Accurate segmentation of nuclei in histopathology images is critical for understanding tissue morphology and aiding in disease diagnosis, particularly cancer. However, this task is challenging due to the high variability in staining and diverse morphological features. In this study, we propose a novel approach that integrates a graph-based attention mechanism into the U-Net architecture. Our method utilizes a state-of-the-art encoder backbone and introduces a Pairwise Node Similarity Attention Module (PNSAM), which computes the similarity between feature channels using a kernel function that inherently applies a dot product to capture spatial information. This module enhances the relationships between local and non-local feature vectors within a feature map obtained from multiple encoder layers, forming a graph attention map. Additionally, we incorporate a channel pruning mechanism that leverages predefined statistical knowledge to select important individual channels for graph attention map creation. The resulting graph attention map enhances encoder features for skip connections. Furthermore, we combine activated features from multiple trainable PNSAM heads to generate a more diverse and robust feature map. We evaluated our novel architecture on three widely recognized datasets: Monuseg, TNBC, and CryoNuSeg. Our method outperformed various state-of-the-art approaches on each dataset, demonstrating its efficiency and robustness in nuclei segmentation tasks. The code is available at: Github