A Literature Review on Segmentation of Medical Images using U-Net and Other Techniques

Image segmentation’s digital image processing subfield has numerous applications in image analysis. The growing science of medical image analysis has made it difficult to segment regions, disorders, or anomalies in medical imaging. Monitoring the growth of conditions like plaque and tumors is made easier with the help of medical image segmentation. Areas of interest (ROIs) from the dataset are extracted as a component of medical image segmentation. such as that from ultrasound imaging (US). Medical image segmentation can take a long time, but recent advancements in software techniques like Deep Learning, Machine Learning, and Artificial Intelligence (AI) are making routine tasks easier to complete. so that it makes it possible to analyze anatomical data with greater precision by separating out only the essential regions. The segmentation of intima carotids and media on an ultrasound image is crucial for reducing the annual number of deaths caused by atherosclerosis. Plaque, or the accumulation of fats, cholesterol, and other substances on and within the artery walls, is known as atherosclerosis. Recently, a number of image segmentation tasks have implemented deep neural network models. In this survey, we look at the previous research that a number of authors have done on segmentation using a variety of methods, such as the U-net (U-Net architecture was developed with minimal changes to CNN architecture) and its variants. For better understanding, we have also studied a few more techniques of Segmentation in addition to U-net.