Deep learning plays a vital role in revolutionizing the healthcare system, primarily in disease diagnosis, enabling the automatic segmentation of clinical images. The manual process of analysis is a tedious and time-consuming task even for experts, which may lead to imprecise evaluation. In this paper, a Transformer Based Semantic Segmentation Network is proposed as a new method for applications in the area of medical imaging. The novelty approach outperforms the majority of state-of-the art models achieving 98.45% of accuracy. It was tested with the “Bacteria detection with dark-field microscopy” dataset, which consists of 366 images of spirochaete bacteria mixed with red blood cells.