The U-Net Model and it’s Variants: A Revolution in Medical Image Segmentation

Effective medical image segmentation is a crucial task for accurate diagnosis and treatment planning. An innovative approach combining manual masking with mathematical processing, where the objective is to refine training data, focusing the algorithm on relevant annotations. The mathematical function converts the area of interest pixel value into a specific colour while other region’s pixel values being converted to either 0 or [0,0,0]. Utilizing this mathematical methodology, a specialized dataset is masked, augmented, and trained on U-Net model. The Relative Root Mean Square Error is 0.2275.