Prolonged immobility and airway patency are the targets of sedation for children undergoing magnetic resonance (MR) imaging in spontaneous breathing. Even though the guidelines of the American Paediatric Association and of the ASA stress the importance of monitoring the breathing pattern along with other standard parameters, no references could be found concerning monitoring of ribcage and abdominal motion in the MR scanner [1,2].
While visual inspection plays an important role in monitoring the respiratory activity of non-intubated patients, in the MR environment this is often made difficult by the presence of clothing and blankets, and it is associated with large errors in the estimation of respiratory rate and volume [3,4].
The contribution of PetCO2 is well established; however, in a non-negligible number of cases, the sedation level can be adequate for the imaging procedure but not sufficient for the patient to tolerate the discomfort caused by the capnometric nasal-oral probe. Furthermore, obstructions of the sampling line due to secretions and moisture can occur, especially in infants, and, although inserting the PetCO2 probe in a facial mask may circumvent these problems, it is often not possible when a head coil is to be used [5-8].
Unfortunately, most devices currently employed to monitor the breathing pattern are inherently incompatible with the electromagnetic fields of the MR scanner.
We propose that remote sensing of pressure in belts applied on the ribcage and abdomen, characterized by low cost and ease of implementation, may provide a way to monitor the breathing pattern. We refer to this technique as remote pressure sensor respiratory plethysmography (RPSRP). A similar technique has recently been shown by Banovcin and colleagues  to deliver results in agreement with the well-established respiratory impedance plethysmography (RIP).
One of the aims of RPSRP is to provide a backup service to measure the respiratory rate (RR) for use when PetCO2 monitoring fails, and to enable assessment of relative changes in tidal volume without the difficulties associated with visual inspection.
Upper airway obstruction is a common cause of hypoxaemia in sedated children. Increased collapsibility of the upper airway causes resistive loading, which leads to asynchrony of thoraco-abdominal motion [10-12]. Although the incidence of upper airway obstruction is increased by the use of intravenous (i.v.) medications, it is not negligible even when drugs considered relatively safe such as oral chloral hydrate are used .
Separate sensing of ribcage and abdominal compartments through RPSRP enables visualization of the Konno-Mead loop, which represents activity and coordination of respiratory muscles [13,14]. As shown in Figure 1, increasing asynchrony is seen as a transition towards a circle and then towards an ellipse whose major axis lies on a line with negative angular coefficient; the degree of asynchrony can be quantified by means of a phase shift (PS) index [10,12,13,15]. It has been shown that, although its predictive value may be limited, the PS index correlates with the severity of airway obstruction in anaesthetized children .
Additionally, RPSRP enables the monitoring of the proportion of inspiratory time over cycle time, commonly identified as TI/TTOT, or duty-cycle. Since it reflects the amount of stress being placed on respiratory muscles, it can be considered as a crude measure of airway obstruction; hence, its monitoring may have the potential to aid in the early identification of respiratory failure [16,17]. While this index has been previously employed to monitor anaesthetized children; however, to our knowledge, data on correlation with the severity of airway obstruction in children are currently not available .
We sought to explore the technical feasibility of breathing pattern monitoring in the MR scanner, and its viability for use in the clinical routine. We report the preliminary findings from testing of a prototype system on volunteers, and from its use along with standard SPO2 and PetCO2 monitoring on children sedated for brain MR imaging.
Materials and methods
Our current engineering prototype, covered by an Italian patent application, consists of 2 cm-wide elastic belts which house tubular air chambers, assembled in such a way as to be sensitive to both longitudinal and transverse motion. In accordance with standards for RIP, nipples and umbilicus have been chosen as anatomical landmarks for positioning; the elastic belts thus positioned are shown in Figure 2. Operators were trained to tighten the belts with a force of about 1 N, using tacks on belts and a thread aligned with the spine of patient to maximize positioning reproducibility.
The air chambers were connected, by means of 12 m of tubing passing through the shielded openings normally provided on the Faraday cage, to a unit (about 20 × 20 × 10 cm in size) outside the magnet room which housed piezo-resistive pressure sensors. After pre-amplification with custom circuitry, and filtering with fLO = 0.05 Hz and fHI = 10 Hz, the signal was sampled at 200 Sa s−1 with 12-bit resolution.
Since each air chamber and tubing is configured as a constant air mass compartment, Boyle's law applies, hence ΔV∝ΔP−1; for our prototype, normally ΔP≈2.9 kPa = 0.42 psi.
After linearization with a third-degree polynomial and smoothing with the Savitzky-Golay method, the signal was displayed as waveforms, and as Konno-Mead loop, with the ribcage component on the y-axis, and the abdomen component on the x-axis. After identification of maxima and minima on the sum of ribcage and abdomen signals, RR, TI/TTOT and peak-to-peak amplitude (A) were computed. The sigh rate (SR) was also obtained, defining a sigh as an isolated breath characterized by at least double signal amplitude. As shown in Figure 3, referring to the interval between a minimum on the ribcage signal and the closest minimum on the abdomen signal as ΔT1, and to the equivalent for maxima as ΔT2, the PS was computed as PS = (ΔT1 + ΔT2)/2TTOT. At the present exploratory stage, these parameters were not provided in real-time, but computed offline with a signal analysis software written in the MATLAB language (Mathworks Inc., Natick, USA).
Approval for testing on children and adult volunteers was given by the internal review board of our institution, and written informed consent was always obtained.
Correlation of RR with values given by a PetCO2 monitor (Millenia 4500, MRI Devices, Orlando, FI, USA) was evaluated on 27 children as described below, and the Pearson coefficient was computed.
In order to preliminarily evaluate the clinical viability of RPSRP monitoring, 27 consecutive patients undergoing MR imaging under sedation were included in the study; patient characteristics data are reported in Table 1. ECG, non-invasive blood pressure, SPO2 and PetCO2 were monitored, and clinical events recorded. Imaging sessions lasted between 20 and 90 min.
Fourteen patients received chloral hydrate orally, the dose ranging between 50 and 100 mg kg−1, up to 1.5 g. Among them, seven reached a sedation score of three on the Skeie scale within 30 min (chloral succeeded, CS group), and seven failed to respond adequately (chloral failed, CF group) and subsequently received, before imaging, i.v. supplementation of 10 mg boluses of sodium thiopental (STP, 1.4-4.1 mg kg−1) titrated to reach the desired sedation score . This sedation protocol is described in detail in . The remaining 13 patients (no chloral, NC group) were sedated with i.v. STP (3.1-4.9 mg kg−1) directly, due to availability of venous access. While in , a strict policy of not administering any medication in the magnet room was envisaged, in the period under consideration in this study, we performed intra-procedural supplementation as soon as motion artefacts appeared on imaging. In this framework, two patients from the CF group and two patients from the NC group received intra-procedural supplementation of i.v. STP (0.5-1.2 mg kg−1).
As reliable measurements of respiratory function are notoriously difficult to obtain in unsedated children, breathing pattern recording was started once the patient was sedated and positioned in the scanner . The operator positioning the belts was blinded to the type of sedation, and the anaesthesiologist managing the sedation was blinded to RPSRP readings and did not remain in the magnet room, while the operator post-processing RPSRP data was blinded to patient identity by using randomly generated patient ID numbers.
A one-way ANOVA was used to evaluate group differences in maximum and average PetCO2, minimum and average SPO2, average RR, TI/TTOT, PS and SR; the Tukey test was used for post hoc comparison. In order to test for confounding effects of age and weight, an ANCOVA was performed using group as factor and age and weight as covariates. Correlations were searched for among total dose of STP, maximum and average PetCO2, minimum and average SPO2, average RR, TI/TTOT, PS and SR.
At this stage, since testing measurements could not be conducted on children due to ethical reasons, correlation between the sum of the amplitude of the two RPSRP waveforms and actual volume was verified by carrying out constant volume measurements on eight healthy volunteers (mean age 25 ± 2.7 yr). Volunteers, lying in supine position, breathed through a high-precision spirometer in 18 runs, each one 30 s long, randomly at 200-1000 mL of fixed tidal volume. The corresponding waveform amplitude was recorded while an operator blinded to RPSRP readings ensured that the fixed volume was respected. The Pearson coefficient was computed for each session. In order to evaluate reproducibility of positioning and tensioning, five additional runs at 500 mL were performed for each volunteer, while an operator blinded to RPSRP readings removed and repositioned the belts between runs. The maximum difference in waveform amplitude was computed for each session.
RR readings from RPSRP on sedated children were found to correlate very strongly with those from PetCO2, with r = 0.999 and P < 0.001; the maximum difference observed was 0.8 bpm.
Among the eight recording sessions performed on adult volunteers, for the correlation between RPSRP waveform amplitude and tidal volume, the Pearson correlation coefficient r ranged between 0.92 and 0.98 (0.96 ± 0.02), P < 0.001; the corresponding slope was between 22.5 μL−1 and 37.7 μL−1 (29.2 ± 5.4 μL−1). This corresponds to a maximum deviation of about 22% from the average slope, and of about 16% in 75% of cases. For the five repeats at 500 mL, the difference in relative amplitude ranged between 20% and 27% (22 ± 2%).
All imaging sessions succeeded, and spontaneous breathing was always maintained; no patient needed intervention to resolve airway obstruction. Three patients (11%) did not tolerate the nasal capnometric probe, and in seven cases (26%) PetCO2 readings were discontinuous. No patient showed any sign of discomfort caused by RPSRP belts.
As shown in Figure 4, it was found that sighs and non-respiratory movements of the torso, undetectable by visual monitoring at a distance, are identifiable on RPSRP waveforms. Transitory non-respiratory movements of the torso, causing interruption in RPSRP readings, were detected in seven patients (those four that received supplementation plus three for whom no motion artefacts were detected); no other interruptions occurred.
Patient characteristics and values for average and maximum PetCO2, average and minimum SPO2, and average RR, TI/TTOT, PS and SR are reported in Table 1. Even though the groups differed significantly in age and weight, the ANCOVA did not reveal any confounding effect (P > 0.1 for all combinations).
Results of post hoc comparisons are given in Table 1. No significant inter-group differences were found in PetCO2, SPO2, RR and PS. However, TI/TTOT was higher in the NC group when compared to the CS group (P = 0.02), the CF group being characterized by intermediate values. Moreover, when compared to the CS group, SR was lower in the CF and NC groups (P = 0.04 and P = 0.03).
A positive correlation was found between total dose of STP and TI/TTOT, with r = 0.4 and P = 0.04; no other significant correlations were found.
Figure 5 provides an example of the degree of variability in thoraco-abdominal coordination, which is found in some neurological patients; the recording shown belonged to a patient in the CF group with a brainstem tumour that did not require intra-procedural supplementation, and that did not display any clinical sign of upper airway obstruction (i.e. no snoring, SPO2 stable at or above 95% and PetCO2 stable below 50 mmHg).
Figure 6a-d report the RPSRP monitoring trends for the four patients who received supplementation after motion artefacts appeared; no obvious pattern that could serve as a predictor of motion artefacts was identified. Transitory increases of TI/TTOT seen in this figure, often accompanied by changes in PS, could be related to increased resistive loading caused by sub-clinical obstruction of the upper airways.
As belts were tolerated by all patients, no interruption in readings occurred besides those caused by transitory non-respiratory movements, and respiratory rate measurements were in accordance with those given by the PetCO2 monitor. Using pressure sensors may be a viable approach to monitor the breathing pattern in the MR scanner, providing a backup source of respiratory rate data for use when the capnometric probe is not tolerated, and when it becomes obstructed. Nevertheless, respiratory rate readings from pressure sensors must be interpreted with great caution when explicit information about airway patency is not available.
Even though measurements conducted on adults at this proof-of-concept stage clearly cannot serve as validation for volume correlation on children, they do show that the amplitude of pressure sensor waveforms has the potential to exhibit a degree of correlation with relative changes in tidal volume, which is better when compared to visual inspection, reportedly associated with 30% error, without the difficulties caused by clothing and blankets obstructing sight . The random error due to re-positioning was found to be comparable with the effect of the large variability seen in the slope of the amplitude-volume fit line. Since calibration procedures used for quantitative respiratory plethysmography are unlikely to be applicable in the context of paediatric sedation, techniques such as RPSRP are not viable to obtain quantitative volume measurements.
In this sample of 27 patients, four received intra-procedural supplementation after motion artefacts appeared; when compared to the results reported in  this may seem a surprisingly high proportion. The cause is that, while in the period under consideration in this study, intra-procedural supplementation was performed as soon as artefacts appeared, in  a different policy was adopted, which was attempted to continue imaging without intervention, and all those cases in which movements were transitory and imaging could be completed were not counted. RPSRP appeared of dubious usefulness for the prediction of movements, due to the absence of potentially predictive markers in the recordings of the patients who needed intra-procedural supplementation, which was characterized by a considerable variability in baseline values and trends. We speculate that the sudden changes in loudness and pitch of scanner noise, which occur between imaging sequences, may have an important role as triggers of patient movement. Since the occurrence of subtle non-respiratory movements can be indicative of a change in the level of sedation, their detection might play a role in prompting intra-procedural supplementation, or shortening of the scanning protocol; however, while non-respiratory movements were detected in seven patients, imaging artefacts were visible in only four of them.
Since no clinically detectable obstruction occurred at this exploratory stage, it was not possible to demonstrate a correlation between clinical observation and respiratory timings. We hypothesize that transitory increases of TI/TTOT, often accompanied by changes in PS, could be indicative of increased resistive loading caused by sub-clinical obstruction of the upper airways; this is to be confirmed by further study [11,16,17]. Even though previous reports indicated correlation between PS and severity of upper airway obstruction, a large baseline variability was found in our sample; as exemplified in Figure 5, considerable variations in the PS can occur even during a clinically uneventful sedation. Individuality changes in neural drive due to the pharmacokinetics and neurological pathology are likely to be important factors [11,21]. One must also take into account that the correlation with severity of obstruction reported in previous studies in children, although statistically significant, was not judged adequate for use as a clinical tool in a previous study on a relatively small sample .
Notwithstanding the small sample size, significant inter-group differences in SR and TI/TTOT were found, while initial sedation scores were equal, and SPO2 and PetCO2 readings were comparable.
TI/TTOT was found to be significantly higher in the NC group, sedated with STP only, when compared to the CS group, which did not receive STP at all. Furthermore, a significant correlation between TI/TTOT and total dose of STP was found. These findings are concordant with evidence that barbiturates such as STP cause dose-dependent collapsibility of airways; this is also true for propofol [22,23].
The SR was lower when STP was administered (CF and NC groups): this is interpreted as indicative of decreased neural drive; since the role of sighing in restoring FRC is well established, its absence may be considered as an early warning of respiratory depression [24,25].
The fact that groups did not differ in average PS is not surprising, given the large variability observed in the absence of events and drifts in other parameters. The usefulness of the PS index in this context is therefore dubious. The baseline breathing pattern was not recorded, that is, the belts were positioned only after the patient was sedated. While availability of baseline data would have made it possible to assess to what extent the observed variability was caused by sedation, it is not clear how this could be reliably performed on unsedated children.
The finding of significant differences in TI/TTOT and SR and the correlation observed between TI/TTOT and dose of STP suggest that these indexes are sensitive to the clinical condition of the patient; however, one must consider that no individual interpretation can be inferred from these preliminary results.
Usage as a backup source of respiratory rate data and as a detector of non-respiratory movements can itself justify the integration of mechanical monitoring of the breathing pattern into pre-existing SPO2 and PetCO2-based protocols. Furthermore, the differences seen in TI/TTOT and SR indicate that they convey information relevant to the state of the sedated patient; this motivates further study on large samples into their potential usefulness as monitoring parameters: limited sample size prevents this proof-of-concept study from reaching any conclusion about sensitivity and specificity.
These results are not specific to our implementation of RPSRP, and can be generalized to existing technologies, such as air cushions normally used for respiratory-gated MR imaging, and other mechanical or optical detectors.
1. Committee on Drugs, American Academy of Pediatrics. Guidelines for monitoring and management of pediatric patients during and after sedation
for diagnostic and therapeutic procedures: addendum. Pediatrics
2. American Society of Anaesthesiologists. Position on Monitored Anaesthesia (approved October 1986); Guidelines for Patient Care in Anaesthesiology (approved October 1967, amended October 1985); Standards for Basic Intra-Operative Monitoring (approved October 1986, amended October 1990); Standards for Post-Anaesthesia Care (approved October 1988, amended October 1990). Park Ridge, IL: American Society of Anaesthesiologists.
3. Krieger B, Feinerman D, Zaron A, Bizousky F. Continuous non invasive monitoring of respiratory rate in critically ill patients. Chest
4. Semmes BJ, Tobin MJ, Snyder JV, Grenvik A. Subjective and objective measurement of tidal volume in critically ill patients. Chest
5. Tsui BC, Wagner A, Usher AG, Cave DA, Tang C. Combined propofol and remifentanil intravenous anaesthesia for poediatric patients undergoing magnetic resonance imaging
. Paediatr Anaesth
6. Agnes NG. Sevoflurane sedation
in infants - a fine line between sedation
and general anaesthesia. Paediatr Anaesth
7. Mason KP, Zgleszewski SE, Dearden JL et al.
Dexmedetomidine for pediatric sedation
for computed tomography imaging studies. Anesth Analg
8. Cortellazzi P, Lamperti M, Minati L, Falcone C, Pantaleoni C, Caldiroli D. Sedation
of neurologically impaired children undergoing MRI: a sequential approach. Paediatrc Anaesth
2007; Epub ahead of print.
9. Banovcin P, Seidenberg J, von der Hardt H. Pressure sensor plethysmography: a method for assessment of respiratory motion in children. Eur Respir J
10. Tobin MJ. Breathing pattern analysis. Intensive Care Med
11. Litman RS, Kottra JA, Gallagher PR, Ward DS. Diagnosis of anesthetic-induced upper airway obstruction in children using respiratory inductance plethysmography. J Clin Monit Comput
12. Tobin MJ, Perez W, Guenther SM, Lodato RF, Dantzker DR. Does rib cage-abdominal paradox signify respiratory muscle fatigue? J Appl Physiol
13. Konno K, Mead J. Measurement of the separate volume changes of rib cage and abdomen during breathing. J Appl Physiol
14. Agostoni E, Mognoni P. Deformation of the chest wall during breathing efforts. J Appl Physiol
15. Mador MJ. Respiratory muscle fatigue and breathing pattern. Chest
16. Tobin MJ. Respiratory monitoring in the intensive care unit. Am Rev Respir Dis
17. Bellemare F, Grassino A. Evaluation of human diaphragm fatigue. J Appl Physiol
18. Brown KA, Reich O, Bates JH. Ventilatory depression by halothane in infants and children. Can J Anaesth
19. Skeie B, Emhjellen S, Wickstrom E, Dodgson MS, Steen PA. Antagonism of flunitrazepam-induced sedative effects by flumazenil in patients after surgery under general anaesthesia. A double-blind placebo-controlled investigation of efficacy and safety. Acta Anaesthesiol Scand
20. Gaultier C, Fletcher ME, Beardsmore C, England S, Motoyama E. Respiratory function measurements in infants: measurement conditions. Working Group of the European Respiratory Society and the American Thoracic Society. Eur Respir J
21. Benchetrit G. Breathing pattern in humans: diversity and individuality. Respir Physiol
22. Strickland TL, Drummond GB. Comparison of pattern of breathing with other measures of induction of anaesthesia, using propofol, methohexital, and sevoflurane. Br J Anaesth
23. Eastwood PR, Platt PR, Shepherd K, Maddison K, Hillman DR. Collapsibility of the upper airway at different concentrations of propofol anesthesia. Anesthesiology
24. Goodman NW, Kestin IG. Sighs and their effect on the breathing of patients anaesthetized with infusions of propofol. Br J Anaesth
25. Patroniti N, Foti G, Cortinovis B et al.
Sigh improves gas exchange and lung volume in patients with acute respiratory distress syndrome undergoing pressure support ventilation. Anesthesiology