Medical image segmentation plays a vital role in diagnosis and treatment planning by extracting clinically relevant information from imaging data. Conventional methods often struggle with variations in anatomical structure and imaging quality, leading to suboptimal segmentation. Recent advancements in Deep Learning, particularly Convolutional Neural Networks (CNNs) and Transformers, have improved segmentation accuracy; however, individual models such as U-Net, ResNet, and Transformer still face limitations in preserving spatial details, extracting deep features, and modeling long-range dependencies. This study proposes a hybrid Deep Learning model that integrates U-Net, ResNet, and Transformer to overcome these challenges and enhance segmentation performance. The proposed hybrid model was evaluated on several publicly available datasets, including BraTS, ISIC, and DRIVE, using Dice Similarity Coefficient (DSC) and Intersection over Union (IoU) as performance metrics. Experimental results indicate that the hybrid model achieved a DSC of 0.92 and an IoU of 0.86, outperforming U-Net (DSC: 0.82, IoU: 0.75), ResNet (DSC: 0.85, IoU: 0.78), and Transformer (DSC: 0.88, IoU: 0.80). Additionally, the model maintained an inference time of 55 ms per image, demonstrating its potential for real-time applications. This study highlights the benefits of combining CNN-based and Transformer-based architectures to capture both local details and global context, providing an effective and efficient solution for medical image segmentation.
