Cardiac Image Segmentation Model Based on Improved Level Set Modeling

In recent years, medical image segmentation based on variational level set methods has been widely used in image processing, but due to image grey level inhomogeneity and features with symbols, the image in segmentation is computationally intensive, computationally time consuming and computationally expensive. These problems are mainly caused by the image grey level inhomogeneity and the dependence on the sign distance function in the gradient descent method. To solve these problems, we propose a new level-set activity contour model that is combined with a fast computational method, the Alternating Direction Method of Multipliers (ADMM), to solve the level-set activity contour model. The problem of inaccurate segmentation results is solved by introducing a new level set active contour model that uses neighbourhood information to accurately segment the region of interest. The model uses local contextual information to mitigate the effect of grey scale variations, thus improving the segmentation accuracy. Since the use of traditional segmentation methods, such as gradient descent, is usually very time consuming, computationally complex and computationally expensive, so an otherwise complex problem can be broken down into several simple sub-problems, which can be solved one by one to solve the entire problem more quickly and accurately by using a fast computational method, the ADMM algorithm. By solving the simple sub-problems and then indirectly solving the original complex problem, this decomposition can solve the problem more efficiently and accurately, thus significantly reducing the computational time and cost. Experimental results validate the robustness of our proposed model against grey scale inhomogeneity in segmented cardiac images. The model demonstrates the ability to evolve curves quickly and accurately depict the contours of cardiac images. To assess the effectiveness of the model, we conducted comparative experiments using Dice coefficient and Jaccard index as evaluation metrics. The experimental data show that our proposed model consistently achieves higher Dice coefficients and Jaccard indices compared to other existing models, indicating superior segmentation performance. In conclusion, our improved level-set contour model combined with a fast computational method, the ADMM algorithm, provides an effective solution to the difficult problems prevalent in medical image segmentation, with significant improvements in accuracy, computation time, and cost-effectiveness. Experimental results show that the new model not only has strong robustness to images with uneven gray scale, but also has higher segmentation efficiency and accuracy, and uses less time.