Digital signal extraction approach for cardiotocography image.

Area of Science:

• Medical Imaging
• Signal Processing
• Obstetrics

Background:

• Cardiotocography (CTG) is vital in obstetrics for monitoring fetal well-being.
• Existing methods often struggle with complex backgrounds in CTG images.
• Accurate extraction of fetal heart rate and uterine contraction signals is crucial.

Purpose of the Study:

• To develop an efficient signal extraction method for binary CTG images.
• To address challenges posed by complex backgrounds in CTG analysis.
• To improve the accuracy of fetal heart rate and uterine contraction data retrieval.

Main Methods:

• Image preprocessing to remove background grids and noise using templates.
• Morphological operations to fill signal curve gaps.
• Projection mapping for curve localization and column scanning for data extraction.
• Amplitude calibration of extracted signals.

Main Results:

• High correlation coefficients (0.9991 ± 0.0030 for FHR, 0.9904 ± 0.0208 for UC) achieved on simulated data.
• Low mean absolute and root mean square errors for both fetal heart rate (FHR) and uterine contraction (UC) signals.
• Validation on clinical CTG images confirmed remarkable similarity to original signals.

Conclusions:

• The proposed method effectively extracts FHR and UC signals from CTG images.
• The technique is robust even in the presence of complex backgrounds.
• This advancement aids in more reliable obstetric monitoring through CTG analysis.