This paper presents a significant approach utilizing self-supervised transformer-based framework for histopathology image segmentation with self-attention mechanism to perform accurate segmentation of multi-organ histopathology images. By utilizing the self-supervised learning, the method learns efficiently the representation of varied features from unlabeled data. The proposed framework uses MonuSeg data from Kaggle for processing. Results reveal that the proposed approach performs better than conventional approaches by handling diverse nuclei shapes and overlapping structures. This method also enhances segmentation accuracy and computational efficiency. The experimentation reveals that the proposed self-learning transformer-based model with attention mechanism for image segmentation is more efficient than the existing methods.