Enhanced Z-axis precision of 3D medical imaging via improved optical flow method
Three-dimensional medical imaging plays a vital role in diagnosis and treatment.However,due to the structure of the slices,the resolution in the longitudinal direction is relatively low,which poses certain challenges for tasks such as three-dimensional reconstruction.This paper addresses the issue of insufficient resolution in the z-axis direction of three-dimensional medical images and proposes a novel method based on frame interpolation techniques.This method improves image continuity by generating intermediate slices,indirectly enhancing the quality of three-dimensional reconstruction.Aiming to overcome model distortion and resolution limitations inherent in traditional image processing methods,a compact encoder-decoder network is designed to integrate intermediate optical flow estimation and feature reconstruction.Furthermore,the network effectively merges global and local information through a feature pyramid structure and employs a dual loss function to optimize image reconstruction quality and the geometric consistency of the feature space.Experimental results demonstrate that,compared to existing frame interpolation methods,this approach significantly improves image quality,structural similarity,absolute error,and perceptual quality,with an average enhancement of approximately 2.7%to 5.8%.These results indicate that the frame interpolation technique proposed in this research performs well on brain tumor datasets,showcasing its potential advantages in enhancing the continuity and detail of medical images.This research offers a promising direction that could positively impact the accuracy of clinical diagnoses and treatment outcomes,providing new insights for further studies in the field of medical image processing.
万方数据
维普
