Research On Intracranial Aneurysm Detection Based On DSA And CTA Images

Intracranial aneurysm(IA)is one of the most harmful cerebrovascular diseases,the symptoms of which are hard to find,with urgent onset process.This disease has a great possibility of sequelae and a high fatality rate.Therefore,the early diagnosis and treatment of IA are particularly important.The current methods for IA examination include Computed Tomography Angiography(CTA),Magnetic Resonance Angiography(MRA)and Digital Subtraction Angiography(DSA).The conventional way of medical images analysis of IA requires the doctor to put marks on them and adjust their display angle,which mean heavier workload and lower accuracy.Therefore,this paper is mainly on the design of DSA and CTA image-based IA detection algorithms,which are to reconstruct the three-dimensional blood vessel structure on the slice data,analyze the structural features of the blood vessels to define the likely focus and finally finish the IA detection.The algorithms are designed to assist doctors in diagnosis and provide help for early treatment of patients.Firstly,the DSA and CTA data used in the experiment are preprocessed in this paper.With blood vessel slice images,morphological image processing and thresholding segmentation methods are used to extract the stratified blood vessel data,which can reduce the interference of the background on the blood vessel structure.On this basis,the surface rendering method is used for 3D reconstruction,and the 3D morphological algorithm is used to calculate the 3D maximum connected component and extract the main body of blood vessels,thereby reducing the influence of small blood vessels and noise on IA detection.Secondly,with the optimized 3D blood vessel reconstruction data,from a new research perspective,a step-based automatic search algorithm for intracranial aneurysm detection is proposed in this paper based on the tubular characteristics of vascular structure.The algorithm intercepts the vascular structure by generating iterative planes,and determines the searching direction by calculating the minimum cross section of the blood vessel.Then,the vascular cross-section and its location are used as features to classify the iterative centers to pinpoint the suspicious locations of the IA.The experimental results show that the algorithm,to some extent,is efficient in IA detection,but affected by the limitations of one-way detection,which leads to the F2-score of 0.707.Ultimately,in order to analyze the overall vascular structure,an aneurysm detection algorithm based on the three-dimensional skeleton of the blood vessel is proposed in this paper.This algorithm calculates the skeleton of the blood vessel structure,and extracts the 3D information of the blood vessels,including features such as the volume,the area of cross section,and deflection angle of the skeleton point.It also uses the method of random forest to classify the blood vessel skeleton points and detect the suspicious areas of the IA.This algorithm avoids the limitations of the step-based algorithm,improving the detection performance and the F2-score to 0.825.To summarize,the experimental results show that the intracranial aneurysm detection algorithms based on DSA and CTA images proposed in this paper can accurately find the location of the IA,which can help doctors to efficiently analyze medical images and bring certain application prospects.