Pediatric appendicitis is one of the most frequent emergency conditions in children, and timely diagnosis is crucial to prevent complications. While ultrasound imaging is commonly used due to its safety and accessibility, its interpretation remains highly operator-dependent, leading to inconsistencies in diagnosis across clinicians and institutions, as reported in prior clinical studies [6]. In this study, the use of Vision Transformers (ViT) is explored for the automated detection of pediatric appendicitis using exclusively real ultrasound images. A Swin Transformer model (“swin tiny patch4 window7 224”) is fine-tuned on a curated dataset of abdominal ultrasound images from pediatric patients at the Children’s Hospital St. Hedwig (Germany). The model achieves an accuracy of 0.89, along with precision, recall, and F1-score all at 0.89, significantly outperforming the average diagnostic accuracy of initial clinical assessments reported in the dataset (approximately 0.68). These results suggest that ViT architectures can effectively capture complex visual patterns in medical imaging and offer reliable support for clinical decision-making in pediatric appendicitis diagnosis.