Performance of regional oxygen saturation monitoring by near-infrared spectroscopy (NIRS) in pediatric inter-hospital transports with special reference to air ambulance transports: a methodological study

Abstract

The aim of the present study was to evaluate the performance of regional oxygen saturation (rSO₂) monitoring with near infrared spectroscopy (NIRS) during pediatric inter-hospital transports and to optimize processing of the electronically stored data. Cerebral (rSO₂-C) and abdominal (rSO₂-A) NIRS sensors were used during transport in air ambulance and connecting ground ambulance. Data were electronically stored by the monitor during transport, extracted and analyzed off-line after the transport. After removal of all zero and floor effect values, the Savitzky-Golay algorithm of data smoothing was applied on the NIRS-signal. The second order of smoothing polynomial was used and the optimal number of neighboring points for the smoothing procedure was evaluated. NIRS-data from 38 pediatric patients was examined. Reliability, defined as measurements without values of 0 or 15%, was acceptable during transport (> 90% of all measurements). There were, however, individual patients with < 90% reliable measurements during transport, while no patient was found to have < 90% reliable measurements in hospital. Satisfactory noise reduction of the signal, without distortion of the underlying information, was achieved when 20-50 neighbors ("window-size") were used. The use of NIRS for measuring rSO₂ in clinical studies during pediatric transport in ground and air-ambulance is feasible but hampered by unreliable values and signal interference. By applying the Savitzky-Golay algorithm, the signal-to-noise ratio was improved and enabled better post-hoc signal evaluation.

Keywords: Air-ambulance; Inter-hospital transports; Monitoring; Near-infrared spectroscopy (NIRS); Pediatric; Savitzky–Golay algorithm.

Fig. 1

The observed percentage of rSO₂ values = 0%, i.e. no signal, for the cerebral sensor (rSO₂-C) and the abdominal sensor (rSO₂-A) during pre-transport observation in the PICU, transport in ground-ambulance and in air-ambulance. Each symbol represents data from one individual patient

Fig. 2

The percentage of reliable rSO₂-C and rSO₂-A values during pre-transport observation in the PICU, ground-ambulance transport and air-ambulance transport. The dotted line indicates 90% reliable rSO₂ values. Each symbol represents data from one individual patient

Fig. 3

Variability in signal expressed as median value of adjacent readings as a function of number of neighbors. Line filled circle: absolute delta NIRS rSO₂-C, dashed line open circle: absolute delta NIRS rSO₂-A

Fig. 4

The Savitzky–Golay filtering technique used to remove noise from the signal in one patient during 1 h of transportation for rSO₂-C. a Unprocessed data, b 5 neighbors, c 20 neighbors, and d 100 neighbors

Fig. 5

The Savitzky–Golay filtering technique used to remove noise from the signal to facilitate interpretation when rSO₂-C and rSO₂-A are affected by changes in altitude during flight. a Unprocessed data for rSO₂-C, b smoothing with 20 neighbors rSO₂-C. c Unprocessed data for rSO₂-A, and d smoothing with 20 neighbors rSO₂-A. The yellow-colored (shadowed) area symbolizes time in air ambulance