Limits...
Automated Software Analysis of Fetal Movement Recorded during a Pregnant Woman's Sleep at Home.

Nishihara K, Ohki N, Kamata H, Ryo E, Horiuchi S - PLoS ONE (2015)

Bottom Line: We will also demonstrate an appropriate way to use the system (Experiment II).We evaluated the agreement of the manual and automated analyses for the same 10-sec epochs using prevalence-adjusted bias-adjusted kappa (PABAK) including quantitative indicators for prevalence and bias.The mean PABAK value was 0.83, which can be considered almost perfect.

View Article: PubMed Central - PubMed

Affiliation: Fatigue and Working Life Research Group, The Institute for Science of Labour, Kanagawa-ken, Japan; Integrated Brain Function Project, Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan.

ABSTRACT
Fetal movement is an important biological index of fetal well-being. Since 2008, we have been developing an original capacitive acceleration sensor and device that a pregnant woman can easily use to record fetal movement by herself at home during sleep. In this study, we report a newly developed automated software system for analyzing recorded fetal movement. This study will introduce the system and compare its results to those of a manual analysis of the same fetal movement signals (Experiment I). We will also demonstrate an appropriate way to use the system (Experiment II). In Experiment I, fetal movement data reported previously for six pregnant women at 28-38 gestational weeks were used. We evaluated the agreement of the manual and automated analyses for the same 10-sec epochs using prevalence-adjusted bias-adjusted kappa (PABAK) including quantitative indicators for prevalence and bias. The mean PABAK value was 0.83, which can be considered almost perfect. In Experiment II, twelve pregnant women at 24-36 gestational weeks recorded fetal movement at night once every four weeks. Overall, mean fetal movement counts per hour during maternal sleep significantly decreased along with gestational weeks, though individual differences in fetal development were noted. This newly developed automated analysis system can provide important data throughout late pregnancy.

No MeSH data available.


Related in: MedlinePlus

Maternal respiration movement.This figure shows maternal respiration influencing the FM channel (37 yrs, 36 gestational weeks). One grid mark indicates one second with 60 seconds per screen. The upper window shows the influence of the mother's breathing (3–5 sec intervals). We made two functions in order to remove respiration artifacts and explained in Step 3. The final decision is shown by the heavy blue line; the red lines indicate removal by the minimal threshold level.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4470661&req=5

pone.0130503.g004: Maternal respiration movement.This figure shows maternal respiration influencing the FM channel (37 yrs, 36 gestational weeks). One grid mark indicates one second with 60 seconds per screen. The upper window shows the influence of the mother's breathing (3–5 sec intervals). We made two functions in order to remove respiration artifacts and explained in Step 3. The final decision is shown by the heavy blue line; the red lines indicate removal by the minimal threshold level.

Mentions: Fig 4 shows artifacts from maternal breathing, which can easily be mistaken for fetal movement. Two steps are necessary to reject maternal breathing artifacts. First we set a notch filter at 0.3 Hz; then, we set a minimum threshold level. During Rapid Eye Movement (REM) sleep, there are large changes in the autonomic nervous system, which includes heart rate and respiration. Maternal breathing movements occur at 3–5 sec intervals and appear as artifacts in the FM channel during maternal REM sleep. Thus, we could not exclude them with only a 0.3 Hz notch filter. We also had to set a minimum threshold level. We tried various threshold levels for the integral values and then referred to the distribution of the intervals between detected signals using the histogram tool (in Part D in Fig 2). The histogram shows the frequency of the various intervals between detected signals. It ranges from 0 sec to 40 sec with steps every 1.0 sec. When the frequency counts for the 3.0 sec, 4.0 sec, and 5.0 sec columns were extremely high, the detected signals included respiration. In that case, we had to change the level of the threshold integral value. After many trials, we found that the optimal values for obtaining fetal movement data with minimal artifacts were 15 for the FM channel and 20 for the MM channel.


Automated Software Analysis of Fetal Movement Recorded during a Pregnant Woman's Sleep at Home.

Nishihara K, Ohki N, Kamata H, Ryo E, Horiuchi S - PLoS ONE (2015)

Maternal respiration movement.This figure shows maternal respiration influencing the FM channel (37 yrs, 36 gestational weeks). One grid mark indicates one second with 60 seconds per screen. The upper window shows the influence of the mother's breathing (3–5 sec intervals). We made two functions in order to remove respiration artifacts and explained in Step 3. The final decision is shown by the heavy blue line; the red lines indicate removal by the minimal threshold level.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4470661&req=5

pone.0130503.g004: Maternal respiration movement.This figure shows maternal respiration influencing the FM channel (37 yrs, 36 gestational weeks). One grid mark indicates one second with 60 seconds per screen. The upper window shows the influence of the mother's breathing (3–5 sec intervals). We made two functions in order to remove respiration artifacts and explained in Step 3. The final decision is shown by the heavy blue line; the red lines indicate removal by the minimal threshold level.
Mentions: Fig 4 shows artifacts from maternal breathing, which can easily be mistaken for fetal movement. Two steps are necessary to reject maternal breathing artifacts. First we set a notch filter at 0.3 Hz; then, we set a minimum threshold level. During Rapid Eye Movement (REM) sleep, there are large changes in the autonomic nervous system, which includes heart rate and respiration. Maternal breathing movements occur at 3–5 sec intervals and appear as artifacts in the FM channel during maternal REM sleep. Thus, we could not exclude them with only a 0.3 Hz notch filter. We also had to set a minimum threshold level. We tried various threshold levels for the integral values and then referred to the distribution of the intervals between detected signals using the histogram tool (in Part D in Fig 2). The histogram shows the frequency of the various intervals between detected signals. It ranges from 0 sec to 40 sec with steps every 1.0 sec. When the frequency counts for the 3.0 sec, 4.0 sec, and 5.0 sec columns were extremely high, the detected signals included respiration. In that case, we had to change the level of the threshold integral value. After many trials, we found that the optimal values for obtaining fetal movement data with minimal artifacts were 15 for the FM channel and 20 for the MM channel.

Bottom Line: We will also demonstrate an appropriate way to use the system (Experiment II).We evaluated the agreement of the manual and automated analyses for the same 10-sec epochs using prevalence-adjusted bias-adjusted kappa (PABAK) including quantitative indicators for prevalence and bias.The mean PABAK value was 0.83, which can be considered almost perfect.

View Article: PubMed Central - PubMed

Affiliation: Fatigue and Working Life Research Group, The Institute for Science of Labour, Kanagawa-ken, Japan; Integrated Brain Function Project, Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan.

ABSTRACT
Fetal movement is an important biological index of fetal well-being. Since 2008, we have been developing an original capacitive acceleration sensor and device that a pregnant woman can easily use to record fetal movement by herself at home during sleep. In this study, we report a newly developed automated software system for analyzing recorded fetal movement. This study will introduce the system and compare its results to those of a manual analysis of the same fetal movement signals (Experiment I). We will also demonstrate an appropriate way to use the system (Experiment II). In Experiment I, fetal movement data reported previously for six pregnant women at 28-38 gestational weeks were used. We evaluated the agreement of the manual and automated analyses for the same 10-sec epochs using prevalence-adjusted bias-adjusted kappa (PABAK) including quantitative indicators for prevalence and bias. The mean PABAK value was 0.83, which can be considered almost perfect. In Experiment II, twelve pregnant women at 24-36 gestational weeks recorded fetal movement at night once every four weeks. Overall, mean fetal movement counts per hour during maternal sleep significantly decreased along with gestational weeks, though individual differences in fetal development were noted. This newly developed automated analysis system can provide important data throughout late pregnancy.

No MeSH data available.


Related in: MedlinePlus