The uterine electromyogram, also named Electrohysterogram (EHG), is a non-invasive technique that has been used for pregnancy and labour monitoring as well as for research work on uterine physiology. This technique is well established in this field. There is however a vast unexplored potential in the EHG that is currently the subject of interdisciplinary research work involving different scientific fields such as medicine, engineering, physics and mathematics.

In this paper, an unsupervised clustering method is applied to a previously obtained set of frequency spectral representations of the respective EHG signal contractions that were previously automatically detected and delineated. An innovative approach using the complete spectrum projection is described, rather than a set of relevant points. The feasibility of the method is established despite the concerns of possible computational burden incurred by the processing of the whole spectrum. Given the unsupervised nature of this classification, a validation procedure was performed whereas the obtained clusters were labelled through the correlation with the common knowledge about the most relevant uterine contraction types, as described in the literature. As a result of this study, a spectral description of the Alvarez contractions was obtained where it was possible to breakdown these important events in two different types according to their spectrum. Spectral estimates of Braxton-Hicks contractions were also obtained and associated to one of the clusters. This led to a full spectral characterization of these uterine events.

子宫肌电图又称电子宫肌电图（EHG），是一种非侵入性技术，已用于妊娠和分娩监测以及子宫生理学研究工作。该技术在该领域中已经建立。但是，EHG具有巨大的未开发潜力，目前是跨学科研究工作的主题，涉及医学，工程，物理和数学等不同科学领域。

在本文中，将无监督聚类方法应用于先前获得的各个EHG信号收缩的频谱表示集，这些频谱表示先前已自动检测和描绘。描述了一种使用完整频谱投影的创新方法，而不是一组相关点。尽管担心整个频谱的处理可能引起计算负担，但该方法的可行性得以确立。考虑到这种分类的无监督性质，执行了验证程序，而获得的簇则通过与最相关的子宫收缩类型的常识的相关性进行标记，如文献所述。这项研究的结果，获得了阿尔瓦雷斯收缩的光谱描述，其中可以根据它们的光谱将这些重要事件分解为两种不同的类型。还获得了Braxton-Hicks收缩的光谱估计值，并将其与这些星团之一相关联。这导致了这些子宫事件的全光谱表征。