c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
journal homepage: www.intl.elsevierhealth.com/journals/cmpb
Time–frequency energy distribution of phrenic nerve discharges during aspiration reflex, cough and quiet inspiration Juliana Knociková ∗ ˇ ˇ Department of Physics, Faculty of Electrical Engineering, University of Zilina, Vel’ky´ diel, Univerzitná 1, 010 26 Zilina, Slovak Republic
a r t i c l e
i n f o
a b s t r a c t
Article history:
Aspiration reflex (AspR) represents a specific inspiratory motor behavior expressed by short,
Received 11 January 2010
powerful inspiratory activity without subsequent active expiration and characterized by
Received in revised form
the ability to interrupt strong tonic inspiratory activity, as well as hypoxic apnea and sev-
24 August 2010
eral other functional disorders. Multiresolution analysis-based determination of spectral
Accepted 29 October 2010
features arising during AspR has not yet been satisfactorily investigated.
Keywords:
pared during the AspR, inspiratory phase of tracheobronchial cough and quiet inspiration.
Anesthesia
Data obtained from 16 adult cats anesthetized with chloralose or pentobarbital were ana-
Aspiration reflex
lyzed using a wavelet transformation, a sensitive method suitable for processing of the
Cat
non-stationary respiratory output signal.
The time–frequency energy distribution of phrenic nerve electrical activity was com-
Inspiration Wavelet transformation
Phrenic nerve energy was accumulated within lower frequency bands in AspR bursts. In AspR, higher frequencies contributed less to the total power, when compared to cough inspiration. Moreover, AspR indicated a stable time–frequency energy distribution, regardless of which of the two types of anesthesia were used. Chloralose anesthesia induced a decrease of parameters in cough and quiet inspiration related to the quantity of energy. The results indicate a specific method of information processing during generation of AspR, underlying its powerful ability to influence various severe functional disorders with potential implications for model experiments and clinical practice. © 2010 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
The quiet respiratory rhythm is generated by a plastic neuronal network located in the rostral ventrolateral medulla [1–3] in adult mammals and affects a variety of respiratory neuronal structures. Solitary rapid and strong inspiratory effort without subsequent active expiration, known as the sniff- and gasp-like aspiration reflex (AspR), can be evoked by mechanical or electrical stimulation of the nasopharynx in cats and other mammals [4,5]. Nasopharyngeal stimulation evokes a
∗
strong pre-motor burst of electric activity in the bulbar inspiratory neurons, followed by similar motor bursts in the phrenic nerves. This short-lasting but very intense inspiratory motor activity is characterized by rapid and strong activation of the inspiratory muscles, resulting in a kind of spasmodic inspiration. Its main functional role, supported by mucociliary transport, is to remove irritants from the upper airways to the hypopharynx through aspiration and their further elimination by cough or swallowing [4,6]. Inspiratory pre-motoneurons are located within the intermedial part of the ventral respiratory group (VRG) and in the
Corresponding author. Tel.: +421 41 513 2301; fax: +421 41 513 1516. E-mail address:
[email protected] 0169-2607/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cmpb.2010.10.014
82
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
dorsal respiratory group (DRG) of the medulla. It has been well documented that the discharge pattern of many medullary respiratory neurons activated during quiet breathing, is also markedly altered during coughing. Thus, the respiratory pattern generating network is involved both in production of quiet breathing and coughing [7]. Subgroups of inspiratory neurons in the nucleus tractus solitarius (NTS), are believed to be the primary site of integration related to this sniff-like reflex [8]. Other investigations imply that a long loop of the medullarypontine-mesencephalic control circuit could be involved in AspR coordination, where the neurons of the lateral medullary tegmental field and the caudal mesencephalic neurons seem to be crucial. As a result of the many similarities in character and intensity derived from the analysis of airway occlusion pressure and airflow in anesthetized cats, a common effector mechanism was suggested for AspR and gasping [9], since common brainstem areas resulting in generation of these reflexes were not directly proved. The inspiratory effort in AspR was traditionally analyzed from airflow records [9,10]. Several results indicated comparable features of motor manifestation during AspR and gasping, differing markedly from the normal breathing pattern. The analysis of phrenic nerve activity via the Short Time Fourier Transform indicated similar power spectra during the AspR and gasping, differing from that found in eupnea [11]. This specification was based on the detection of characteristic spectral signs of high frequency oscillations (HFOs) which were often described as signatures of respiratory network activity [12], being specific for various respiratory efferent systems [13]. However, due to the changing degrees of stationarity in recordings of inspiratory motor outputs, use of methods involving time–frequency localization of spectral features seems to be useful [14]. Moreover, there is a lack of information regarding comparison of spectral components between AspR and other defensive airway reflexes. Thus, this comparison is necessary because the nasopharyngeal stimulation evokes high-frequency impulses in the afferent fibers of the glossopharyngeal nerve, reaching nearly 400 Hz [15] and provocation of such a powerful AspR can interrupt various severe functional disorders [4]. The sniff- and gasp-like aspiration reflex (AspR), elicited by nasopharyngeal stimulation, manifests through marked respiratory, cardiovascular, neuromuscular and bronchomotor changes [4]. AspR can be provoked in any phase of the respiratory cycle and in very deep stages of general anesthesia of different types and during hypothermia, contrary to other airway defense reflexes, such as coughing and sneezing. The powerful AspR is able to interrupt strong tonic inspiratory activity in the phrenic nerve, i.e. apneusis, as well as hypoxic apnea and several other functional disorders in model experiments in animals, similar to “autoresuscitation by gasping” in infants in danger of imminent death from sudden infant death syndrome (SIDS). Various degrees of arousal reaction and resetting of central mechanisms for several vital functions can be elicited as a result of nasopharyngeal stimulation [9,10]. Therefore, AspR provides a unique model for interruption of various functional disorders, at least in animal models. Generating mechanisms and dynamics of time–frequency varying components arising in respira-
tory motor outputs during AspR are, however, still not fully understood. In this study, we analyzed the time–frequency distribution of the whole energy stored within phrenic bursts emerging during AspR, the inspiratory phase of the tracheobronchial (TB) cough and quiet inspiration. A method of wavelet analysis suitable for processing of non-stationary respiratory neural activity was used. The aims of the study were: 1. To investigate the time–frequency energy distribution of phrenic nerve activity during AspR using wavelet transformation Because of unique inspiratory motor behavior, specific time–frequency energy distribution of phrenic nerve activity is expected in AspR. Method of wavelet transformation, suitable for analysis of non-stationary respiratory outputs, helps to clarify the controversy presented in literature concerning spectral features of aspiration reflex. 2. To compare the properties of phrenic nerve pattern during AspR with the inspiratory phase of tracheobronchial cough and quiet inspiration Strong AspR without subsequent active expiration probably represents basic initial burst of inspiratory activity. Creation of tracheobronchial cough pattern requires activation of additional circuits that are responsible for preparation of strong expulsive effort. Therefore, wider spectral range of phrenic activity is expected during inspiratory phase of tracheobronchial cough and quiet inspiration under comparison with aspiration reflex. 3. To assess the effect of two different types of anesthesia on the character of phrenic nerve activity during the examined respiratory behaviors According to the ability of AspR to be elicitated in any phase of the respiratory cycle, as well as in very deep stages of general anesthesia, to interrupt hypoxic apnea and other functional disorders, it would be hypothesized that aspiration reflex presents a stable respiratory pattern, and its energetic distribution is not markedly sensitive to narcosis, contrary to cough or quiet inspiration.
2.
Materials and methods
2.1.
Data recording and pre-processing
The experiments were performed on 16 adult cats (3.0 ± 0.52) kg of both genders. Eight cats underwent pentobarbital anesthesia (PTBT; Pentobarbital, Spofa) with an initial intraperitoneal (i.p.) dose (35–40) mg kg−1 . Supplementary doses of anesthetic were given when required. Eight cats were anesthetized with alpha-chloralose (CH; Merck, (40–50) mg kg−1 ) given i.p. Tracheal airflow, respiratory rate (RR), end-tidal CO2 concentration (ETCO2 ) and blood pressure (BP), esophageal pressure (EP) were continuously monitored. Body temperature
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
was maintained at 38 ± 0.5 ◦ C. Both, AspR and the cough reflex were elicited mechanically by touching the nasopharyngeal or the tracheobronchial (TB) mucosa respectively, repeatedly 3–5 times in 1-s intervals. An elastic nylon fibre (0.4 mm) or a soft plastic catheter was introduced through a pharyngostomy (performed approximately 1 cm above the laryngeal structures) and tracheostomy (a window was done between the upper tracheal rings), respectively, in order to elicit both the reflexes. To produce tracheobronchial coughing, it was inserted through the cannula in the tracheal wall up to the level of tracheal bifurcation and gently moved in a rostrocaudal direction. Stimulation of the lower airways evoked repeated coughs. Series of AspR were evoked by gentle repetitive stimulation by thin nylon fiber via pharyngostomy, contacting the nasopharyngeal mucosa. The right phrenic nerve was dissected from the surrounding tissues and its sheaths were removed. These procedures have been described previously in details [11,16]. The electrical activity of the phrenic nerve was analyzed in 51 AspRs, 43 inspiratory phases of TB cough and 42 quiet inspirations. The animal care and use conformed to the guidelines accepted by the European Community, and the particular laws and regulations of the Slovak Republic. Only the single bursts of phrenic nerve activity were considered. Cough reflex is expressed by aligned inspiratory–expiratory bursts with deep inspirations followed by a forceful expulsive expirations. In this study, cough behavior was characterized by large bursts of phrenic activity with a deep negative wave of air flow (or EP) followed by a forceful expulsive expirations with strong positive peak of EP. Contrary, aspiration reflex presents strong inspiratory activity without subsequent active expiration (Fig. 1). The phrenic nerve was placed on a bipolar silver electrode and immersed in a pool of parafin oil. The electrode was connected to a low-noise amplifier Iso-DAM8. The signals were amplified, band-pass filtered (10–1000) Hz and digitalized. Minimal sampling frequency of 2 kHz was used. After recording, software filtration was performed off-line for the frequency range from 30 Hz to 1000 Hz. The 50 Hz frequency was cut off through a stop-band filter.
2.2.
Data analysis
Phrenic neural activities were examined and compared during inspiratory motor patterns (Fig. 1A and B) of AspR, TB cough and eupnoea. The multiresolution analysis resulting in a signal interpretation both in the time and frequency domains, was realized through a continuous wavelet transformation. It may be implemented using the following formula:
∞
W(s, ) =
x(t) −∞
1 s, (t) dt = √ s
∞
x(t) −∞
∗
t − s
dt
(1)
where is the transforming function called the mother wavelet, is the wavelet translation, and s is the wavelet scale. Thus, the presented equation allows the calculation of wavelet coefficients. After the wavelet transformation was performed, the particular phrenic bursts were exposed using the wavelet scalogram (Fig. 1C). In the scalogram, the wavelet coefficients are depicted as a function of time and the wavelet
83
scale. There is a dependence between the wavelet scale and the frequency. Lower wavelet scales are expressed as high frequencies, and vice versa. Pseudo-frequencies expressed in Hz may be induced from the wavelet scales: Fa ≡
Fc aTs
(2)
where a is a scale, TS is the sampling period, and Fc is the center frequency of the wavelet function calculated by approximation. As an effort to quantify an energy stored in the phrenic bursts during different motor behaviors, integration of the area under the wavelet coefficients curve was performed in a vertical and horizontal direction (Fig. 2). The integrated values were indicated as a function of time (vertical integration) and frequency (horizontal integration). In time domain, we detected the duration (TIME) of particular electrical nerve activity, the power maximum (MAXT ) and time of its occurence (TIMEMAX ), standardized to duration of the activity. In frequency domain, we calculated the total power (PTOT ), its maximal value (MAXF ) and frequency of its occurance (FREQ). The total power was divided in six frequency bands. The first band represents the frequencies above 500 Hz, the second band (300–500) Hz, third band (100–300) Hz, fourth (80–100) Hz, fifth (50–80) Hz, and the sixth band contains frequencies below 50 Hz. We detected the energies stored in each frequency band (FB1–FB6), and a measure of their contributions to the total power (in percentage; PFB1–PFB6). The analysis was performed with a self-developed computer program using MATLAB programming environment.
2.3.
Statistics
Unpaired t-test, Mann–Whitney test, one-way ANOVA and Kruskal–Wallis test (nonparametric ANOVA) were used for the statistical data evaluation. Differences were considered significant for p < 0.05. In aspiration reflex, cough and quiet inspiration, we tested effect of general anesthesia. The unpaired t-test was used to verify the hypothesis that two population means (under pentobarbital or chloralose anesthesia) are equal. The individual values were not paired or matched with one another. In the case of non Gaussian distribution, a non-parametric alternative, Mann–Whitney test, was performed. For comparison of time–frequency features of aspiration reflex, cough and quiet inspiration, one-way analysis of variance (ANOVA) was performed to test whether analyzed three groups are identical by comparison of the sample means. This aspect was evaluated under chloralose, as well as under pentobarbital anesthesia. The Kruskal–Wallis test, a non-parametric appropriate alternative, does not assume normality, and instead of comparing sample means, it compares sample means of ranks.
3.
Results
We ascertained differences in the time–frequency energy distribution of the phrenic nerve activity between AspR, the inspiratory phase of TB cough and quiet inspiration. Data were
84
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
Fig. 1 – Different expressions of the respiratory output activity. Part (A) illustrates an airflow record of the aspiration reflex preceded and followed by a quiet breath. The corresponding phrenic nerve activity is indicated in part (B), below on the left side. Wavelet scalogram of the phrenic nerve activity during AspR is drawn on the right side (part C). The energy of the phrenic nerve burst is expressed as a function of time and the wavelet scale (related to the frequency). Part (D) illustrates basic relationships between different inspiratory patterns in PTBT anesthesia. Strong time-domain energy is a typical feature of AspR through relatively short duration. Duration and parameter MAXT are standardized to quiet inspiration representing fundamental values. MAXT – maximum of time-domain power (result of vertical integration), a.u. – arbitrary units. Wavelet scale and wavelet coefficients are dimensionless values. * p < 0.05, ** p < 0.001.
taken from the experiments performed both in the alphachloralose and pentobarbital anesthesia and then compared.
3.1.
Comparison of AspR and quiet inspiration
The AspR was characterized by typically shorter duration under both types of anesthesia (p < 0.0001). Maximum of power in scale domain (MAXF ) reached extremely high values in AspR (p = 0.0002 for CH, and p = 0.0004 for PTBT anesthesia) and has occurred in the lower frequencies (FREQ; p = 0.0145 for CH, and p = 0.0192 for PTBT). Maximum of power occurred later (TIMEMAX ) within duration of the quiet inspiration (p = 0.0236), compared to phrenic nerve burst duration in the AspR (expressed in percentage). This shifting was detected only in PTBT anesthesia. AspR was characterized by extremely high level of total power (p = 0.0004 for CH, and p = 0.0029 for PTBT). Comparison between AspR and quiet inspiration showed increased
AspR activity for frequencies under 100 Hz (from p = 0.0002 to p = 0.04) under both types of anesthesia (Figs. 2 and 3). Frequencies over 100 Hz contributed to the total power (PFB1–PFB3) more during quiet inspiration in both types of anesthesia (from p < 0.0001 to p = 0.0329). The sixth frequency band contributed to the total power more (PFB6) in AspR (p < 0.0001 for CH, and p = 0.0208 for PTBT). There was not noticed any difference related to the fifth frequency band (50–800) Hz in regard to its contribution to the total power.
3.2. cough
Comparison of AspR and inspiratory phase of TB
Phrenic nerve activity lasted much shorter time in AspR (p < 0.0001) under both types of anesthesia (Fig. 1D). Maximum of power in frequency domain (MAXF ) decreased in AspR under chloralose anesthesia (p = 0.0002). In this anesthesia, the mentioned maximum has occurred at lower frequency
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
85
Fig. 2 – Phrenic nerve activity. The neurogram is expressed as time-expanded wave form, wavelet scalogram and horizontal and vertical scalogram integration (from top to bottom) in quiet inspiration (A), aspiration reflex (B) and the inspiratory phase of TB cough (C) in cats under PTBT anesthesia. The most pronounced cumulation of the energy within lower frequency bands (higher scales) is typical for AspR. The TB cough indicates higher relative power in the higher frequency bands, compared to AspR. a.u. – arbitrary units. Wavelet scale is a dimensionless value.
86
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
Fig. 3 – Wavelet scalograms of the phrenic nerve electrical activity in aspiration reflex (A), the inspiratory phase of TB cough (B) and quiet inspiration (C) in cats under chloralose anesthesia. During AspR, the energy is cumulated within lower frequency bands (higher wavelet scales) of the phrenic nerve activity, compared to the inspiratory phase of coughing and eupnea. TB cough is characterised by long and more pronounced “ramp-like” preparatory inspiration, the maximum of power was found at the end of the phrenic nerve activity. Wavelet scale is a dimensionless value.
during the inspiratory phase of cough (p = 0.0002). In time domain, there was a stronger maximal intensity (MAXT ) of AspR (p < 0.0001) in PTBT anesthesia (Fig. 1D). Moreover, under this anesthesia, maximum has occurred earlier (TIMEMAX ) in AspR (expressed in percentage; p = 0.0006). Under comparison of AspR and inspiratory phase of TB cough, there was no significant difference in the total power (PTOT ), however, lower absolute energy for frequencies over 200 Hz (Fig. 2) in AspR was noted under both types of anesthesia (from p < 0.0001 to p = 0.0383). In CH anesthesia, there was some increased energy detected within interval from 30 Hz to 50 Hz which was typical for AspR (p = 0.0477). Frequencies over 80 Hz contributed to the total power less in AspR (from p < 0.0001 to p = 0.0332). The sixth frequency band (FB6; (30–50) Hz) contributed to the total power more in CH anesthesia during the AspR (p < 0.0001). The fifth frequency band (FB5; (50–80) Hz) contributed to the total power in AspR as much as in the inspiratory phase of cough.
3.3. Comparison of inspiratory phase of TB cough and quiet inspiration Time duration of phrenic nerve activity was approximately the same both in the inspiratory phase of cough and quiet inspiration in cats under PTBT anesthesia (Fig. 1D). However, the quiet inspiration was shorter at CH anesthesia (TIME; p = 0.0346). Maximum of power in the time domain (MAXT ) reached
higher values (p = 0.0107 for CH anesthesia, and even more p < 0.0001 for PTBT anesthesia). It occurred later (TIMEMAX ) during the inspiratory phase of coughing (expressed in percentage; p = 0.0091 for CH and p = 0.0028 for PTBT anesthesia). The total power of the phrenic nerve activity and its maximal value were lower during eupneic inspiration, compared to inspiratory phase of cough (p = 0.0091 for CH, and p = 0.0007 for PTBT anesthesia). This decrease is visible in all frequency bands: FB1–FB6 (from p < 0.0001 to p = 0.0472). Higher frequency bands (PFB1–PFB3) contributed to the total power more in quiet inspiration in both CH and PTBT anesthesia (from p = 0.0009 to p = 0.0426). The inspiratory phase of the cough is characterized by the energy aggregation at the end of the burst duration (Fig. 3B).
3.4.
Effect of anesthesia
In this study, the effect of two anesthetic types was compared during three phrenic nerve respiratory patterns. At AspR the phrenic nerve activity indicated high total power and the most stabile time–frequency energy distribution. Anesthesia influenced the duration of the phrenic nerve activity only during the inspiratory phase of coughing. It lasted longer during CH, compared to PTBT anesthesia (p < 0.0001). Alphachloralose resulted in a decrease of all parameters quantifying the intensity. Thus, the maximum of power in frequency domain (MAXT ) decreased notably for AspR (p = 0.001). Similar
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
87
results were found in both the inspiratory phase of coughing (p = 0.0128) and quiet inspiration (p = 0.0107), especially for lower frequency bands. This maximum appeared in lower wavelet scale (higher frequency; FREQ) during the cough inspiration under CH anesthesia. Maximum of power in time domain occurred earlier (TIMEMAX ) in the cough inspiratory phase (p = 0.0041) and at quiet inspiration (p = 0.0234). Moreover, CH anesthesia induced a decrease of the absolute values of intensity for all frequency bands (FB1–FB6) in all analyzed groups (p = 0.0001 to p = 0.0387), except the frequency bands (30–50) Hz in all groups and (100–300) Hz in quiet inspiration. Anesthesia also influenced the relative powers (contribution of the frequency band to the total power; PFB1–PFB6) in lower frequency bands under 100 Hz (from p = 0.0005 to p = 0.0223) during quiet inspiration and the preparatory “inspiration” of the cough. At AspR the anesthesia type has not influenced the relative frequency powers of the phrenic nerve activity (Fig. 4), thus confirming its high resistance against used anesthetics.
4.
Discussion
The most important finding of this study is that the phrenic nerve activity during AspR shows a specific time–frequency energy distribution, compared to both the cough and eupneic inspirations. Moreover, the energy distribution in AspR remained stable, irrespective of which of the two anesthetics were used in the examined cats. These results indicated that there is shift in spectral power towards lower frequency oscillations typical for AspR and that this feature was not affected by the anesthetics used in the examined cats unlike tracheobronchial cough or quiet eupnea. Thus, the conclusions are in accordance with an earlier proposal that, within the brainstem, there may exist a specific neuronal network and mechanism(s) responsible for generation of AspR [4]. Similarly, functional magnetic resonance imaging studies in humans have indicated that voluntary sniffing, coughing and quiet breathing occupy different brain regions [17]. Analysis of electrical signals from the respiratory motor nerve outputs is usually performed as an investigation of high frequency oscillations (HFOs) which are accepted as a signature of the central pattern generator of breathing [1,12]. In cats, respiratory output discharges indicate HFOs usually within the range from 50 Hz to 100 Hz [13,18]. However, there is a great variability in different respiratory motor behaviors. In this study, the whole energy distribution within the phrenic nerve bursts was explored in the time and frequency domains, using a continuous wavelet transformation [19]. Unlike a traditional Fast Fourier Transformation, the wavelet method is considered to be more suitable for analysis of respiratory output patterns. The main reason is that Fourier transformation uses constant length of weighing function (window) which results in constant time–frequency resolution. The continuous wavelet transform was created as an alternative method of non-stationary signal processing. It decomposes the signal into modified versions of the mother wavelet. The mother wavelet is a kind of weighing function and has limited duration with zero as its mean value. In this study, “Morlet’s wavelet” was chosen as the mother wavelet [20].
Fig. 4 – Effect of anesthesia. Changes in the frequency rates (power rate of the pertinent frequency band to the total power of the burst expressed as percentage) are described as a result of anesthetics on the aspiration reflex (A), inspiratory phase of TB cough (B) and quiet inspiration (C) in cats. Red line: CH anesthesia, blue line: PTBT anesthesia. CH anesthesia resulted in a decrease of the total power, most visible in lower frequency bands, but not in higher frequency bands. The AspR represented the most stable model of the neural inspiratory activity. The obvious differences in the energy distribution caused by anesthesia were found in the inspiratory phase of coughing. * p < 0.05, ** p < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
The most important advantage of wavelet transformation is that the mother wavelet (an alternative of the window in Fast Fourier Transformation) changes its scale and time position completely. Therefore, it is continuously adapted to the analyzed signal properties, and an optimal time–frequency resolution may be reached during the analysis [21]. Thus, lower frequencies (slow oscillations) are more easily dis-
88
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
tinguished in the frequency domain because of the wider window. Conversely, for higher frequencies (fast oscillations), there is a better time resolution because the window is narrower. Determination of the frequency composition of neuronal activity is an important part of our analysis. A specific time–frequency energy distribution and marked energy accumulation within the lower frequency bands were detected in AspR. Damping of the contribution of the higher frequency bands to the total energy was a typical feature for the inspiratory phase of the cough in comparison with quiet inspiration. The described behavior has also been observed by other authors [22]. Nevertheless, the most important result coming from the multiresolution analysis is a specific time–frequency energy distribution and marked energy accumulation within the lower frequency bands detected in AspR. Marchenko et al. detected similar behavior in adult decerebrate rats during gasping development [23] and explained it as a possible reconfiguration of the respiratory network during “switching” from eupnea to gasping. Our study confirms this loss of high frequency oscillations as a typical feature for AspR in the same way as it was described during gasping. Crucial findings observed by Kocsis et al. during experiments on newborn rats and kittens [24] clarified that there are synchronized bursts corresponding to medium frequency oscillations in inspiratory nerve activity. HFO were highly attenuated in the phrenic nerve activity of newborn rats. Considering such behavior and our conclusions, the AspR could probably arise from the activity of the basic oscillator, a dominant structure in the generation of respiratory activity in newborn mammals, unlike in subsequent periods, resulting in network dominance presenting typical HFO activity in adult inspiratory nerve discharges. The basic rhythm generator, predominant during first developmental stages, does not require circuitry generated HFO typical of high O2 demands and probably becomes dominant in hypoxia/hypercapnia conditions [25,26]. It is well known that frequency components depend markedly on different experimental conditions. The type of anesthesia is one of the parameters greatly affecting the signal spectra. Results from this study indicate decreased values of all parameters quantifying the intensity of the signal due to alpha-chloralose anesthesia. This type of narcosis influenced the percentage contribution of frequency bands to the total power of the burst both in cough and quiet inspirations, being obvious especially in lower frequency bands. However, in AspR, the frequency distribution of the energy remained stable for all analyzed frequency bands, regardless of the type of anesthesia used. Signal transmission among neurons is realized through synapses where the signal undergoes a dynamic process of frequency de-/remodulation. It is generally accepted that the bulbar respiratory polysynaptic network is especially sensitive to general anesthesia [4]. In addition, this respiratory network shares at least its motoneuron pool with the basic circuitry producing the polysynaptic cough reflex [8]. It was shown that general anesthetics (such as pentobarbital or alphachloralose) directly hyperpolarized the respiratory neurons, changing their membrane channel properties. The responses of the central nervous system (CNS) to barbiturates or alpha-
chloralose were found to be dependent mostly on the chloride gradient, by stimulation of K+ leak channels [27]. The study by Korpáˇs and Tomori [28] confirmed similar findings that the polysynaptic (respiratory or cough) neuronal networks were more sensitive to anesthesia, while the “paucisynaptic” AspR networks have remained very stable. This may reflect, at least in part, a lower number of synapses involved in AspR. Moreover, the response of neuronal activities to the elicitation of AspR was examined during hyperoxia and compared to that created in severe hypoxia [29]. No difference was found between the neuronal responses to the reflex during the two mentioned conditions. The explanation could indicate a specific method of frequency modulation of sensory inputs at a central integration of AspR. The potency of neuromodulators to influence signal transmission was proved to be a frequencydependent process as an additional level of regulation [30]. Such an ability might explain the powerful potential of the AspR, based on very high-frequency oscillations (200–400 Hz) of afferent signals from the nasopharynx in cats activating the brainstem inspiratory center to evoke AspR and through very dense synaptic contacts and different modulators to normalize various dysfunctions of both hypofunctional character, e.g. hypoxic apnea and precoma [11], and hyperfunctional states, e.g. bronchoconstriction [6], apneusis and laryngospasm [9]. Phrenic nerve activity was found to be different in quiet breathing or gasping at an early stage of severe hypoxia. Using wavelet transformation, Akay [31] showed that hypoxia silences neural activity in the early phase of the phrenic neurogram, regardless of maturation in juvenile piglets. Qualitative changes of respiratory motor output activity during different pathological processes were also demonstrated through non-linear methods of analysis. Using non-linear methods of dynamic analysis, Dragomir et al. [32] indicated quantitative changes of respiratory signals during severe hypoxia in neonatal piglets. Using the method of approximate entropy, they demonstrated a reduced signal complexity during coughing and swallowing. These defense airway reflexes can effectively remove the fluid and mucus from the airways, and together with the Kratchmer apneic reflex and the laryngeal chemoreflex may prevent fluid aspiration, a process which is particularly important in the perinatal period [33]. Hence, one reason for sudden infant death syndrome may be a failure of these defense mechanisms. Dragomir et al. discuss a decreased signal complexity as a synchronous neural activity of a homogenous group of neurons. In the light of information theory, eupneic respiratory activity is considered to be a more random behavior. Several results indicate that hypoxia silences the neuronal activity and gasping may develop [10,11]. Gasping developing spontaneously during asphyxia or induced reflexly, can resuscitate some infants from imminent death by SIDS [34]. From recent investigations of breathing mechanisms, there has been an increased importance of selecting appropriate animal models for clinically related research. In this paper, all procedures were performed on adult cats because of typical pattern properties in AspR. The AspR has very similar neurophysiological characteristics to gasping [9], being able to evoke an arousal reaction and to reset the central control mechanisms of several vital functions. Fung et al. [29] showed that modulation of neuronal activities to gasping
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
induced by anoxia was identical to that induced by pharyngeal stimulation in either hyperoxia or severe hypoxia. In the aspiration reflex induced by pharyngeal stimulation [11], the neuronal responses are similar to that in gasping. This conclusion is compatible with the idea that the medullary gasping mechanism is recruited by pharyngeal stimulation. From the clinical and experimental point of view, it can reverse hypoxic apnea and even resuscitate cats from pre-comatose states [10]. Tomori et al. [11] described phrenic and hypoglossal nerve parameters of the power spectrum in gasping and aspiration reflex, differing markedly from that found in eupnea. They concluded that multiple sites and/or generators for ventilatory neurogenesis may exist in the brainstem, and these are involved in gasping and eupnea development. Moreover, mechanical stimulation of naso- and oropharynx activities provokes a potent reflex by which mechanisms generating eupneic ventilatory activity are suppressed and those for gasping are released [35]. The behavior described can support such an idea that medullary mechanisms which are critical for the neurogenesis of gasping are separate from those responsible for generating eupneic ventilation activity.
5.
[2]
[3]
[4]
[5]
[6]
[7]
Conclusion [8]
The electrical activity of the phrenic nerve in AspR has high maximal power, accentuated within the lower frequency bands and its parameters are not significantly influenced by anesthesia in contrast to the inspiratory phase of breathing or coughing. The inspiratory phase of TB cough showed higher frequency components. Chloralose anesthesia caused a decrease in parameters related to the quantity of energy of cough and quiet inspiration. These results support the idea of specific information processing at a stage of central neural integration of AspR. Described mathematical analyses indicates the importance of investigations in the fields of biomedical engineering, neurobiophysics and respiratory physiology. With further research, this interdisciplinary field may yield implications for model experiments, and medical practice.
[9]
[10]
[11]
[12]
[13]
Competing interests The author has no competing interests.
[14]
Acknowledgements [15]
The author would like to gratefully thank to Prof. Zoltán ˇ Tomori, M.D., D.Sc. (University of Pavol Jozef Safárik, Slovakia) for valuable suggestions and useful discussions. Helpful participation of co-workers (Department of Med. Biophysics, Comenius University, Slovakia) during experimental data recording is also acknowledged.
[16]
[17]
references [18] [1] J.L. Feldman, Neurophysiology of breathing in mammals, in: F.E. Bloom (Ed.), Handbook of Physiology, The Nervous
89
System, IV, Intrinsic Regulatory Systems of the Brain, American Physiological Society, Bethesda, MD, Washington, DC, 1986, pp. 463–524. R. Shannon, D.M. Baekey, K.F. Morris, B.G. Lindsey, Ventrolateral medullary respiratory network and a model of cough motor pattern generation, Journal of Applied Physiology 84 (June) (1998) 2020–2035. D.M. Baekey, K.F. Morris, S.C. Nuding, L.S. Segers, B.G. Lindsey, R. Shannon, Ventrolateral medullary respiratory network participation in the expiration reflex in the cat, Journal of Applied Physiology 96 (June) (2004) 2057– 2072. Z. Tomori, The sniff-like aspiration reflex, in: J. Korpáˇs, Z. Tomori (Eds.), Cough and Other Respiratory Reflexes, S. ˝ Karger, Basel/Munchen/Paris/London/New York/Sydney, 1979, pp. 224–250. ˇ cka, ˇ R. Bena Z. Tomori, The sniff-like aspiration reflex evoked by electrical stimulation of the nasopharynx, Respiration Physiology 102 (December) (1995) 163–174. Z. Tomori, J.G. Widdicombe, Muscular bronchomotor and cardiovascular reflexes elicited by the mechanical stimulation of the respiratory tract, The Journal of Physiology 200 (January) (1969) 25–49. R. Shannon, D.M. Baekey, K.F. Morris, Z. L.I, B.G. Lindsey, Functional connectivity among ventrolateral medullary respiratory neurons and responses during fictive cough in the cat, The Journal of Physiology 525 (May) (2000) 207–224. H.J. Batsel, A.J. Lines, Bulbar respiratory neurons participating in the sniff reflex in the cat, Experimental Neurology 39 (June) (1973) 469–481. ˇ M. Kurpaˇs, Comparison of inspiratory Z. Tomori, V. Donic, effort in sniff-like aspiration reflex gasping and normal breathing in cats, The European Respiratory Journal 6 (January) (1993) 53–59. ˇ cka, ˇ ˇ R. Tkácová, ˇ Z. Tomori, R. Bena V. Donic, Reversal of apnoea by aspiration reflex in anaesthetized cats, The European Respiratory Journal 4 (October) (1991) 1117–1125. ˇ V. Donicová, ˇ Z. Tomori, L.M. Fung, V. Donic, W.M. St. John, Power spectral analysis of respiratory responses to pharyngeal stimulation in cats: comparisons with eupnoea and gasping, The Journal of Physiology 485 (June) (1995) 551–559. M.I. Cohen, Neurogenesis of respiratory rhythm in the mammal, Physiological Reviews 59 (October) (1979) 1105–1173. M.I. Cohen, W.R. See, C.N. Christakos, A.L. Sica, High-frequency and medium-frequency components of different inspiratory nerves discharges and their modification by various inputs, Brain Research 417 (August) (1987) 148–152. M.H. O’Neal III, E.T. Spiegel, K.H. Con, I.C. Solomon, Time–frequency representation of inspiratory motor output in anesthetized C57BL/6 mice in vivo, Journal of Neurophysiology 93 (March) (2005) 1762–1775. B.S. Nail, G.M. Sterling, J.G. Widdicombe, Patterns of spontaneous and reflexly-induced activity in phrenic and intercostal motoneurons, Experimental Brain Research 15 (1972) 318–332. ˇ ˇ Distinct Z. Tomori, I. Poliacek, J. Knociková, J. Jakuˇs, V. Donic, generators of aspiration and expiration reflexes, their localizations, mechanisms and effects, in: Proceedings of European Respiratory Society Annual Congress, Berlin, 2008. K. Simonyan, Z.S. Saad, T.M. Loucks, C.J. Poletto, C.L. Ludlow, Functional neuroanatomy of human voluntary cough and sniff production, Neuroimage 37 (August) (2007) 401– 409. C.N. Christakos, M.I. Cohen, R. Barnhardt, C.F. Shaw, Fast rhythms in phrenic motoneuron and nerve discharges, Journal of Neurophysiology 66 (September) (1991) 674–687.
90
c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 0 2 ( 2 0 1 1 ) 81–90
[19] J. Sadowsky, The continuous wavelet transform: a tool for signal investigation and understanding, The Johns Hopkins APL Technical Digest 15 (1994) 306–318. [20] Y. Meyer, Wavelets: Algorithms and Applications, Society for Industrial and Applied Mathematics, Philadelphia PA, 1993. [21] A.H. Najmi, J. Sadowsky, The continuous wavelet transform and variable resolution time–frequency analysis, The Johns Hopkins APL Technical Digest 18 (1997) 134–140. [22] S. Yanaura, J. Kamei, K. Goto, T. Hosokawa, M. Misawa, T. Hukuhara, Analysis of efferent discharges of the phrenic nerve during cough reflex, Japanese Journal of Pharmacology 32 (October) (1982) 795–801. [23] V. Marchenko, R.F. Rogers, Selective loss of high-frequency oscillations in phrenic and hypoglossal activity in the decerebrate rat during gasping, American Journal of Physiology. Regulatory, Integrative and Comparative Physiology 291 (November) (2006) R1414–R1429. [24] B. Kocsis, K. Gyimesi-Pelczer, R.P. Vertes, Medium frequency oscillations dominate the inspiratory nerve discharge of anesthetized newborn rats, Brain Research 818 (February) (1999) 180–183. [25] K. Schmidt, G. Bohmer, Involvement of fast synaptic inhibition in the generation of high-frequency oscillations in central respiratory system, Brain Research 485 (April) (1989) 193–198. [26] V. Marchenko, R.F. Rogers, Time–frequency coherence analysis of phrenic and hypoglossal activity in the decerebrate rat during eupnea, hyperpnea, and gasping, American Journal of Physiology. Regulatory, Integrative and Comparative Physiology 291 (November) (2006) R1430–R1442. [27] L. Rasmussen, J. Moller, Central nervous system dysfunction after anesthesia in the geriatric patient, Anesthesiology Clinics of North America 18 (March) (2003) 59–70.
[28] J. Korpáˇs, Z. Tomori, Cough and Other Respiratory Reflexes, ˝ S. Karger, Basel/Munchen/Paris/London/New York/Sydney, 1979. [29] M.L. Fung, Z. Tomori, W.M. St. John, Medullary neuronal activities in gasping induced by pharyngeal stimulation and hypoxia, Respiration Physiology 100 (June) (1995) 195– 202. [30] H.T. Ito, E.M. Schuman, Frequency-dependent signal transmission and modulation by neuromodulators, Frontiers in Neuroscience 2 (December) (2008) 138– 144. [31] M. Akay, Hypoxia silences the neural activities in the early phase of the phrenic neurogram of eupnea in the piglet, Journal of Neuroengineering and Rehabilitation 2 (November) (2005) 32. [32] A. Dragomir, Y. Akay, A.K. Curran, M. Akay, Investigating the complexity of respiratory patterns during the laryngeal chemoreflex, Journal of Neuroengineering and Rehabilitation 5 (June) (2008) 17. [33] L. Van der Velde, A.K. Curran, J.J. Filano, L.A. Darnail, D. Bartlett Jr., J.C. Leiter, Prolongation of the laryngeal chemoreflex after inhibition of the rostral ventral medulla in piglets: a role in SIDS, Journal of Applied Physiology 94 (May) (2003) 1883–1895. [34] R. Sridhar, B.T. Thach, D.H. Kelly, J.A. Henslee, Characterization of successful and failed autoresuscitation in human infants including those dying of SIDS, Pediatric Pulmonology 36 (August) (2003) 113–122. [35] M.L. Fung, W.M. St. John, Z. Tomori, Reflex recruitment of medullary gasping mechanisms in eupnoea by pharyngeal stimulation in cats, The Journal of Physiology 475 (March) (1994) 519–529.