Quantitative Analysis of Fetal Actocardiogram: Update

Actocardiogram (ACG): The ACG simultaneously traced FHR curve and fetal movement signals. The MHz level continuous wave (CW) ultrasound was used, of which SPTA intensity was as weak as 1 mW/cm2. The fetal movement Doppler signals obtained by 2 MHz source ultrasound was 20-50 Hz. The ultrasound detected Doppler fetal heart beat signals at the same time, of which Doppler frequency was 100 or more HZ, i.e. single ultrasonic probe detected two phenomena, FHR and fetal movements. Fetal movement Doppler signal was separated by a band-pass filter, and formed low frequency spikes, of which amplitude was parallel to that of fetal movement (Figures 1 and 2) [1]. Uterine contraction was also simultaneously recorded to prepare the CTG function.


Introduction
Although fetal movements were evaluated by maternal perception or abdominal motion detected by mechanical actograph at abdominal surface in old time, the results were insufficient to scientific fetal movement study [3]. A direct detection of fetal movement at fetal surface was required, and the device was invented by Maeda in 1984 [1]. Although the CTG was diagnosed mainly by FHR pattern classification [4], several vague problems and diagnostic difficulty were found in the past. It was, therefore, necessary to analyze the CTG with quantified techniques and numeric evaluation, and the result was the creation of fetal state from non-reactive FHR, the differentiation of physiological sinusoidal FHR from true ominous sinusoidal FHR, and so on, which were related to the lack of fetal movement study. Although some problems were solve by the real-time B-mode ultrasound [7], most diagnostic difficulties were solved by the handling of ultrasonic Doppler fetal movement signals [1]. The prototype actocardiograph was handmade by Maeda (Figures 1 and 2) and reported after the confirmation of its basic properties [1].

Methods
The FHR was quantified by the FHR score, and fetal movements by the ACG. There were various commercial ACG models after the prototype ACG (Figure 1), including MT-320, MT-325, MT-332, MT-333U, MT-430, MT-516, MT-517, MT-522 and MT-540 (TOITU, Tokyo). Recent ACG works were mainly studied by the ACG records offered by friends of the author, therefore, deep appreciation is expressed to the kind offering. Four parameters were quantitatively analyzed in the quantified studies on fetal behavior [2]. Physiologic sinusoidal FHR was differentiated from true ominous one by fetal periodic respiratory and swallowing movements [8]. Developmental mechanism of acceleration and LTV was studied. Short and long fetal outcomes were studied by the A/B ratio [9]. New prognostic value of the loss of LTV was clarified in this report.

Quantitative diagnosis of CTG by FHR score
It was the first trial in 1960s to quantitatively analyze intrapartum CTG instead of visual pattern classification ( Figure 3). The percentage of low Apgar score gave evaluation scores from 1 to 4 in abnormal FHR changes ( Table 1). The method was statistically a goodness measure. The evaluation scores were summarized in 5 min to obtain the FHR score [2], which was compared to Apgar scores and significant correlation was found (Figure 4) [5]. The CTG analysis with FHR score were programmed in the computer for automated CTG diagnosis [5,6,10].
Fetal state is abnormal, if FHR score is 10-19, and highly abnormal if the score is 20 or more. Thus the FHR score comprehensively evaluates fetal status.

ACG spike height
Subject moving distance The amplitude of fetal movement spikes recorded at fetal ACG chart was linearly parallel to the moving distance of ACG subject (a steel ball moved in the water). The ACG precisely records the movement of the subject [1].

Figure 3:
Quantified analysis of fetal heart rate change to determine the FHR score in non-interventional labors in 1969 [3], where FHR was manually analyzed, and it was computerized in 1970s.

FHR
Apgar score <7 (%) Evaluation score  Table 1: FHR score is the sum of evaluation scores in 5 min, which were determined by the incidence of low Apgar cases [3].

Other quantitative computerized CTG analyses
Artificial neural network computer analysis: Objective decision was made by neural network analysis, because of the network training with actual FHR data of known outcome cases instead of experts' knowledge [10].
Training of neural network computer: The neural network softwear was trained by 8 FHR data (baseline FHR, variability, sinusoidal heat rate, dip number, dip duration, nadir FHR, recovery time and lag time) of 3 normal, 3 intermediate and 14 pathological outcome cases for 10,000 times by Noguchi and colleagues to obtain 100% correct internal check.

Diagnosis of neural network:
Trained softwear was installled into other computers to diagnose new cases with three outcome probabilities to be normal, suspicios and pathological.
Accuracy of neural network diagnosis: It was confirmed by the FHR score, which was high in pathologic, modrate in suspicios and low in normal outcome probabilities.

Frequency analysis of FHR traces:
Since FHR record is composed of various frequency signs, e.g. straight baseline, the acceleration, deceleation, sinusoidal change and long term variability,and so on, the fast Furier transform (FFT) frequenxcy analysis was suitable in the quantitative FHR analysis.
Main subject of the FFT frequency analysis was the differentiaion of true ominus sinusoidal FHR from physiological benign sinusoidal change. The true ominous sinusoidal FHR was diagnosed when the low frequency area to total spectral area (La/Ta) ratio was 39% or more and at the same time peak power spetral densiy (PPSD) was 300 or more bpm 2 /Hz, while the values were lower in the physiological false sinusoidal FHR than true one [11]. In addition, the loss of LTV less than resting fetal state was diagnosed when the La/Ta was less than 15% and the PPSD was less than 60 bpm 2 /Hz [12].

Quantified diagnoasis of fetal behavior by the ACG Visual diagnosis of fetal behavior with actocardiogram:
Resting fetal state is visually diagnosed when there is no acceleation, but no fetal movement burst was registered, and baseline variability is preserved. Active fetal state is characterized by fequent FHR aceleation, which are synchronized to fetal movement bursts, and baseline variability is as large as 5 to 24 bpm. Highly active fetal state is diagnsed by long duraion of acelertion synchronized to long lasting fetal movements. The longest acceleration lasted for about 5 minutes. Intermediate state showed rare FHR accelerations but not zero and accompanied rare fetal movement burst [2].

Quantified analysis of fetal behavior:
The fetal ACG was analyzed by four parameters as follows; Four fetal movement parameters Four parameters' values were determined in the fetal ACGs of nomal fetuses in the late stage of pregnancy (Table 2). Fetal behavior will be determined by 4 parameters in new cases. The importance of A/B ratio was confirmed, because it is not influenced by fetal behavior in the evaluation of fetal statuses.

Difference of ACG parameters in normal fetus and fetal hypoxia:
There was significant difference of quantified parameters between normal and hypoxic fetuses, i.e. occupancy was 32.67% in normal fetus and 10.00% in hypoxia, frequency was 0.65 and 0.24 cpm, A/B ratio was 1.03 and 0, respectively [13].

Differentiation of physiologic sinusoidal FHR from true omioua one by the ACG
The CTG hardly differentiated physiologic sinusoidal FHR fom the truly ominous one. A physiologic one was easily diagnosed to be harmless in fetal ACG when the periodic fetal movements synchonized to the sine wave-like FHR, e.g. periodic fetal respiratory or mouth movements provocate physiologic sinusoidal FHR ( Figure 5) [8].
Developmental mechanism of FHR acceleration and LTV studied by the ACG Acceleration: A motion increases heart rate in human physiology. It was clear also in the fetus, i.e. a fetal movement burst accompanies triangular FHR acceleration (Figure 6), where the movement preceds acceleration for approx. 7 sec. The acceleration was lost in the non-   reactive FHR, whicn follows severely hypoxic FHR changes some days later, including the braycardia, late deceleration and the loss of variability [14].

The mechanism to produce triangular shape of acceleration
A triangular curve developed in electric wave bursts after passing thriugh an integral circuit with 7 sec delay time constant. Also triangular heart rate curves developed in the 1 min lasting continuos leg motions in an adult [15]. The electric and physiologic simulations explained the triangular FHR developing mechanism, which was the brain excitation The mechanism to develop FHR variability in the brain A repeated moderate fetal movements provoked synchronized FHR accelerations (Figure 6), and minor fetal movements also provoked FHR baseine changes (Figure 7). These facts indicate that the long term variability (LTV) develops as the response of fetal brain to minor fetal movements. Since the FHR delay was 7 sec, the location to respond the movement is the same as FHR acceleration, i.e. it is the mid brain [16].

Prenatal quantitative diagnosis of outcome with the A/B ratios of ACG
The A/B ratio was standardized dividing the total duratin of accleration by the total duration of movement bursts (Figure 8). ACGs of 15 common fetal disorders were quantitatively analysed with A/B ratios, and compared to the Apgar scores and the numeric long term outcome after births [9].    in the mid brain by the movements, i.e., Terao et al. [16] reported that the acceleration developed in the mid brain.
The short term outcome analysed by 1 and 5 min Apgar score closely correlated A/B ratios, i.e. Y (1 min Apgar)=7.68X (A/B ratio)-1.75, R 2 =0.85, p<0.001, Y (5 min Apgar)=6.44X+0.58, R 2 =0.68, p<0.001, and Y (numeric long term outcome)=6.42X+0.05, R 2 =0.71, p<0.001. Short and long term outcomes were abnormal, when the A/B ratio was less than 1. The fact was remarkable that a spastic quadriplegia case was found among cases of lower A/B ratio than 1. Therefore, A/B ratio of ACG is a useful parameter to expect fetal short and long term outcomes. Regression equations are listed [9].

COMMENT Comprehensive evaluaton of the fetus
Fetal diagnosis with the pattern classification tends to the single item diagnosis but not to cover the total condition of the fetus. In quantitative diagosis, the FHR score predicts Apgar score and umbilical cord arterial blood pH even in the first stage of labor. Neural netork analyais reports the pathological outcome probability. In the ACG, the A/B ratio predicts pathological Apgar score and long term outcome after the birth. Automated comprehensive diagnosis and outcome prediction are big advantage of quantitative analysis of FHR and ACG.

Advantage of fetal movement record
Since the change of movement signal of ACG is the same as those of fetal movement (Figure 2), the inflence of fetal movement on FHR is precisely evluated, i.e. fetal movement precedes FHR change, e.g. periodic fetal movements provoked sine wave-like FHR changes ( Figure 5), where the physiologic sinusoidal FHR is differentiated from the truly ominous sinusoidal heart rate by the ACG. Resting state of the fetus is differentiated by the absence of fetal movement bursts.from non-reactive FHR. The loss of acceleration against feal movement burst shows hypoxic suppression of fetal brain predicting the severe hypoxia.

Frequency analysis of FHR traces
The tecnique was introduced with the purose to automatically diagnose truly ominous sinusoidal FHR, which appeared without fetal movement record, and quantitatively confirmed by the FFT frequency analysis [11]. It was a surprize to diagnose severe loss of FHR variabilty less than the resting fetal state also by a FFT frequency analysis of FHR record [12].

The fetal state in severe loss of variability (LTV)
Since anencephalic fetal ACG recorded neither acceleration nor LTV, normal fetal brain is responsible to the development of acceleration acceleration in adult exercise [15], and the fetal acceleration was reported to be produced in the midbrain in the studies on anencephalic fetus [16]. The FHR acceleration disappeared in the non-reactive FHR, while the LTV was preserved, and severe hypoxic FHR changes appear some days later in non-reactive FHR cases, including bradycardia, late deceleration and the loss of LTV. The outcome of C-section in this state was ominous, if compared to reactive FHR [14], i.e. fetal damage was heavy in cases of the loss of LTV, which will be severe brain damage. In an intrapartum fetus of heavy loss of LTV less than 1 bpm similar to anencephalic fetus ( Figure 12) whose mother refused C-section and a severely depressed neonate with apnea was vaginally delivered, where 1 min Apgar was 3 due to the brain damage. Therefore, a severe loss of LTV means such brain damage as the loss of bran in anencephalic fetus. A severe hypoxic loss of FHR variability was rare, but the outcome will be ominous with probable CP caused by neuronal cell necrosis. Thus, the C-section after the severe loss of variability may not guarantee totally healthy neonate, but there will be the risk of possible brain    The comparison of CTG ( right) of a severe fetal asphyxia (hypoxia), of which FHR was a late deceleration and the loss of variability (LTV), which was the same as the baseline of the anencepaly (left), of wich LTV ampltude was less than 1 bpm. Fetal brain damage in the asphyxia was as severe as the loss of brain in the anencephly [15].
and LTV (Figures 9-11), but not in the brain cortex because no heart rate change was recognized by the person who developed the triangular damage followed by cerebral palsy. That is why earlier C-section is recommended before severe loss of variability [15]. The ideal timing of C-section will be shown after the estimation of hypoxic impact on fetal brain, e.g. by the hypoxic index, which is determied with the hypoxia duration (min) divided by the PaO 2 (%) where nadir FHR (bpm) is adopted instead of PaO 2 , because of close relation (R 2 =1) in rabbit heart rate and PaO 2 ( Figure 13) [15]. Since the hypoxic index of 3 fetuses after the loss of variability were 25-26, a C-section is recommended when the hypoxic index reached 20-24, while this threshold hypoxic index value should be studied in more cases.

Abnormal FHR in general insults without hypoxia
Abnormal FHR changes were reported in cytomgalovirus (CMV) infected fetus [17] and in congenital syphilis [18]. Therapeutic decision still seemes controversial in the occasion because of no hypoxia. However, as discussed above, such abnormal FHR as severe loss of LTV is caused by fetal brain damage, the FHR changes can develop by viral or bacterial toxin or oher general insults. As the loss of LTV means severe damage of fetal brain, FHR changes in various general insults should be treated by the same strategy as hypoxic FHR changes, i.e. C-section before the loss of variability. The A/B ratio of ACG will be helpful for the outcome prediction.

Computerization of fetal monitoring
The FHR score calculation, neural network analysis of FHR, frequency analysis of FHR, discusssed in this report, were progrmmed and working in the centralized fetal monitoring, improving perinatal status [5,6,10]. Quantified ACG and hypoxic index will be added to the system in the near future. An attending obstetrician will input continuous FHR into the computer system, and receive analyzed results rapidly and directly from the system. The time consuming visual CTG analysis will be changed to the computerization to receive objective, comprehensive and precise diagnostic data [6].

Conclusion
Since the ACG and its quantitative analysis proposed new field of fetal evaluation, which is totally objective in quantitative as well as visual analysis of ACG as discussed in this report, the subjective visual FHR pattern diagnosis, whch was vague with big interobserver difference, is definitely improved and the fetal management is greatly progressed by the introduction of quantified FHR analysis and fetal ACG, where C-section was recommeded to be performed before the loss of FHR variability. In addition, computerized automatic feal diagnosis, which utilizes various quantified analyses, extensively improves obstetric statuses.