Study On The Process Optimization Of Neuron Reconstruction Based On User Behavior Analysis

The brain neural network plays a decisive role in human thinking,emotion and behavior,and has important value in the research of brain diseases and brain-like intelligence.Neurons are the basic unit of brain network.Rebuilding the topology of neurons and analyzing their morphological characteristics are important directions in the research of brain networks.Over the past decade,with the gradual maturation of neural labeling and optical imaging technology,scientists have been able to obtain whole brain 3D images at the level of single neurons,laying the data foundation for neuronal morphology reconstruction.The next question is how to recognize and segment the morphological structure of neurons from these images and reconstruct the digital 3D model.At present,the gold standard for neuron reconstruction is artificial reconstruction,which can guarantee the accuracy to the greatest extent,but the efficiency is extremely low.In recent years,scientists have been committed to the exploration of automated neuron reconstruction methods,which is a hot research issue in neuro-optic image processing.A large number of fully automatic reconstruction tools have been developed,and the reconstruction speed has also been greatly improved.However,due to the complex and diverse structure of neurons,dense labeling,and serious image signal interference,it is difficult to obtain sufficient accuracy for the fully automatic method.Therefore,the semi-automatic method has become the mainstream technical solution for current neuronal reconstruction and related application research.It uses automated reconstruction algorithms to improve reconstruction efficiency,while adding manual intervention to reduce the error rate.Currently,it takes 15 to 70 hours to reconstruct each neuron using a semi-automatic method,and the reconstruction speed is an average of 0.1 to0.6 cm per hour.In addition to the two influencing factors of automatic algorithm performance and manual operation experience,the semi-automatic human-computer interaction process is also an important factor that affects the reconstruction speed,but this is usually considered to be an engineering problem related to software development,and the attention of the academic community is relatively low.In order to effectively obtain the interactive process of neuron reconstruction,this paper carried out related research on process mining,put forward a block-oriented process mining scheme,and realized automatic modeling of the reconstruction process.This scheme simplifies the complex loop structure through the idea of "divide and conquer",which can handle complex neuron reconstruction process.In addition,according to the steps of log collection,preprocessing and conformance checking in the scheme,the incomplete logs and various kinds of noise interference are excluded,and the process model excavated can truly reflect the actual interaction process.Based on this set of schemes,this article processed more than 100 reconstruction logs and accumulated analysis data for subsequent work.In order to use the extracted process model to optimize the neuron reconstruction process,this paper proposes data-oriented and process-oriented user behavior analysis and optimization schemes.Among them,the data-oriented analysis clusters the interactions in the model and divides them into five categories,and optimizes them according to the characteristics of each category.The process-oriented analysis locates the unreasonable structure in the model through the method of model replay,and uses the user's operation rules and behavior habits to simplify the model.This solution effectively simplifies the human-computer interaction process and improves the efficiency of human-computer interaction.Based on the above work,in view of the fact that the efficiency of manual intervention of reconstruction tools is generally low,based on the open source reconstruction tool GTree,a semi-automatic neuron reconstruction plug-in has been developed,which can not only ensure the accuracy of reconstruction through manual supervision,but also Simplify the way of human-computer interaction and reduce the degree of manual intervention.Compared with before optimization,the plug-in rebuild speed increased by 19%.In summary,the work of this paper basically achieves the purpose of research.Using the established neuron reconstruction process optimization scheme based on user behavior analysis,you can obtain a process model that reflects neuron reconstruction and optimize it accurately,which effectively improves the human Machine interaction efficiency.It shows that only through the joint advancement of automatic algorithms,manual operation experience and human-computer interaction process can the efficiency of neuron reconstruction be maximized.