In this paper, we explore the application of self-supervised learning, specifically the DINO (DIstillation of NOisy labels) model, for the classification of parasitic eggs in microscopic images. Traditional methods for detecting these eggs rely on microscopy, which is both time-consuming and requires skilled personnel. To address these limitations, we utilized a dataset containing over 12,000 images representing 9 types of parasitic eggs. The DINO model, which leverages a teacher-student framework without the need for labeled data, demonstrated superior performance with an accuracy of 87%, compared to 80% achieved by the Vision Transformer (ViT). This suggests that self-supervised learning, particularly with models like DINO, offers a promising alternative for medical image classification tasks, especially in scenarios where labeled data is scarce.