Research On Fast Segmentation Algorithm Of Vertebral CT Image Based On Improved SNIC

Medical image processing plays a very important role in doctors’ clinical diagnosis and is a hot issue in the field of image processing.In the computerassisted system,the segmentation of vertebral CT images is very important for diagnosing the condition of the spine.It not only replaces the manual segmentation of experts,saves manpower,but also provides a basis for subsequent 3D reconstruction.Because many traditional image segmentation algorithms are inefficient when segmenting medical images with a large number of pixels,the segmentation effect is poor.Therefore,this paper proposes a fast segmentation algorithm for vertebra CT images based on improved SNIC.The first is the research on the preprocessing of CT images of vertebrae.Through theoretical analysis and experimental comparison,a preprocessing method suitable for superpixel segmentation in this paper is selected.After that,the SNIC superpixel algorithm was studied.According to the characteristics of the algorithm and the vertebra CT image,the SNIC algorithm was improved from three directions: First,the method of restoring the original scale after the down-scale segmentation was used to improve the segmentation speed.Second,a dynamic parameter which can be adaptive according to the characteristics of different local areas is proposed to replace the original parameters,eliminating the process of manual parameter selection and improving the segmentation effect of the algorithm.The third is to add texture feature information to the algorithm calculation to make the algorithm pay more attention to the texture features of the vertebrae.Finally,this article studies the application of this algorithm on vertebral CT images.Using the improved SNIC algorithm to segment the CT images of vertebrae,and selecting the appropriate number of superpixel according to the segmentation performance;Select the feature of super pixel and extract the feature to form the feature vector;select an appropriate number of clusters and use FCM algorithm to cluster the feature vectors,merge superpixel according to the clustering results to achieve segmentation of vertebra CT images.The experimental verification part mainly conducts segmentation experiments on vertebra CT images in medical images,compares and evaluates the algorithm in this paper with other image segmentation algorithms,and calculates the evaluation coefficient to verify the effectiveness of the algorithm.The experimental results show that the vertebra CT image segmentation method combining the improved SNIC superpixel algorithm and the clustering algorithm proposed in this paper can ensure the segmentation accuracy and improve the segmentation speed.