Neural Network Analysis and Evaluation of the Fetal Heart Rate

Noguchi, Yuichi; Matsumoto, Fujihiko; Maeda, Kazuo; Nagasawa, Takashi

doi:10.3390/a2010019

Cited by 15 publications

(6 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The accuracy of neural network diagnosis was confirmed by simultaneously obtained FHR score in new cases. The FHR score was significantly higher in the group with a probability of a pathological outcome, intermediate in the group with a probability of an intermediate outcome and significantly low in the group with a probability of a normal outcome [10,11].…”

Section: Artificial Neural Networkmentioning

confidence: 86%

“…Trained software was installed onto other computers and used for the diagnosis of new cases, where the eight FHR parameters in three succeeding 5-minutes periods were entered into the software, with the output results being the probability that an outcome would be normal, intermediate or pathological. The results were objective and easily understood without education because they were reported as a percentage [7,10,11].…”

Section: Artificial Neural Netwrok (Ann) Systemmentioning

confidence: 99%

See 1 more Smart Citation

Central Computerized Automatic Fetal Heart Rate Diagnosis with a Rapid and Direct Alarm System

Maeda¹,

Utsu²,

Noguchi³

et al. 2012

TOMDJ

Self Cite

View full text Add to dashboard Cite

Aims: This study aimed to simplify fetal monitoring, reduce inter-observer differences and false-positive diagnosis and monitor a large number of births simultaneously. Methods: Fetal signals from several births were transmitted to a central computer via local area network (LAN) or telemetry and analyzed using a multichannel timesharing system. Fetal heart rate (FHR) abnormalities were detected by using three programs: the experts' knowledge system, power spectral analysis and artificial neural network. Abnormal results were automatically communicated directly to the attending doctor. Instead of an FHR chart recorder, the original fetal signals were stored on the computer and re-processed on demand. Results: a maximal FHR score in the first stage of labor indicated a low Apgar score, and correlated with umbilical blood pH. The fetal distress index derived from the FHR score was three or more in cases of fetal acidosis. The neural network yielded probabilities of fetal outcome that coincided with the FHR score, and the neural index derived from these probabilities predicted fetal outcome. Pathological sinusoidal FHR and severe loss of FHR variability were automatically diagnosed by power spectral analysis. Perinatal mortality was 1.1 in 1.000 births, which was significantly lower using this central computerized system than the previous system, and no cases of cerebral palsy were reported 2 months after delivery. Conclusion: The central computerized automated fetal monitoring system improved fetal outcomes even in institutions dealing with a large number of births.

show abstract

Section: Artificial Neural Networkmentioning

confidence: 86%

Section: Artificial Neural Netwrok (Ann) Systemmentioning

confidence: 99%

Central Computerized Automatic Fetal Heart Rate Diagnosis with a Rapid and Direct Alarm System

Maeda¹,

Utsu²,

Noguchi³

et al. 2012

TOMDJ

Self Cite

View full text Add to dashboard Cite

show abstract

“…For instance, rule-based classifiers have been addressed in [8] and neural networks in [13]. In [14], three different techniques were exploited for feature selection (principal component analysis, group of adaptive models evolution, and neural networks) followed by direct correlation of welldiscriminating feature sets with FHR pathology.…”

Section: A Backgroundmentioning

confidence: 99%

Fetal Heart Rate Classification Using Generative Models

Dash

Quirk

Djurić

2014

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

This paper presents novel methods for classification of fetal heart rate (FHR) signals into categories that are meaningful for clinical implementation. They are based on generative models (GMs) and Bayesian theory. Instead of using scalar features that summarize information obtained from long-duration data, the models allow for explicit use of feature sequences derived from local patterns of FHR evolution. We compare our methods with a deterministic expert system for classification and with a support vector machine approach that relies on system-identification and heart rate variability features. We tested the classifiers on 83 retrospectively collected FHR records, with the gold-standard true diagnosis defined using umbilical cord pH values. We found that our methods consistently performed as well as or better than these, suggesting that the use of GMs and the Bayesian paradigm can bring significant improvement to automatic FHR classification approaches.

show abstract

“…Performance of the neural network computer: The recognition results obtained by the neural network with teaching input after training for 10,000 times, all data carried the probabilities of 0.998 (or 99.8 %) to1.000 (or 100 %) ( Table 2). Therefore, this well-trained neural network can be used to recognize new patients not included in the teaching patterns that is to say "open recognition" [12].…”

Section: Artificial Neural Networkmentioning

confidence: 99%

Artificial Neural Network System in the Automated Diagnosis of Fetal Heart Rate

Meda¹,

Noguchi²

2012

J Health Med Inform

View full text Add to dashboard Cite

Fetal heart rate (FHR) tracing with uterine contraction is the most commonly utilized for fetal monitoring during pregnancy and in the labor. Automated computerized diagnosis of FHR was intended and recently introduced artificial neural network analysis because of its very objective nature. FHR tracing was quantified and FHR score was obtained, FHR frequency power spectrum was analyzed to diagnose sinusoidal FHR to apply in artificial neural network analysis. The neural network was composed of a soft ware with three layers associated with back-propagation system, and it was trained with 8 FHR parameters including the sinusoidal FHR for 10,000 times to obtain 100 % correct internal check. Trained network soft ware was copied with new computer to diagnose new subjects. Diagnostic input was the 8 FHR parameters of 3 periods of 5 min, and output was probabilities to be normal, intermediate and pathologic outcomes in percentage, of which diagnosis was correct in the comparison to simultaneously calculated FHR score that was high in pathologic outcome probability, moderate in intermediate and low in normal outcome probability. Neural network index derived from pathologic and normal outcome probabilities was useful in the outcome prediction of prolonged fetal monitoring.

show abstract

Neural Network Analysis and Evaluation of the Fetal Heart Rate

Cited by 15 publications

References 31 publications

Central Computerized Automatic Fetal Heart Rate Diagnosis with a Rapid and Direct Alarm System

Central Computerized Automatic Fetal Heart Rate Diagnosis with a Rapid and Direct Alarm System

Fetal Heart Rate Classification Using Generative Models

Artificial Neural Network System in the Automated Diagnosis of Fetal Heart Rate

Contact Info

Product

Resources

About