Research on Coronary CT Image Segmentation Algorithm Based on Deep Learning
To assist doctors in diagnosing coronary artery diseases better and improve the segmentation accuracy of coronary CT images,three deep learning network models,Transformer,DeepLab V3+and U-Net,were studied.First,the coronary images were preprocessed with enhancement,denoising and normalization to reduce segmenta-tion difficulty.Then,three deep learning network models were used to perform pixel-level segmentation of coronary images,and their segmentation effects were compared.The Transformer model achieved the best performance in PA,MPA,and MIoU evaluation metrics,demonstrating significant advantages in the accuracy of medical image segmentation,especially in fine boundary area segmentation.Experiment results show that this method is effective and feasible,significantly outperforming the DeepLab V3+and U-Net models in the same conditions.
NETL
NSTL
万方数据
