The field of medical image analysis is rapidly evolving, with a focus on developing innovative methods to improve image segmentation, classification, and understanding. Recent research has emphasized the importance of capturing anatomical structure and relationships in medical images, leading to the development of new architectures and techniques that incorporate spatial and relational information. These advances have resulted in significant improvements in performance and interpretability, enabling more accurate and reliable analysis of medical images. Notable papers in this area include: CORAL, which proposes a coordinative learning framework to capture local and global structure in volumetric images, achieving state-of-the-art performance on benchmark datasets. AGENet, which introduces a novel framework that incorporates spatial relationships through edge-aware geodesic distance learning, demonstrating improvements over existing methods. DCMM-Transformer, which presents a new ViT architecture that incorporates a Degree-Corrected Mixed-Membership model to capture latent anatomical groupings in medical images, showing superior performance and generalizability. GCA-ResUNet, which proposes an efficient segmentation network that integrates Grouped Coordinate Attention into ResNet-50 residual blocks, achieving high accuracy and computational efficiency. RS-CA-HSICT, which introduces a hybrid deep learning approach that leverages the strengths of CNN and Transformer for enhanced MPox detection, reporting high classification accuracy and F1-score.