The Fetal heart rate (FHR) is a valuable source of information which is routinely monitored to assess fetal well-being via cardiotocography (CTG). The use of CTG to diagnose Intra Uterine Growth Restriction (IUGR), one of the most common fetal pathologies, has been proposed due to its low cost and noninvasiveness. Recently, Siamese Neural Networks (SNN) have been introduced as an effective strategy for metric learning, due to their capability to learn complex embeddings which can be used for classification. Despite being more difficult to train compared to traditional deep neural networks, SNNs have the inherent advantage of requiring less training data and can be extended to handle multi-class problems without need to retrain the entire model. In this paper, we present preliminary results on the use of SNNs for IUGR identification in CTG recordings, which explore the suitability of these models in the field of CTG interpretation. The developed model comprises three identical branches based on the ResNet architecture, each fed with a separate input tuple, i.e., FHR and Fetal Movement Profile signals. The branches are connected via an output layer that computes the Euclidean distance among the embeddings of the three inputs (i.e., the triplet). The dataset employed was extracted from a wider CTG database and was balanced with respect to the class and gestational age. 466 recordings were used for training and validation and 92 as a hold-out test set. The trained SNN is used to project the data in a lower dimensional space. The embedded data are then classified using a 1-nearest neighbor. The proposed approach allowed to discriminate between IUGR and physiological pregnancies with an initial accuracy score of 73% on the test set, which aligns with previous studies and shows the validity of the proposed approach, that is open to further improvement.