Research On Three-dimensional Visualization Based On Multi-atlas Segmentation

In the tissues and organs of the human body,the brain is undoubtedly the core site,and the human brain has an extremely complex internal structure,through the computer image to explore the study of human brain tissue is an important research direction outside the anatomy.However,the structure of the human brain is very complex,and there is a big difference,making the commonly used magnetic resonance imaging diagnosis has been a certain impact,especially in the image segmentation is particularly prominent.Therefore,the organization segmentation and three-dimensional visualization of human brain image of nuclear magnetic resonance have become a hot issue in image processing.In order to solve the segmentation of human brain images,this paper adopts the method of multi-map registration.The core idea of this method is to use the template map and the target image generated by the registration of the deformation parameters,the need to split the organizational structure mapping on the mark image,they get a multi-map segmentation results,and then all the results of the integration of the segmentation,And then successfully split out the organization of interest.At present,the results of multi-map segmentation are often carried out without targeted three-dimensional reconstruction,lack of corresponding three-dimensional visualization algorithm support,and multi-map segmentation based on three-dimensional visualization.So that the results can not be intuitively displayed in front of people.In view of the above situation,this paper compares the three-dimensional visualization algorithm,and selects the volume rendering algorithm to reconstruct the segmentation result.The principle of body drawing method is studied,and the method of light projection.The transfer function in the volume rendering is to map the three-dimensional volume data into optical properties,so the design of the transfer function directly affects the effect and quality of the three-dimensional display.In this paper,we propose a transfer function design for multi-map segmentation:1.The reasonable transfer function is based on the image data type of the result of multi-map segmentation.2.The design provides the function of manually adjusting the transfer function The controls.Secondly,taking into account the complexity and systematic nature of the multi-map segmentation algorithm,this paper designs and develops a three-dimensional visualization software platform based on the results of the transfer function of multi-map segmentation and the three-dimensional visualization integration of multi-map segmentation results.And provide a system development software platform for further study of multi-map segmentation method.Software platform with QT development tools,VTK visualization tools,ITK image algorithm toolkit,can achieve DICOM,NII and RAW and other medical image format read and volume rendering display.And the division of the medical image extraction process,the registration process and the integration process and three-dimensional visualization work in one set,the system platform function is complete,has a high research and application development reference value.