Diagnosis of cardiotocographic sinusoidal patterns by spectral analyses

• Published in: Biomedical signal processing and control Vol. 93; p. 106174
• Main Authors: Savirón-Cornudella, Ricardo, Laliena Bielsa, Antonio, Esteban-Escaño, Javier, Calvo Torres, Javier, Chóliz Ezquerro, Marta, Castán Larraz, Berta, Díaz de Terán Martínez-Berganza, Elisa, Rodríguez Castaño, María José, Álvaro Navidad, Miguel, Andeyro García, Mercedes, Whyte Orozco, Jaime, Castán Mateo, Sergio, Esteban, Luis Mariano
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.07.2024
• Subjects: Cardiotocography; Fetal anemia; Pseudosinusoidal pattern; Sinusoidal pattern; Fetal anemia; Cardiotocography; Pseudosinusoidal pattern; Sinusoidal pattern
• ISSN: 1746-8094; 1746-8108
• DOI: 10.1016/j.bspc.2024.106174

•Sinusoidal pattern in fetal heart rate monitoring is linked to severe anemia or hypovolemia.•Computational spectral analysis distinguishes sinusoidal and pseudosinusoidal patterns from controls.•The use of the energy proportion variable predicts sinusoidal patterns with adverse perinatal effects.•Classification tree differentiates sinusoidal from pseudosinusoidal patterns.•Implementing model in labor units enhances diagnosis and management of suspected sinusoidal-like patterns. The sinusoidal pattern in cardiotocographic (CTG) monitoring shows a sinus-shaped signal longer than 30 min without short-term variability. It is commonly linked to fetal morbidity, particularly severe fetal anemia. Pseudosinusoidal patterns resemble sinusoidal patterns but without adverse fetal outcomes. This study aims to characterise sinusoidal and pseudosinusoidal patterns using spectral analysis. A multicenter study case-control was conducted between January 2012 and February 2023. Maternal characteristics, perinatal data, and CTG parameters through spectral analysis were examined. The spectrum of the electrocardiographic signal was calculated, and the proportion of energy (PE), short- and long-term variability, amplitude, and the differences between sinusoidal, pseudosinusoidal, and control groups were compared. A predictive model for signal type was built using a classification tree. 60 CTG records were collected, including 38 controls. Of the 13 sinusoidal patterns detected, all exhibited a sinusoidal pattern with a PE ratio > 0.3, 9 of them (69 %) had a PE ratio > 0.5, and 4 (31 %) were in the range of 0.3–0.5. Among the 9 cases diagnosed as pseudosinusoidal, all had a sinusoidal pattern with a PE within the range of 0.3–0.5. Every control exhibited a PE < 0.3, except for one case. Short-term variability demonstrated limited discriminatory capability, while long-term variability showed a strong discriminatory capacity. For the classification tree, accuracy diagnosis was 92.3 %, 88.8 %, and 97.3 % for the sinusoidal, pseudosinusoidal, and control groups, respectively. Computerised spectral analysis and the variable PE within the frequency range of 1.8–3.5 are reliable parameters to discriminate sinusoidal patterns.