Research And Realization On Color Perception Of Medical Images Based On Multi-features Fusion

Medical imaging is a widely used technology in both clinical and research,but most of them presents gray-scale images,which means that it uses light and shade to express physiological information.As is well-known,it is difficult for naked eyes to distinguish subtle changes of gray-scale,which leads to the neglect of some important information in medical images.Compared to gray-scale image,color image with rich and clear detail can effectively display and present medical information and achieve better observation,cognition and understanding.In order to realize the goal of fast color perception of medical images and maximize the useful information in medical images,an improved medical image color perception method based on multi-features fusion and hierarchical density peak clustering is proposed in this thesis.Besides,simulation system of color perception equips proposed method has been applied to the color perception of 3D medical image.The main contributions of this thesis are listed as follows:1.Considering the problems of medical image presents image content in the form of scale image,and high time complexity of visualization in high-dimensional medical image data,a gradient feature fusion method is adopted to extract useful information as much as possible to fully present the structure and detail information of the original image in this thesis.Then,some representative pixels of the local region are selected by using the method of density peak based on hierarchical clustering,which has been proved that it can reduce the time complexity to O(MN)where M is much less than N.In addition,the experiments on CT,MRI,PET,DTI and Ultrasound verify the feasibility and validity of the proposed algorithm.2.The improved algorithm based on multi-features fusion and hierarchical density peak clustering has been applied to achieve 3D medical image color perception.Specifically,3D feature of each voxel in the 3D medical image will be exacted firstly.Then,the method randomly selects several voxels as representative voxels and give color information to those voxels.Finally,it transmits color information to all voxels according to the neighborhood relationship between the voxels of existing color information and the rest,and yields 3D color medical image.The experiments on the sequences of the 3D CT and MRI show that the color of different slices in sequences are consistent.Besides,the visual effect after 3D reconstruction with ImageJ is favorable,which proves the feasibility and effectiveness of the improved algorithm.3.A simulation system equips the proposed methods mentioned above is designed for medical image color perception.The system achieves color perception of multi-modal medical images.Users input any modal medical image data such as CT,MRI,PET,DTI,ultrasound image,3D CT sequence,3D MRI sequence and other modal medical image,and set parameters,then the system can output colorized medical images with rich color information and detail in real time.