across the electrode array just prior to (at least15 s) and immediately 15 s after each stimulation mode. It is supposed that the spectral distribution immediately after high frequency stimulation reflects most of the effect on the LFP power. Suppression or enhancement of power after
L.J. Bour et al. / Brain Stimulation xxx (2015) 1e12
5
Figure 4. For each contact point, the power spectral density of the LFPs at baseline in the unfolded two dimensional array display in patient 6 is shown. Here, the power spectral density between 0 and 40 Hz inside the STN has been split up into A) a low frequency part (
Contents lists available at ScienceDirect
Brain Stimulation journal homepage: www.brainstimjrnl.com
Original Research
Directional Recording of Subthalamic Spectral Power Densities in Parkinson’s Disease and the Effect of Steering Deep Brain Stimulation L.J. Bour a, *, M.A.J. Lourens a, R. Verhagen a, R.M.A. de Bie a, P. van den Munckhof b, P.R. Schuurman b, M.F. Contarino a, c, ** a
Department of Neurology and Clinical Neurophysiology, Academic Medical Center, University of Amsterdam, the Netherlands Department of Neurosurgery, Academic Medical Center, University of Amsterdam, the Netherlands c Department of Neurology, Haga Teaching Hospital, the Hague, the Netherlands b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 September 2014 Received in revised form 3 February 2015 Accepted 6 February 2015 Available online xxx
Background: A new 32-contacts deep brain stimulation (DBS) lead, capable of directionally steering stimulation, was tested intraoperatively. Objective: The aim of this pilot study was to perform recordings from the multidirectional contacts and to investigate the effect of directional current steering on the local field potentials (LFPs). Methods: In eight patients with Parkinson’s disease, after standard microelectrode recording and clinical testing, the new lead was temporarily implanted. The 32-channel LFP recordings were measured simultaneously at different depths and directions before and after directional stimulation. Results: The spatial distribution of LFPs power spectral densities across the contact array at baseline marked the borders of the subthalamic nucleus (STN) with a significant increase in beta power and with a mean accuracy of approximately 0.6 mm in four patients.The power in the 18.5e30 Hz frequency band varied across different directions in all patients. In the three cases that showed improvement of rigidity, this was higher when current was steered toward the direction with the highest LFP power in the beta band. Subthalamic LFPs in six patients showed a differential frequency-dependent suppression/ enhancement of the oscillatory activity in the 10e45 Hz frequency band after four different ‘steering’ modes as compared to ring mode, suggesting a higher specificity. Conclusions: Through a new 32-contact DBS lead it is possible to record simultaneous subthalamic LFPs at different depths and directions, providing confirmation of adequate lead placement and multidirectional spatial-temporal information potentially related to pathological subthalamic electrical activity and to the effect of stimulation. Although further research is needed, this may improve the efficiency of steering stimulation. Ó 2015 Elsevier Inc. All rights reserved.
Keywords: New 32-contact DBS lead Subthalamic nucleus Local field potentials Spectral analysis
Introduction The subthalamic nucleus (STN) is the predominantly chosen target for deep brain stimulation (DBS) in patients with Parkinson’s
Abbreviations: STN, subthalamic nucleus; GPi, Globus Pallidus interna; DBS, deep brain stimulation; LFP, local field potential; MER, micro electrode recording. * Corresponding author. Department of Neurology and Clinical Neurophysiology D2-108, Academic Medical Centre, P.O. Box 22660, 1100 DD Amsterdam, the Netherlands. Tel.: þ31 (0)20 5663515; fax: þ31 (0)20 5668197. ** Corresponding author. Department of Neurology and Clinical Neurophysiology D2-124.1, Academic Medical Centre, P.O. Box 22660, 1100 DD Amsterdam, The Netherlands. Tel.: þ31 (0)20 5668117; fax: þ31 (0)20 5668197. E-mail addresses: [email protected] (L.J. Bour), [email protected] (M. F. Contarino). http://dx.doi.org/10.1016/j.brs.2015.02.002 1935-861X/Ó 2015 Elsevier Inc. All rights reserved.
disease [1e8]. Abnormal oscillatory electrical activity of ensembles of neurons in the STN, observable in the amplitude modulation of the firing patterns of these neurons in a frequency range from approximately 10e35 Hz, is related to the severity of the parkinsonian state [9e12]. The amount of spectral power is related to the clinical status of the patient and the level of anti-parkinsonian medication [1,10,13e22]. Furthermore, STN DBS is capable of modulating this oscillatory activity [22e25]. Oscillatory amplitude modulation is not only observed in the spiking patterns of STN neurons, but also can be recorded from the local field potentials (LFP), which are the low frequency (1 TU) of the amplifiers makes them rather insensitive to impedance variations of the source (Porti7-32et, TMS International BV; http://www.tmsi.com/ products/item/porti). The 32-contact lead was then extracted and the Medtronic 3389 DBS lead was implanted for chronic stimulation. All patients underwent a post-operative CT scan. Signal analysis LFP signals were recorded against a common reference, i.e., the average potential of all active contact points with a sample frequency of 2048 Hz. Analog data were filtered from DC to 500 Hz. For initial inspection of the data, they were digitally band-pass filtered between 3 Hz and 80 Hz with a non-causal 4th order zero phase Butterworth filter (forward and reverse IIR filtering) and segments of interest were selected. To avoid circular convolution artefacts, at the start and the end of the recording an extra second of data was added (data padding), which afterward was removed. For subsequent detailed analysis of the spatiotemporal distribution of the LFPs inside the STN and the effect of DBS on the LFPs, segments of
4
L.J. Bour et al. / Brain Stimulation xxx (2015) 1e12
Figure 3. In four patients the average depth power of n 2-s epochs for two frequency bands, i.e., 3e10 Hz (blue dashed line) and 11e40 Hz (green line), is illustrated as a function of depth. The number of epochs was 77, 9, 7, and 50 for patients 1, 2, 3, and 4 respectively. For convenience the dorsal border of the STN, as detected by MER, is positioned at þ0.00 mm. The upper and lower black dashed lines indicate the dorsal and ventral border of the STN as determined by MER, respectively. The red error bars around each mean are the comparison intervals determined by the multi-comparisons Tukey Kramer method (significant level is a ¼ 0.001). For balanced one-way ANOVA, significant difference between two mean values means no overlap of their comparison interval. For all patients ANOVA showed a significant variability across depth (P < 0.005 for both frequency bands). Post-hoc contrast showed a significantly higher power in the 11e40 Hz measured inside than outside the STN for all patients (P < 0.0001). For the 3e10 Hz band power was significantly higher inside than outside only in patients 3 and 4 (P < 0.001). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
interest were selected from the original data and were band-pass filtered between 8 and 80 Hz with the same filter as described above. In the LFP signals sometimes artefacts were observed in the forms of short ‘jumps’ in the data, which are probably hardware related. To account for artefacts, an automatic artefact detection algorithm was applied [45] and visually inspected afterward. In short, after data padding and subtraction of the mean the segments were band-pass filtered and the amplitude envelop of the filtered segments were calculated using the Hilbert transformation. If the average of the z-score of the amplitude envelope of all channels exceeded a predefined value, it was supposed there was a common artefact present in all channels. The artefact detection procedure results in the start and end indices of epochs, which can be used to extract artefact-free data epochs from the original data. Only artefact-free data epochs longer than 2 s were selected for further analysis. Subsequently, the selected artefact-free data epochs were filtered in the same way as the segments in the artefact detection method, however, now after filtering the epoch means were subtracted, i.e., ignoring data padding. Furthermore, a 50 Hz discrete
Fourier transform filter was applied to remove powerline artefacts [45] and the added data parts again were removed. Finally data were down-sampled to 256 Hz. The Welch method with 50% overlap and a spectral resolution of 0.5 Hz was used for the estimation of the power spectral densities. If there was more data than one epoch (length of at least 2 s), power spectra of all epochs were averaged. Calculations of the power spectra were performed on monopolar derivations where each contact was referenced to the average of all active contacts (common reference). Based on the PSD measured in the patients of this study, which partially reflected also subdivisions used in the literature [14], the spectrum between 3 and 45 Hz was subdivided into four bands; 1) from 3 to 9.5 Hz 2) from 10 to 18 Hz 3) from 18.5 to 30 Hz 4) from 30.5 to 45 Hz. To evaluate the effect of stimulation, power spectral densities were calculated across the electrode array just prior to (at least15 s) and immediately 15 s after each stimulation mode. It is supposed that the spectral distribution immediately after high frequency stimulation reflects most of the effect on the LFP power. Suppression or enhancement of power after
L.J. Bour et al. / Brain Stimulation xxx (2015) 1e12
5
Figure 4. For each contact point, the power spectral density of the LFPs at baseline in the unfolded two dimensional array display in patient 6 is shown. Here, the power spectral density between 0 and 40 Hz inside the STN has been split up into A) a low frequency part (
