Experimental Report on Medical Image Segmentation (Including Source Code)

Experimental report on medical image segmentation (including source code), primarily covering: analysis of edge extraction effects using different operators (Sobel, Prewitt, Roberts, Laplacian); threshold segmentation techniques; and segmentation discussion using watershed method.

This experimental report aims to investigate various medical image segmentation approaches. We initially implemented edge detection using Sobel, Prewitt, Roberts and Laplacian operators (employing convolutional kernels for gradient computation), followed by detailed comparative analysis of their performance characteristics. Subsequently, we explored threshold segmentation applications, examining optimal threshold selection strategies (including Otsu's method for automatic threshold determination) to achieve accurate segmentation. Finally, we conducted in-depth analysis of watershed segmentation effectiveness (implementing marker-controlled watershed transformation to prevent over-segmentation) and its medical imaging applications.

Through this experimental study, we seek to deepen understanding of fundamental medical image segmentation methodologies and provide reference for future research and practical implementations.