Evaluation of fetal hypoxia during the 1st stage of labor, based on the combination of Cardiotocography, Fetal Pulse Oximetry and Fetal Doppler Velocimetry, using self-organising neural networks

Aim: The aim of this study is the early and objective diagnosis of fetal hypoxia during labour, based on the simultaneous application and analysis of cardiotocography (CTG), fetal pulse oximetry (FPO) tracings and Fetal Doppler Velocimetry indices.

Materials and Methods: Our study included 142 primigravidas in term, with uncomplicated singleton pregnancies. In every case CTG and FPO was used during labour. The Doppler index (PI) was recorded for the descenting aorta (AOR–ΡΙ) and renal artery (RA–ΡΙ) of the fetus. The acid-base balance of the fetus was evaluated with pH and base deficit-Bdecf of the umbilical artery. The women were divided into two groups: Group A (pH<7.15, Bdecf>–10 mmol/l, n=19) and Group B (pH>7.15, Bdecf<–10 mmol/l, n=123).

Analysis: In this study, we analyzed the data of the last 20 min of the 1^st stage of labour and calculated the power of the FHR fluctuations (P_VLF), in frequencies between 0.008–0.015 Hz, using the matching pursuit algorithm. We also calculated the total time in which fetal arterial oxygen saturation was<30%. During the same time, AOR-PI and RA-PI were estimated between uterine contractions, with a time interval of 5 min.

Results: Comparing the two groups, we observed that the values of all 4 parameters of the study were elevated in Group A and the difference was statistically significant in every parameter (table 1). We also estimated the prognostic value of every parameter and constructed the according ROC-curve (table 2). After the application of the SOM network, we constructed a 2-cluster matrix (table 3) and the overall prognostic value of all 4 parameters was estimated (table 4).

Conclusion: The simultaneous application and analysis of CTG, FPO and Doppler indices, can distinguish the acidemic fetuses with high sensitivity and specificity and can introduce a method of early and objective diagnosis of fetal hypoxia, resulting in reduction of perinatal mortality and morbidity and unnecessary interventions.

Table 1: Study parameters values in mean±Standard Error

	Group A (n=19)	Group B (n=123)	p
tFsPO2<30% (min)	4.80±0.74	0.81±0.29	<0.001
PVLF	220.0±27.1	65.0±10.8	<0.001
AOR-PI	2.42±0.12	1.64±0.05	<0.001
RA-PI	2.11±0.09	1.47±0.03	<0.001

Table 2: Prognostic value of the study parameters. AUC: Area Under Curve

	Sensitivity (%)	Specificity (%)	Likelihood Ratio	AUC
tFsPO2<30% (min)	67.6	86.8	5.15	0.84±0.07
PVLF	65.3	84.7	4.28	0.80±0.08
AOR-PI	89.1	83.5	5.43	0.93±0.03
RA-PI	94.9	84.1	5.97	0.95±0.02

Table 3: Topographic distribution of the cases using the SOM network matrix.

	Group A (n=19)	Group B (n=123)
Cluster 1	2	108
Cluster 2	17	15

Table 4: Overall prognostic value of all parameters, using the SOM network matrix.

Sensitivity (%)	Specificity (%)	Likelihood Ratio
89.4	87.8	7.32

Literatur: 1. The prognostic value of Doppler measurements in fetal descending aorta in conjunction with fetal pulse oximetry recordings during labor. Kassanos D., Loghis C., Vasios G., Hintipas E., Bournas N., Kallinteri C., Thomopoulos P., Salamalekis E. Proceedings of the 8th World Congress of Perinatal Medicine, September 2007, Florence, Italy 2. Computerized analysis of fetal heart rate variability using the matching pursuits technique as an indicator of fetal distress during labor. E. Salamalekis, E. Hintipas, I. Salloum, G. Vasios, C. Loghis, N. Vitoratos, Ch. Chrelias, G. Creatsas. J Matern Fetal Neonatal Med. 2006; 19(3): 165-9 3. Computerised intrapartum diagnosis of fetal hypoxia based on fetal heart rate monitoring and fetal pulse oximetry recordings utilising wavelet analysis and neural networks. Salamalekis E, Thomopoulos P, Giannaris D, Salloum I, Vasios G, Prentza A, Koutsouris D. BJOG. 2002 Oct;109(10):1137-42.

Doppler Velocimetry - Fetal hypoxia - fetal pulse oximetry - neural networks