A Comparative Analysis of U-Net-Based Segmentation Techniques for MRI Images

Image segmentation is essential in the areas of computer vision and image analysis. It allows to extract important information by separating an image into meaningful sections. Accurate image segmentation is critical for identifying objects, barriers, or anomalies in real-time in applications such as autonomous cars, medical imaging, and surveillance systems. Overall, the capacity to partition images into semantically meaningful regions enables to replicate human-like perception. In this paper, we have carried out an extensive comparative study of various image segmentation techniques and found unet performs better on MRI images as compared to other techniques. The techniques are analysed based on various performance metrics such as Dice coefficient, sensitivity, specificity, and accuracy.