This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Premature labor and delivery is an important public health problem as it represents a major and often preventable cause of morbidity and mortality in newborns. Prematurity, often associated with permanent neurodevelopmental disability, occurs when a baby is born before the 37th week of gestation. As a consequence, the immature newborn may require long-term neonatal intensive care at a substantial emotional and financial cost to families and hospitals. According to the European Perinatal Health Report, preterm birth rates varied widely among European countries and ranged from 5.5 to 11.4%, meaning that about half a million babies are born prematurely in Europe every year (Euro-Peristat Project, 2004). The highest percentage of preterm births was registered in Austria (11.4%), followed by Germany (8.9%) and was lowest in Ireland (5.5%) and Lithuania (5.3%). The variation in very preterm births, before 32 weeks of gestation, was less pronounced, and rates for most countries fell within a range of 0.9 to 1.1%. Some of the variation between countries may be due to differences in the way that gestation is determined. According to a previous study, about 85% of infant mortality is accounted for by preterm labor (Maul et al., 2003).
One of the major challenges faced by obstetricians is a proper diagnosis of labor which could be especially useful in the prediction of labor for patients at high risk for premature delivery. To date there is no accurate and objective method available for MMG to predict the onset of labor or to distinguish between false and true labors (Arulkumaran et al., 1991; Garfield et al., 2001; Iams, 2003; Peaceman et al.,1997; Rabotti et al., 2010a). The objective assessment of uterine activity could also significantly contribute to the timely recognition of eventual complications. A high percentage of preterm births are the result of either (i) induced delivery or delivery by Caesarean section, also called C-section1, (ii) spontaneous preterm labor with intact membranes or (iii) preterm premature rupture of the membranes. Common reasons for these types of deliveries are the intrauterine growth restriction (IUGR) or hypertensive disorders of pregnancy, i.e., pre-eclampsia and eclampsia (Goldenberg et al., 2008). IUGR refers to the poor growth of a fetus while still in the mother's uterus. Pre-eclampsia is a medical condition characterized by pregnancy-induced hypertension in the mother accompanied by proteinuria (an excess of serum proteins in the urine). Eclampsia, usually occurring in a patient who already developed pre-eclampsia, is an acute life threatening complication of the pregnancy characterized by the appearance of tonic-clonic seizures.
At present, little is know about the pathogenesis of preterm labor. It is speculated that preterm labor might be the result of either an early idiopathic activation of the normal labor process or the result of pathological insults (Goldenberg et al., 2008; Leman et al., 1999; Maner et al., 2003). To reduce the incidence of preterm delivery, current obstetric interventions focus on inhibiting premature contractions by the administration of tocolytic agents, which temporarily delay the delivery. However, tocolytic therapy involving agents to control the contractibility of the myometrium, e.g. magnesium sulfate and others, is expensive and exposes the patient to unnecessary risks. Therefore, most physicians rely on cervical change before initiating this therapy. In a recent study the use of tocolytic agents is questioned, as their administration fails to demonstrate improvements in neonatal outcome (Kenyon and Peebles, 2011). Instead, the progesterone has been suggested as a future therapeutic candidate to reduce the risk of early birth (Fonseca et al., 2007; Kenyon and Peebles, 2011). However, there is a lack of studies to demonstrate a long-term reduction in neonatal morbidity and mortality from administration of progesterone.
In a review conducted by Garfield and colleagues studies showed that the uterus (myometrium) and cervix pass through a conditioning step in preparation for labor. This step consists of changes in the electrical properties that make muscles more excitable and responsive to produce forceful coordinated contractions (Garfield et al., 2001). At a certain point this process becomes irreversible and delivery cannot be delayed for more than a few days even with tocolytic agents (Garfield et al., 2001). Preterm labor management is of an even greater concern. Early treatment may result in more effective suppression of preterm labor. This often leads to unnecessary treatment and hospitalization of patients who are not in true preterm labor. Accurate diagnosis of preterm labor is only possible through cervical changes such as dilation or effacement. However, even noticeable dynamic cervical change may not be an accurate indicator of true labor, because a high percentage of women with established cervical change do not deliver pre-term when not treated with tocolytics (Cox et al., 1990; Maner et al., 2006). In addition, hospital admissions during false labor are associated with considerably high costs.
Therefore, the development of assessment tools to identify and characterize the course of labor, as well to support accurate prediction of the delivery, would be of great help to physicians.
In order to improve knowledge about the mechanism involved in labor, it is essential to understand the physiological mechanisms that regulate uterine contractions, identify parameters in order to track the stages of labor and relate the changes in such parameters to clinical outcomes.
The physiology of uterine contractions
Anatomy and biology of the uterus
The uterus is part of the female reproductive system, located inside the pelvis, dorsal to the urinary bladder and ventral to the rectum. The uterus has an important role in the development of the fetus, by providing a safe environment throughout the gestation and later, as term approaches, in the expelling of the fetus through intense contractions.
The anatomical structure of the human non-pregnant uterus is presented in Figure 1.1. The uterus has a length of approximately 8 cm. Two fallopian tubes enter near its top, dividing the uterus in two parts: the fundus (above the fallopian tubes) and the body (below the fallopian tubes). The narrower, lower end, of the body of the uterus is called the cervix and is made of fibrous connective tissue, with a much firmer consistency compared to the body of the uterus.
The uterine wall (see Figure 1.1) consists of a well-differentiated (i) innermost lining layer also called the endometrium, (ii) a thick muscular inner layer called the myometrium and (iii) an outer layer named the serosa or perimetrium (Chard, 1994).
The myometrium is composed of (i) the outer longitudinal muscle layer and (ii) the inner circular layer. The longitudinal layer consists of bundles of smooth muscle cells which are aligned with the long axis of the uterus. The muscle cells of the circular layer are arranged concentrically but more diffusely around the longitudinal axis of the uterus (Chard, 1994; Csapo, 1954). Early studies have shown that the longitudinal layer is continuous with the circular one and that during pregnancy a coordinated contraction of both layers occurs (Osa and Katase, 1975; Tomiyasu et al., 1988).
For the current work, the particular layer of interest is the myometrium which is responsible for the induction of uterine contractions (Garfield et al., 1998).
Organization of the smooth muscle cell
To understand uterine contractility, a review of the morphology and the electrophysiology of the uterus is necessary.
The myometrium, located between the endometrium and perimetrium, mainly consists of uterine smooth muscle cells (see Figure 1.2), also known as uterine myocytes, arranged in overlapping tissue-like bands (the exact arrangement is still a highly debated topic). In addition, this type of smooth muscle can maintain force for prolonged time periods with very little energy expenditure. In regard to smooth muscles, physiologists point out that, important differences exists between various smooth muscle tissues of the same species and between anatomically and functionally comparable smooth muscles of related species (Fischer, 1944). While striated muscles are organs with a comparable locomotive function and consist of only muscle tissue, smooth muscles are generally, only with few exceptions, elements contributing together with other tissues to the anatomy of the whole organ. The smooth muscle is composed of small fibers, usually 5 to 10Î¼m in diameter and approximately 20 to 500Î¼m in length. In most organs, smooth muscle cells are functionally connected by the so called gap junctions (functional syncytium, see Figure 1.3). The gap junction is a structure composed of two symmetrical portions of the plasma membrane from two different adjacent cells.
Three types of filaments have been identified in the uterine smooth muscle cells (i) a thick filament consisting of myosin molecules, (ii) a thin filament consisting of actin molecules and (iii) an intermediate filament which consists mainly of desmin molecules (see Figure 1.3). The actin and myosin filaments possess different striated arrangements compared to skeletal muscles and have a high number of actin filaments which are attached to dense portions of smooth muscle, known as dense bodies (Chard, 1994). The dense bodies are connected by intermediate filaments, which form an elastic cytoskeleton coupled to the extracellular connective tissue. In addition, the actin and myosin filaments are organized in so called sarcomeres (see Figure 1.3) which are not aligned inside the smooth muscle due to the irregular arrangement of the dense bodies. The amount of actin filaments which are surrounding a myosin filament is variable, but larger compared to those found in the skeletal muscle.
The basic structure of the smooth muscle actin filaments is similar to the structure of actin filaments found in the skeletal or heart muscle. However, the smooth muscle actin filament is considerably longer. Similarly, the smooth muscle myosin filaments are longer compared to their skeletal or heart muscle counterparts and display a rectangular cross-section (Klinke et al., 2009). The myosin molecules are linearly arranged and are characterized by an antiparallel assembly rather than a helical geometry (see Figure 1.3).
Contractile apparatus of the uterine myometrium
Cells typically exhibit a voltage potential difference with the inside of the cell being more negative than the outside due to the concentration of potassium (K+) ions within the cell. In other words, they maintain a voltage difference across the cell's plasma membrane, known as the membrane potential. An excitable patch of membrane has two levels of membrane potential (i) the resting potential, which is the value the membrane potential maintains as long as nothing perturbs the cell, and (ii) a higher value called the threshold potential.
While most cells remain at a constant potential and do not vary with time, two types of cells, smooth muscle and neuronal, are electrically active. Early cell experiments indicated that ionic current flow is voltage dependent (Anderson, 1969). The shortlasting event in which the electric membrane potential of a cell rapidly raises and falls is called an action potential (AP). For a better understanding of the uterine contractile activity, in the following section, a description of the mechanisms eliciting the AP of the individual nerve cell will be provided.
Figure 1.4 illustrates the ideal AP for typical cells (the actual recordings of APs can be distorted compared to the schematic view due to variations in the electrophysiological assessment techniques). In general, a cell's resting potential is approximately -70mV due to the K leakage currents and remains at that value until a stimulus is applied.
When a stimulus of sufficient magnitude (above the threshold of âˆ’55mV ) is applied, the membrane voltage depolarizes. Voltage-gated channels then open and cause quick inward currents of natrium (Na+) and calcium (Ca2+) ions. This rapid inward flow of positive ions is responsible for reversal of the voltage polarity and contributes to the rising phase of the AP. At approximately +40mV the Na voltage-gated channels close and voltage-gated K channels open causing the falling phase of the process or repolarization. Because of the large K currents leaving the cell, a transient negative shift known both as hyperpolarization or as a refractory period occurs. This mechanism prevents an action potential from travelling back towards the way it came. Once the voltage-gated K channels close, Na and K pumps continue to balance the ion flow until the resting potential is reestablished (Lodish et al., 2007). The same terminology can be used to describe the activation of the uterine cells. A more detailed description is provided in the following sections.
The contractile effort of the myocytes is periodic and arises from the propagation of the electrical activity through the muscle cells in the form of action potentials. The APs are a result of ionic current flow into and out of the cell membrane. The exchange of Na+, K+, and Ca2+ ions across the cell membrane constitute the ionic currents. Their unequal distribution across the cell membrane creates a potential difference. The ions can be pumped across the cell membrane, against the concentration gradient, by energy that is supplied by the breakdown of adenosine triphosphate (ATP) into adenosine diphosphate (ADP). ATP is a multifunctional nucleoside triphosphate, often referred
to as the molecular unit of currency of intracellular energy transfer (Knowles, 1980). Its main role is to transport energy for metabolism within cells. ATP is produced by photophosphorylation and cellular respiration and is used by enzymes and structural proteins in many cellular processes including muscle action (Knowles, 1980).
At rest, the uterine smooth muscle cell has the following ionic distribution (i) the concentration of Na+ and Ca2+ ions is higher outside the cell (compared to inside) and (ii) the concentration of K+ ions is higher inside the cell (Chard, 1994). This distribution of ions corresponds to the resting membrane potential. The myometrial cell's resting potential can range from -40 to -60mV due to the hormonal state (Parkington and Coleman, 2001; Sims et al., 1982). It has been shown that the uterine myometrial cells are electrically coupled by the so-called gap junctions which consist of proteins (Garfield et al., 1995; Maner et al., 2006). When grouped, these proteins create channels of low electrical resistance between cells that facilitate pathways for the efficient conductance of action potentials. There are a few of these channels throughout pregnancy, indicating a poor coupling and decreased electrical conductance. However, as women approach term, the gap junctions increase and form an electrical syncytium which is a prerequisite for proper contractions (Garfield et al., 1995).
The contraction of the myometrium is mainly governed by changes of the Ca2+ concentration between extra and intracellular space and by the release of Ca2+ from intracellular depots. Specifically, that involves a rise in the intracellular Ca2+ concentration from a resting level of about 10âˆ’7M to approximately 10âˆ’5M. The source of Ca2+ can be (i) intracellular, where Ca2+ ions are released from intracellular depots, (ii) extracellular, where Ca2+ ions flow into the cell following their electrochemical gradient in response to a change in membrane permeability or (iii) a combination of both (Chard, 1994). Conversely, a reduction of intracellular free Ca2+ ions terminates the contraction. The reduction of Ca2+ ions occurs either as a result of re-uptake into cellular depots or efflux into the extracellular space (Chard, 1994).
However, because the sarcoplasmatic reticulum of the smooth muscle cell is poorly developed, the source of Ca2+ causing the contraction is mainly extracellular (Sanborn, 2000). Consequently, diffusion of Ca2+ into the cell occurs when the concentration of Ca2+ in the extracellular fluid exceeds 10âˆ’3M. The average time required for the diffusion is between 200-300ms (approximately 50 times longer than the diffusion measured in skeletal muscle fibers) (Guyton and Hall, 2010). Compared to skeletal muscle, smooth muscle cells have many more voltage-gated Ca channels and much fewer voltage-gated Na channels. As a consequence, the generation and propagation of APs in smooth muscle is mainly regulated by Ca channels (in skeletal muscles this activity is mainly regulated by the Na channels) which open and close significantly slower compared to the Na channels. This slow acting mechanism accounts for the slow onset of contraction and relaxation of the smooth muscle tissue in response to the electrical stimulus.
The dynamic of Ca2+ concentration is controlled by voltage and receptor controlled channels. This implies that the contraction process of the myometrium can be influenced by changes in the electrical properties and/or the receptor environment. As action potentials propagate over the surface of a myometrial cell, the depolarization causes voltage-dependent Ca2+ channels to open. When this occurs, Ca2+ enters the muscle cell traveling down its electro-chemical gradient to activate the myofilaments and elicits a contraction by increasing the size and/or number of actual portals for Ca2+ entry (Aguilar et al., 2010). Thus, the increase in Ca2+ is considered to be the primary catalyst of the chemomechanical process of smooth muscle contraction. The repolarization is the result of both the K+ ion efflux and the inactivation of Na+ channels. Figure 1.4 shows the relationship between the membrane AP and the inward/outward ion flux in the cell. The inward current carried by Na+ and Ca2+ ions is responsible for the cell polarization and the outward current carried by K+ and Ca2+ ions induces the cell repolarization.
Previous studies have also established that the myometrial electrical activity governs myometrial mechanical contractions (Maner et al., 2006; Tezuka et al., 1995). This behavior was demonstrated in electromyogram (EMG) recordings of the human uterus during pregnancy (Devedeux et al., 1993). Contraction bursts that occur prior to the onset of labor are often perceived by the mothers as periods of perceived contractility.
Studies have shown that in various species the gap junctions are sparse throughout pregnancy but increase during delivery (Garfield and Hayashi, 1981; Garfield et al., 1977, 1978; Lodge and Sproat, 1981; Miller et al., 1989; Miyoshi et al., 1998). It was also observed that these gap junctions disappeared within 24 hours of delivery. Gap junction proteins are thought to align themselves and create low-resistant channels (of approximately 1nm) between the cytoplasm of adjacent cells (see Figure 1.3) to form a pathway for the passage of APs (Garfield et al., 1998, 1977; Miller et al., 1989). The increase in the gap junctions and their electrical transmissions provides better coupling between the cells, resulting in synchronization and coordination of the contractile events of the various myometrial regions in the uterus. The results provide clear evidence that the propagation of the electrical activity over the entire myometrium due to the increase in gap junction areas at term is related to successful progress of labor and delivery of the fetus.
The distribution of gap junctions is not necessarily homogeneous in tissue and the relation between the junction pattern and intercellular communication is poorly understood. The most studied gap junction subunit is the connexin 43 (Cx43), a 43-kD protein expressed in myocardium and myometrium as well as in other cells. The results suggest that transcription of the Cx43 is induced by activating the proteine kinase C (PKC) in human myometrial cells (Garfield et al., 1998). The PKC is involved in controlling the function of other proteins through phosphorylation and can be activated by an increased concentration of Ca2+ (more details are given in section 22.214.171.124).
Studies conducted by Garfield and colleagues have demonstrated that throughout the pregnancy gap junctions are present at a very low density providing an indicator of poor coupling and limited electrical conductance (Garfield et al., 1977). Conversely, contractile uterine activity during term or pre-term labor is characterized by the presence of a large amount of gap junctions between the myometrial cells (Garfield et al., 1977; Miller et al., 1989). In addition, a study carried out by the same group has shown that the presence of gap junctions is controlled by the regulation of progesterone and estrogen in the uterus. Specifically, progesterone down-regulates and estrogen upregulates the myometrial gap junction density (Miller et al., 1989).
As previously described, smooth muscle cell voltage-dependent Ca2+ channels open as depolarization occurs, allowing Ca2+ ions to enter the muscle. Once voltage-dependent Ca2+ channels open, a single action potential can initiate a twitch contraction (quick shortening of the muscle) as shown in Figure 1.3. A twitch contraction does not develop force, instead the repeated discharge of the APs contributes to the increase in amplitude of the contraction. In other words, the increments in tension (which are triggered by individual APs) will accumulate as a result of the intracellular free Ca2+ ions when APs are discharged at a rate higher than 1 Hz (Marshall, 1962).
The contraction of skeletal muscles is initiated by the nervous system: the moto-neuron triggers an action potential which propagates through the neuromuscular junction to the muscle plate, causing the contraction of the muscle fiber (Guyton and Hall, 2010). Interestingly, neuromuscular junctions that are present in the skeletal muscle do not occur in smooth muscle. In the myometrium the mechanism by which an AP is generated is different. It is believed that the AP burst can originate from any uterine cell. The initiating cells are referred to as pacemaker cells and they can shift from one contraction to another (Lodge and Sproat, 1981). The concept of a pacemaker in the myometrium has been considered and investigated for many years. It has been suggested that the uterus is myogenic in that it contracts in vivo and in vitro without the need for external stimuli and that any myometrial cell is capable of acting either as a pacemaker or pace-follower (Kao, 1959). However, research employing a variety of histological techniques have not yielded clear evidence of the presence of cells with the histological and electrophysiological properties of a functional pacemaker (Gherghiceanu and Popescu, 2005; Hinescu et al., 2006; Hinescu and Popescu, 2005; Suciu et al., 2007). The key issue about the origin of the electrical impulse, which initiates the myometrial contraction and the regulation of its direction of propagation, remains unclear in both the pregnant or non-pregnant uterus (Aguilar et al., 2010).
Contraction of the smooth muscle cell
The molecular basis of the smooth muscle contraction is represented by the cyclic interaction between the myosin head and the actin filament. During contraction the myosin heads tilt and drag the actin filament (over a small distance of approximately 10nm), while during relaxation the actin filaments are released. Later on, the myosin heads can rebind to another part of the actin filament and slide along further. A schematic representation is given in Figure 1.3. This mechanism is also known as the cross-bridge cycle, during which the hydrolysis of ATP occurs (Word, 1995). The chemical energy released by the ATP-hydrolysis is transformed by the myosin molecule into active muscular strength. Hence, myosin is regarded as the motor protein of the muscle (Klinke et al., 2009). In addition, cross-bridges also allow the myosin to pull an actin filament in one direction while simultaneously pulling another actin filament in the opposite direction. As a consequence, smooth muscle cells can contract as much as 80% of their length (skeletal muscles can contract less than 30% of their length) (Guyton and Hall, 2010).
The myosin heads are made of heavy and light protein chains. It has been shown that the contraction and relaxation are regulated by phosphorylation (acquisition of a phosphate group) and dephosphorylation (removal of a phosphate group by hydrolysis) of the myosin light protein chain. Phosphorylation of the regulatory light chain of myosin by Ca2+/CaM - dependent myosin light-chain kinase (MLCK) plays an important role in smooth muscle contraction (Word, 1995). In particular, contraction is initiated by a phosphorylation of myosin by which ATP is degraded into ADP (Guyton and Hall, 2010; Word, 1995). Relaxation is characterized by a low concentration of Ca2+ ions, inactivation of MLCK and dephosphorylation of the myosin light chain (Guyton and Hall, 2010).
As already described in the previous sections, compared to skeletal muscle cells, smooth muscle cells have a larger amount of voltage-gated Ca channels and fewer voltage-gated Na channels. Therefore, the generation and propagation of APs in smooth muscle is mainly regulated by Ca channels which open and close significantly slower compared to the Na channels. As a result, there is a very short onset time for contraction and relaxation of the smooth muscle in response to the electrical stimulus (Csapo and Goodall, 1954). This also corresponds to a slow cycling speed of the myosin crossbridges in smooth muscle, which is 10 to 300 times slower than in the skeletal muscle (Guyton and Hall, 2010).
To summarize, the myometrium is composed of smooth muscle cells. The basic process controlling the uterine contraction is the underlying electrical activity in the form of APs. APs propagate between muscle cells and open ion channels allowing the influx of calcium ions to produce a contraction. APs occur in groups and form a burst of activity which in humans can last more than a minute (Garfield et al., 2005) with the burst frequency around 0.1 Hz. The AP frequency within a burst has been reported to range between 0.1 - 10 Hz (Devedeux et al., 1993) but most of the studies focus on the frequency range 0.1 - 3 Hz (Leman et al., 1999; Marque et al., 1986) or 0.3 - 1 Hz (Garfield et al., 2005; Maner et al., 2003; Rabotti et al., 2010a, 2009). The number of bursts in a given time determines the frequency of a uterine contraction. Consequently, the duration of a burst determines the duration of the uterine contraction and the force generated by the whole uterus is determined by the propagation of APs from cell to cell and the amount of muscle mass involved (Garfield, 1984; Marshall, 1962). Each burst stops before the complete relaxation of the uterus (Marshall, 1962). The electrical properties and the excitability of the myometrial cells can be altered by agents that directly stimulate or inhibit uterine contractions.
Measurement of uterine contractile activity
To date, several different prognostic techniques have been developed to measure uterine contractions. The most commonly used clinical approaches are (i) the tocography (TOCO) see Figure 1.5 (Smyth, 1957), (ii) the intrauterine pressure catheter (IUPC), (iii) the use of biomarkers such as fetal fibronectin (fFN), white blood-cell count (WBC), or corticotropin releasing hormone (CRH) and (iv) the Bishop Scoring System.
The TOCO is the most commonly used non-invasive technique to measure the mechanical deflections produced by the uterine contractions. It is very simple to use and risk free for both, the mother and fetus. The tocodynamometer, a strain-gauge based measurement device, records the deflection of the maternal abdomen during a uterine contraction. The strain on the tensometric transducer, which is strapped on the patient's abdomen, is proportional to the strength of the contraction. The TOCO technique, typically performed in the third trimester of the pregnancy, is widely used by physicians in over 90% of women admitted to labor and delivery units. However, the information obtained is limited, may not detect all uterine contractions and can not differentiate contractions that will subside spontaneously from those that will lead to delivery. In addition, the frequency of contractions does not reflect the force of the labor (Buhimschi et al., 2003). Another major drawback of this instrument is the susceptibility to maternal motion artifacts. As a consequence of these uncertainties, patients will be either all treated as having preterm contractions or their treatment will be delayed until cervical change occurs (Garfield et al., 1998). In addition, tocodynamometric assessment of uterine contractile activity is very difficult in obese patients.
The IUPC represents a golden standard in the current obstetrical practice as it can provide an accurate assessment of uterine contractions. The IUPC is an invasive method that measures the intrauterine pressure via a catheter (Devedeux et al., 1993). It can be used only after a certain dilatation of the cervix is reached (late stages of pregnancy, close to delivery). A change in pressure inside the uterus is reflected by displacement of the fluid in the catheter.
While the IUPC is more reliable and accurate than TOCO, it is an invasive procedure that requires the rupture of the amniotic membranes, thus limiting its use to patients with complicated delivery. Therefore, the risk that complications might occur when using the IUPC increases significantly. Due to the poor predictive power of the TOCO and the invasive nature of the IUPC, neither technique has been beneficial in the accurate prediction of preterm labor or the diagnosis of true labor at term. The predictive power of biomarkers was also investigated. The fFN is a fibronectin protein produced by fetal cells, found in the birth canal of pregnant women, between the chorion and the decidua. The use of fFN has been proposed for the prediction of preterm labor and the management of women with symptoms of preterm labor. Some studies reported successful use of fFN in the prediction of actual premature birth (Iams, 2003; Lockwood, 2001), while others suggest that fFN has limited predictive value and conclude that there is no sufficient evidence to recommend its use (Berghella et al., 2008; Garfield et al., 2001; Hellemans et al., 1995; McNamara, 2003).
In a recent study, Hill and colleagues used recursive partitioning to identify gestational age-specific threshold values for infectious and endocrine biomarkers to predict preterm delivery (Hill et al., 2008). They have found that according to gestation age, two biomarkers, namely corticotropin releasing hormone (CRH) and white blood-cell count (WBC), provide a relatively high prediction accuracy for preterm delivery. Although studies of biomarkers have improved the understanding of the mechanisms of disease leading to spontaneous preterm birth, most potential biomarkers (of preterm birth) investigated in women with predisposition to preterm labor are similar with respect to diagnostic performance and accuracy (Goldenberg et al., 2005). That is, negative predictive values are superior to positive predictive values and the tests are usually more specific than sensitive (Berghella et al., 2008; Hill et al., 2008; McGregor et al., 1995). A high negative predictive value has also been observed for salivary estriol (McGregor et al., 1995).The Bishop Scoring System is a pre-labor scoring system to assist in predicting whether induction of labor will be required (Bishop, 1964). The total score is achieved by assessing the following five components during vaginal examination: position of the cervix, cervical dilatation, cervical effacement, cervical consistency and fetal station. However, the Bishop score was not found to contribute to a reduction in preterm labor.
The current state-of-the-art in labor monitoring can be summarized as follows (i) intrauterine pressure catheters provide the best information but with limited usability due to their invasive nature (rupture of the amniotic membranes), (ii) presently available uterine monitors such as TOCO are uncomfortable, less accurate and depend on the examiner for proper placement, and (iii) the predictive power of biomarkers is not sufficient for a successful diagnosis and (iv) the Bishop Scoring System has not lead to a reduction in preterm labor. While cervical change and the frequency of contractions are probably the two most frequently used clinical methods for assessing labor, there is still a high amount of controversy regarding the best way to evaluate and quantify the uterine contractile activity.
Two methods are currently employed to record the electrophysiological activity of uterine contractions: (i) the electromyography/electrohysterography (EMG/EHG), recorded by electrodes attached to the abdomen (see Figure 1.5), and (ii) a newly established method, the magnetomyography (MMG), based on recordings of the magnetic fields that correspond to electrical fields. These techniques measure the electrical/magnetic activity on the surface of the maternal abdomen, which is a result of a sequence of bursts or groups of action potentials that are generated and propagated in the uterine smooth muscle tissue - the myometrium. The term EHG is more specific, as it strictly refers to the electrical activity of the myometrial cells. However, in the context of uterine activity, the terms EHG and EMG are often used to described the same activity. Therefore, for the sake of simplicity, the term EMG will be used throughout this thesis to refer to the electrical activity of the uterus.
The uterine electrical activity, which reflects the original process of muscle fiber excitation due to the propagation of the APs, can be measured by means of internal and abdominal surface electrodes (see Figure 1.5). The EMG recordings of uterine activity date back to 1931 when Otto Bode used a galvanometric device to record the activity of the human uterus during labor for the first time (Bode, 1931). His work unveiled a technique with great potential as it provides a non-invasive and inexpensive assessment of uterine contractile activity. However, decades later, EMG measurements are still not adopted in obstetric practice, despite sustained scientific evidence that the EMG signal is representative of the electrophysiological changes occurring in the myometrium (Buhimschi et al., 1997; Buhimschi and Garfield, 1996; Buhimschi et al., 2003; Garfield et al., 2005; Maner et al., 2003).
Noninvasive EMG recordings have been extensively studied with an emphasis on both time (Buhimschi et al., 1998; Duchene et al., 1990; Verdenik et al., 2001) and frequency domain (Doret et al., 2005; Garfield et al., 2005) parameters for the prediction of labor. Garfield and colleagues performed simultaneous recording of the EMG activity directly from the uterus and the abdominal surface of rats (Buhimschi and Garfield, 1996). They were able to conclusively show that the signals recorded from the abdominal surface correspond to those generated in the uterus, suggesting that similar techniques can be used in humans. The recording of EMG activity on the human uterus using abdominal electrodes was also reported (Garfield et al., 1998).Other studies have shown that the power density of uterine EMG bursts in patients during active labor peaked at 0.71 ± 0.05 Hz as compared to non-laboring term (0.48 ± 0.03 Hz) patients (Garfield et al., 2005, 1998; Maner et al., 2006). In addition, the power density peak values were comparatively low for non-labor patients with respect to patients in active labor. The EMG has a high temporal resolution. However, because of the differences in the conductivities of tissue layers, the EMG signals are often filtered during their propagation to the surface of the abdomen.
All electrophysiological phenomena that occurs inside the human body are characterized by the flow of ion currents which can be detected by a measurement of potentials inside or on the surface of the human body (Baumgartner, 1995; Preissl, 2005; Rogalla and Barker, 1991). While the EMG technique allows successful assessment of uterine contractions, another method of observing biomagnetic signals is magnetomyography (MMG). MMG, the magnetic counterpart of the EMG, is a noninvasive technique that measures the magnetic fields associated with APs (see Figure 1.6).
MMG measurements are conducted externally, using sensitive magnetic sensors called superconducting quantum interference devices (SQUID). A SQUID is a very low noise magnetic field sensor, which converts the magnetic flux threading a pickup coil into voltage allowing the detection of weak magnetic signals. Since the SQUIDs rely on the physical phenomena found in superconductors they require cryogenic temperatures for proper operation. Therefore, the array of SQUIDs is immersed and cooled in liquid helium (at approx. âˆ’270" C), in a special vessel called a dewar.
The MMG recordings possess some important properties which renders them suitable for the analysis and characterization of the human biomagnetic activity. Compared to EMG signals, the MMG signals are detectable outside the boundary of the skin without making any contact with the body. Also they are independent of conductivity geometry, i.e. tissue conductivity. The measured electrical activity arises from the volume currents flowing (in the body) to the electrode sites and not directly due to the primary current generators. Therefore, the uterine EMG signals suffer from some degree of attenuation by the time they reach the surface of the maternal abdomen. By contrast, magnetic field recordings are more strongly coupled to the primary currents and are much less dependent on tissue conductivity boundaries. Thus, compared to the electric field, the magnetic field observed outside the human body offers a much more precise representation of the underlying activity.During a MMG recording detection of the signal is made outside the boundaries of the skin, that is, without any electrical contact with the body. Further, the MMG recordings provide a higher spatio-temporal resolution compared to EMG, as the spatiotemporal resolution obtained from the abdominal surface electrode recordings is limited based on the electrical properties of the abdomen and the practical difficulties of placingnumerous electrodes on the mother (Eswaran et al., 2009). In addition, the MMG recordings are independent of any kind of references, which ensure that each sensor mainly records localized sources.
MMG measurements span a frequency range from about 10 mHz to 1 kHz and field magnitudes from about 10 fT for spinal cord signals to about several pT for brain rhythms.
Figure 1.7 illustrates a few examples of magnetic field strength to provide the reader information regarding the magnitude of the MMG signals. It should be noted that the Earth field magnitude is about 0.5 mT and the urban magnetic noise about 1 nT to 1 T. This corresponds to a factor of 1 million to 1 billion larger than the MMG signals. Such large differences between signal and noise requires a very high accuracy of noise cancellation. Therefore, the system must be placed in a magnetically shielded room, to avoid the interference of the strong environmental noise and external magnetic fields with the biomagnetic fields generated by human organs.
First MMG recordings of spontaneous uterine activity were reported by Eswaran and colleagues, demonstrating the feasibility of the technique (Eswaran et al., 2002). The measurements were carried out with a system called SARA (SQUID array for reproductive assessment), CTF Systems Inc. Port Coquitlam Canada (see Figure 1.8). The system consists in primary superconductive quantum interference device (SQUID) magnetic sensors that are spaced approximately 3 cm apart in a concave array that covers the maternal abdomen from the pubic symphysis to the uterine fundus and laterally over a similar span. The surface of the array is curved to match the shape of the gravid abdomen. To allow SQUIDs to record the MMG signals, the mother must simply lean forward against the smooth surface of the array. Thus the SARA system is capable of obtaining electrophysiological data from the entire fetus and maternal reproductive system in a passive, consistent and non-invasive manner.
Because strong environmental noise and external magnetic fields can interfere with the human biomagnetic fields, SARA was installed in a magnetically shielded room (Vakuumschmelze; Hanau, Germany).
There are currently two such systems installed worldwide. The first SARA system, called SARA I, is operational since May 2000 at the University of Arkansas for Medical Sciences (UAMS), at the SARA Research Center, Little Rock, USA. As of September 2008 a second SARA system, called SARA II, is in operation at the University of T¨ubingen, fMEG Center, T¨ubingen, Germany. Compared to the system installed at the UAMS, SARA II provides a slightly better coverage of the perineal region, consequently covering a slightly larger surface of the gravid abdomen. For an overview of the organization and labeling of the magnetic sensors in both systems refer to the Appendix, Figure A.1.
Automatic detection of uterine contractile activity
The automatic detection of uterine contractions has been attempted by different studies. Radhakhrishnan and colleagues attempted to detect uterine contractions in EMG using higher-order zero-crossing analysis (Radhakrishnan et al., 2000). In addition to EMG measurements the authors performed TOCO and IUPC recordings on the investigated datasets to compare the outcomes of their proposed method. They have observed that the discriminating power of the higher-order crossing counts D(i) decreased as the order i increased (i > 2).
The method was applied to a subject, in non active labor, in the 38th week of gestation, see Figure 1.9. In (a) the processed uterine EMG is presented. In (b) the peaks of the normalized first-order zero-crossing (FOZC) counts coincide with the segments where contraction occurs. Figure 1.9 (c) presets the TOCO recording from the same subject.
The authors discuss the feasibility of this technique and conclude that (i) the firstorder zero crossing method is the most suitable to discriminate between contractile and non-contractile events and (ii) the results are consistent with the TOCO and IUPC recordings. However, the method lacks threshold detection and does not take into account the frequency characteristics of the signal.
Based on the assumption that the uterus remains quiescent throughout most of the pregnancy but close to the time of delivery its electrical activity increases, a generalized synchronization index, as an indicator to track the spatial patterns of uterine myometrial activity, was proposed by Ramon and colleagues in (Ramon et al., 2005). The synchronization of a sensor pair was inferred from a statistical tendency to maintain a constant phase difference over a given period of time (one should note that during that same time frame the analytic phase of each sensor may change markedly). The phase differences between two sensors were computed by subtracting the cumulative linear phase (the unwrapped phase of the analytic signal) of one sensor from the other. Synchronization between the phases of two signals was computed using Shanon's Entropy Function (Shannon and Weaver, 1998). where pi is the relative frequency of finding the phase difference modulus of 2â‡¡ in the ith bin (N represents the number of bins). To compute the phase difference in a 20 s stepping window, a total of N = 100 bins were used. Because e(t) can vary between zero and emax = ln(N), further normalization was necessary which yielded the generalized synchronization index (see Figure 1.10). The authors computed the generalized synchronization index in regional coil-pairs for six pregnant women with a gestational age (GA) ranging from 29 to 40 weeks. For any given coil, the coil-pair consisted of the surrounding six coils. This resulted in 21 unique combinations of coil pairs and synchronization indices.
The index was computed for a 4 minute segment (which included at least two contractions) and at points when the subjects reported perceivable contractions. For a comparative analysis, the magnetic field data was randomly shuffled in each channel and followed by the computation of the synchronization indices. The authors show that the synchronization indices of the unshuffled (original) data are much higher compared to the synchronization indices of the randomly shuffled data (red horizontal line in Figure 1.10) and conclude that the synchronization indices (see Figure 1.10) and their spatial distributions describe uterine contractions. Figure 1.11 shows the spatial patterns of the synchronization indices across the inspected windows and computed within a 20 s stepping window. The intensity scale is normalized in the range of zero (blue) to one (red), with red areas indicating higher synchronization.
In a recent study La Rosa and colleagues used a multiple change-point estimator along with the K-means clustering algorithm to detect uterine contractions in selected regions (see Figure 1.12) of a SQUID sensor array (La Rosa et al., 2008). The channels were modeled by a time varying auto regressive (AR) model. A segmentation algorithm based on the Schwarz Information Criterion (SIC) was used to estimate the timeinstants of changes in the parameters. To discriminate contractions, features such as the time segment power, i.e., root mean square (RMS), and the dominant frequency component, i.e., first-order zero-crossing (FOZC), were evaluated. A discrete-time binary decision signal, indicating the presence of a contraction, was created by applying an unsupervised clustering algorithm to classify the RMS values. The detection of multiple change points was performed using a binary search algorithm. Figure 1.13 (a) illustrates a down sampled and filtered (0.2 - 0.4 Hz) MMG signal with vertical grid lines indicating the estimated change points. In Figure 1.13, (b) and (c) show the RMS and FOZC computed on the estimated time segments. It can be observed that the FOZC poorly estimates the presence of a contraction in the investigated frequency range. Figure 1.13 (d) presents the clusters obtained after applying the K-means algorithm on the RMS values. Figure 1.13 (e) shows the binary decision signal and in (f) the time-intervals are illustrated where the mother reported the presence of a contraction.
However, in practice, the approach performed well only for a group of channels from the same patient, suggesting that a different model for segmentation should be attempted. Moreover, this approach does not take into account information from all of the sensors. In addition, the K-means clustering technique needs a-priori specification of the number of clusters in the data. The authors use three cluster centroids to discriminate the contractions from noise. From the very nature of this setting, this approach will always erringly identify some segments as contractions even though they may not be present in the data. On the other hand, if the dataset contains a long single contraction, by construction, their approach will label some parts of the uterine contraction as noise.
Finally, since the frequency content of the electrical burst activity corresponding to the uterine contractions changes with time (Garfield et al., 2005), a simple spectral analysis (RMS for this particular case) may not be sufficient to objectively capture the uterine contractile activity.