Contour Model Research Of Peripheral Nerve Internal Non-splitting Merging Phase Fascicular Groups From MicroCT Images

Peripheral nerve injury is a common clinical disease that can lead to severe paralysis and neurological dysfunction.The existing peripheral nerve models do not study the internal neural fascicular groups structure and lack modeling studies on the characteristics and mathematical patterns,resulting in high disability rates after nerve repair surgery.During surgery to repair peripheral nerve injury,precise docking of neural fascicular groups with the same functional properties will restore nerve conduction and control to a large extent,and the contour model of neural fascicular groups can provide a localization benchmark for docking nerves.Therefore,the contour modeling of fascicular groups in Micro CT images of peripheral nerves has become one of the key problems.Because of the complex internal structure of peripheral nerves,several splitting merging behaviors may occur in a short distance space（1～5 mm）.Therefore,this paper only focuses on the modeling of the neural fascicular contours in the non-splitting merging phase,and uses the spline curve method and the Fourier model method to construct the neural fascicular contours model and conduct comparative analysis,respectively,in order to realize the construction of the model for constructing the neural fascicular contours in the non-splitting merging phase and to explore the patterns of the parameters in the model.This paper is supported by Natural Science Foundation of Guangdong Province,China（No.2018A0303130137）,National Natural Science Foundation of China（No.61975248）and Opening Project of Key Laboratory of High-Performance Computation of Guangdong Province,China（No.TH1528）.The main work of this paper is as follows:（1）Modeling the contour of neural fascicular using two spline curve methods,Bezier curves and B spline curves were used to model the neural fascicular contours,and the feasibility of constructing neural fascicular contours model using spline curves was investigated.（2）In order to study the laws of the parameters of the neural fascicular contour model,this paper proposes to use Fourier model to construct the neural fascicular contours.The advantages and disadvantages of the Fourier model and B spline model methods are compared and analyzed.（3）The experiments of statistical analysis of Fourier model parameters were conducted to obtain the probability density function of each parameter separately.And the Fourier model parameters of the continuous contour of a single neural fascicular group are analyzed and the parameter variation curves are obtained.（4）Explore the laws of Fourier model parameters,conduct experiments on the reconstruction of neural fascicular contours,and verify the correctness of the results.The experimental results show that:（1）Higher-order Bezier curves are not suitable for the construction of the neural fascicular contour curve model.The high complexity of the higher-order Bezier curve model is difficult to control,and it is difficult to ensure continuity at the model connections.（2）The number of control points in the B spline model can significantly affect the accuracy of the model,and the accuracy of the model enters the steady state when the number of control points reaches 21 or more.（3）The third-order quasi-uniform B spline model with 21 control points can be used to construct the neural fascicular contour,and this method has low model complexity while having high accuracy.（4）Compared with the neural fascicular contour constructed using the B spline model and the Fourier model:the former model has slightly higher accuracy than the latter,but the parameters of the B spline model are not regular.（5）The probability density distributions of the Fourier model fundamental frequency parameters b₁,c₁and harmonic parameters all satisfy the t-distribution with position/scale,and the fundamental frequency parameters a₁and d₁satisfy the segmented Gaussian distribution.The scale range of the fundamental frequency parameters is much larger than that of the harmonic parameters,and the Fourier model constructed with only a small number of fundamental frequency parameters can capture the overall shape of the neural fascicular contour.（6）The Fourier model parameters can be clustered into 64 classes by the AP clustering algorithm,at which time the Dice coefficient of constructed neural fascicular contour image reaches 92.09%,the Hausdorff distance is less than 5.25 pixels,and the relative boundary error average is 1.26%.In summary,this paper shows that the 4th-order Fourier model can be used to model the neural fascicular contours in the non-dividing/merging phase of Micro CT images,and the parameters in the model conform to the t-distribution with position/scale or segmented Gaussian distribution.The research in this paper provides an important reference key data support to realize the precise docking of neural fascicular groups with the same functional properties in nerve injury repair surgery and the diagnosis and treatment of peripheral nerve diseases.