Automatic Tracing Of Neurofilaments Under Fluorescence Microscopy Based On Digital Image Processing Technology

Video surveillance technology is a technology based on computer technology that identifies moving objects in video information,which has been widely used in sports,biology,traffic surveillance,human-computer interaction,daily life,video indexing,national defense,vehicle navigation,astronomical observation,college classroom monitoring,image retrieval and other fields.In recent years,video surveillance technology has also achieved good results in the medical field.When the neurofilaments are overexpressed,the transmission channels of the neural information are overloaded and the signal cannot be effectively transmitted to the receiving tissue,which will cause the patient's skeletal muscles to atrophy and eventually paralyze or die.The traditional method is to manually label neurofilaments in video recorded under fluorescence microscopy.However,this method requires high labor costs and causes human error.So this paper proposes an algorithm based on image processing technology to automatically track neurofilaments.The experimental part of this paper consists of three modules: image preprocessing module,image segmentation module and feature extraction module,which were used to analyze the neurofilaments movement.First,digitization was used to process neurofilaments motion video,which translated video information into digital image information that the machine could recognize.Then,neurofilaments images would be restored and normalized,and all images would be converted into standard patterns.Next,the image enhancement was used to process neurofilaments images.Second,image segmentation was used to process the neurofilaments images preprocessed.The complete target neurofilament was segmented based on threshold segmentation algorithm and association detection algorithm.The position of the target neurofilament in the next frame of the image was extracted based on the detection of the contact between the two frames.According to the characteristics of the neurofilament itself,the k-means clustering algorithm is used to segment the neurofilament in the target region.Then,neurofilaments features would be detected and extracted.The k-means clustered image was converted to a binary image.The maximum contour of the target neurofilament was extracted and dissolved into points which were used as the interpolation points of the snake model to extract the real contour of the target neurofilament.Spatial characteristics of neurofilaments between two adjacent images was established to determine the direction of neurofilaments.Finally,the trajectory of the neurofilaments were established so as to achieve the purpose of tracking neurofilaments.In the experimental part,the image processing tracking algorithm was compared with Mean Shift tracking algorithm,Kalman filter tracking algorithm and Particle filter tracking algorithm.It was found that the neurofilament tracking algorithm based on image processing was obviously superior to the other three algorithms and has better tracking accuracy and robustness.