IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 56, NO. 6, JUNE 2009
1587
Identification of the Dynamic Relationship Between Intrapartum Uterine Pressure and Fetal Heart Rate for Normal and Hypoxic Fetuses Philip A. Warrick∗ , Member, IEEE, Emily F. Hamilton, Doina Precup, and Robert E. Kearney, Fellow, IEEE
Abstract—Labor and delivery are routinely monitored electronically with sensors that measure and record maternal uterine pressure (UP) and fetal heart rate (FHR), a procedure referred to as cardiotocography (CTG). Delay or failure to recognize abnormal patterns in these recordings can result in a failure to prevent fetal injury. We address the challenging problem of interpreting intrapartum CTG in a novel way by modeling the dynamic relationship between UP (as an input) and FHR (as an output). We use a nonparametric approach to estimate the dynamics in terms of an impulse response function (IRF). We apply singular value decomposition to suppress noise, IRF delay, and memory estimation to identify the temporal extent of the response and surrogate testing to assess model significance. We construct models for a database of CTG recordings labeled by outcome, and compare the models during the last 3 h of labor as well as across outcome classes. The results demonstrate that the UP–FHR dynamics can be successfully modeled as an input–output system. Models for pathological cases had stronger, more delayed, and more predictable responses than those for normal cases. In addition, the models evolved in time, reflecting a clinically plausible evolution of the fetal state due to the stress of labor. Index Terms—Biosignal interpretation and diagnostic systems, biosignal modeling, linear and nonlinear dynamical models, signal and image processing.
I. INTRODUCTION HE LIFELONG disability that can result from oxygen deprivation during childbirth is rare but devastating for families, clinicians, and the health-care system. The hallmark indications that significant fetal cerebral hypoxia has occurred during labor are metabolic acidosis and neurological signs such
T
Manuscript received June 27, 2008; revised November 27, 2008 and January 19, 2009. First published February 20, 2009; current version published June 10, 2009. This work was supported by the LMS Medical Systems, Inc., and by the Natural Sciences and Engineering Research Council of Canada (NSERC). Astericks indicates corresponding author. ∗ P. A. Warrick was with LMS Medical Systems, Inc., Montreal, QC H4A 3S5, Canada. He is now with the Department of Biomedical Engineering, McGill University, Montreal, QC H3A 2B4, Canada (e-mail: philip.
[email protected]). E. F. Hamilton is with the Department of Obstetrics and Gynecology, McGill University, Montreal, QC H3A 1A1, Canada, and also with LMS Medical Systems, Inc., Montreal, QC H4A 3S5, Canada (e-mail: emily.hamilton@ lmsmedical.com). D. Precup is with the School of Computer Science, McGill University, Montreal, QC H3A 2A7, Canada (e-mail:
[email protected]). R. E. Kearney is with the Department of Biomedical Engineering, McGill University, Montreal, QC H3A 2B4, Canada (e-mail:
[email protected]). Digital Object Identifier 10.1109/TBME.2009.2014878
as altered levels of consciousness or seizures. Between 1 and 7 in 1000 fetuses experience oxygen deprivation during labor that is severe enough to cause fetal death or brain injury [1]–[3]; the range of this estimate reflects considerable regional variation and some clinical debate on the definition of brain injury. Unfortunately, noninvasive methods to directly measure the fetal acid–base status and cerebral oxygenation do not exist. Consequently, clinicians must rely upon indirect measures of oxygen delivery and neurological function. A standard approach is cardiotocography (CTG), which measures maternal uterine pressure (UP) and fetal heart rate (FHR); these signals are the cumulative result of many concomitant physiologies. Visual pattern recognition and inference are the basis of clinical interpretation. However, these are inconsistently applied [4]. Furthermore, classical patterns have low specificity. Because significant hypoxia is rare, false alarms are common, leading physicians to disregard truly abnormal signals. Indeed, approximately 50% of birth-related brain injuries are deemed preventable, with incorrect CTG interpretation leading the list of causes [3], [5]–[7]. The social costs of such errors are massive: intrapartum care generates the most frequent malpractice claims and the greatest liability costs of all medical specialties [8]. Thus, there is great motivation to find better methods to discriminate between healthy and hypoxic conditions. Clinicians’ interpretation of intrapartum CTG signals relies on the temporary decreases in FHR (FHR decelerations) in response to uterine contractions. FHR decelerations are mainly due to two contraction-induced events: 1) umbilical cord compression and 2) a decrease in oxygen delivery through an impaired utero-placental unit. There is general consensus that deceleration depth, frequency, and timing with respect to contractions are indicators of both the insult and the ability of the fetus to withstand it. Fig. 1 shows an example CTG demonstrating this response, where an FHR deceleration follows shortly after the onset of each of the four successive uterine contractions. Hypothesis-driven modeling based on this understanding of physiology and clinical interpretation has focused on contraction and deceleration detection [9]–[14]. In this paper, we propose a new approach that focuses on the dynamic relationship between UP (as an input) and FHR (as an output). Although the FHR is subject to numerous influences (i.e., it is the result of a multiple-input system), UP is the only input that is accessible by external monitoring and routinely used in clinical practice; indeed, clinicians already interpret certain UP–FHR relationships as indications of pathology. We use a
0018-9294/$25.00 © 2009 IEEE Authorized licensed use limited to: McGill University. Downloaded on June 18, 2009 at 13:09 from IEEE Xplore. Restrictions apply.
1588
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 56, NO. 6, JUNE 2009
Fig. 2. Data processing for estimation of UP–FHR dynamics. Preprocessing cleans and segments the UP and FHR into 20-min epochs of input U and ˆ delay d, and memory M . The output f . Nonparametric SI estimates the IRF h, significance filter validates the resulting models. Fig. 1. CTG stimulus response from four contraction–deceleration pairs over 10 min. (a) UP signal with contraction onsets (“C”) indicated. (b) FHR signal with deceleration onsets (“D”) indicated.
system identification (SI) approach to estimate system dynamics in terms of an impulse response function (IRF). This model represents very low-frequency (VLF) FHR energy (i.e.,
