Research On Trigeninal Nerve Fiber Bundle Segmentation Algorithm Based On Brain Atlas

With the improvement of social medical awareness and medical level,trigeminal nerve diseases caused by vascular or tumor compression are gradually paid attention by researchers.The identification of the trigeminal nerve and the positional relationship between the trigeminal nerve and the lesion before surgery can effectively improve the operation.The success rate reduces the risk of surgery.At present,researchers have done a lot of research on fiber imaging of trigeminal nerve,and formed a complete set of artificial recognition methods.However,existing methods require medical workers to manually sketch single or multiple regions of interest on magnetic resonance images based on knowledge of trigeminal anatomy,and then manually filter the fiber tracking results to obtain independent trigeminal nerve fiber bundles.This method has two limitations: first,the researcher is easily interfered by subjective factors during the operation,which affects the accuracy of the research results to a certain extent;second,the researcher's drawing method is not fixed,and the standards are not uniform And the reproducibility that is difficult to achieve,resulting in certain limitations of this traditional manual method.In view of the above problems,this paper proposes an automatic segmentation algorithm of trigeminal nerve fibers based on brain atlas based on the white matter clustering atlas method and brain tissue segmentation tool.The specific work and results of this paper are as follows:1.Aiming at the problem of artificially delineating regions of interest and artificially screening fibers in traditional artificial methods,this paper use the fiber data of multiple groups of brain samples to establish a data-driven fiber clustering map to automatically segment the fiber data of new samples,and directly obtain the trigeminal nerve fiber bundle.In the experiment,data of 20 groups of young healthy people were selected as sample data.First,the brain stem was extracted as a region of interest for deterministic fiber tracking using the Freesurfer software segmentation tool.Secondly,through multi-sample registration and spectral clustering of 20 groups of fiber data,a data-driven fiber clustering map was created.According to the small characteristics of the trigeminal nerve,the trigeminal nerve fiber bundle was labeledby sub-classifying the brainstem fiber bundle during the establishment of the fiber map.Finally,the brain stem fiber data of the new sample was automatically segmented using fiber atlases to obtain trigeminal nerve fiber bundles.The results show that the proposed method successfully segmented 5 sets of trigeminal nerve fiber bundles,and the traditional manual method successfully identified 4 sets of trigeminal nerve fiber bundles.The weighted Dice coefficients between the two results are 0.865,0.939,0.824,and 0.942.Experimental results show that this method can effectively avoid the influence of human factors,and greatly improve the work efficiency of neurosurgeons and cranial nerve researchers.2.Aiming at the small and difficult to divide trigeminal nerve fiber bundles,based on the white matter clustering atlas method,the idea of secondary classification is added.In the study,multiple groups of brainstem fiber data were selected,and fiber registration was performed on the multiple sample data.Then classified and located the trigeminal nerve fiber bundles on the registered brainstem fiber data to reduce the scope of research.Finally,the range was classified twice,and the trigeminal nerve fiber bundle was accurately labeled to complete the creation of a data-driven fiber atlas.3.In this paper,an automatic segmentation method is used to perform trigeminal nerve fiber bundle imaging experiments on patients with vascular compression and tumor compression trigeminal neuralgia.The experimental results not only successfully segmented the trigeminal nerve fiber bundle in the patient's brain,but also clearly demonstrated the positional relationship between the fiber bundle and the lesion.This paper discusses the problems existing in the traditional artificial trigeminal nerve fiber imaging methods,and proposes corresponding improvement methods.By comparing experimental results,both methods can track the trigeminal nerve fiber bundle.However,the tracking results of traditional manual methods are easily disturbed by human factors,and the entire process is labor-intensive,time-consuming and artificially dependent.The method in this paper uses data-driven fiber clustering atlas to automatically segment the trigeminal nerve fiber bundle,effectively avoiding the influence of human factors,and greatly improving the work efficiency of neurosurgeons and cranial nerve researchers.