Skull defect reconstruction method based on three-dimensional convolutional neural network

Abstract

The invention discloses a skull defect reconstruction method based on a three-dimensional convolutional neural network. The skull defect reconstruction method comprises the following steps: S1, constructing a healthy skull three-dimensional model by utilizing complete head CT data of a healthy human body; S2, generating a plurality of groups of simulated patient data of skull defects according tothe healthy skull three-dimensional model; S3, generating training data by utilizing the simulated patient data; and S4, constructing a three-dimensional convolutional neural network, and training thethree-dimensional convolutional neural network by using the training data to obtain an automatic reconstruction network model of the defective skull. According to the invention, a large number of virtual patient models are generated by utilizing CT volume data of a healthy person, and an automatic skull defect reconstruction model is trained through deep learning to complete automatic reconstruction of skull defects, so that automatic reconstruction of the defective skull is realized.

Description

technical field [0001] The invention relates to a method for reconstructing a skull defect, in particular to a method for reconstructing a skull defect based on a three-dimensional convolutional neural network. Background technique [0002] With the development of computer technology, the combination of CT three-dimensional reconstruction technology and 3D printing technology can reconstruct the prosthesis faster and more accurately according to the shape and structure of the skull defect, which provides support and convenience for repair surgery, and is the first choice for cranial repair surgery. A revolutionary advance. Doctors use high-precision CT to scan and three-dimensionally reconstruct the patient's skull to determine the shape and size of the patient's skull defect, and then use 3D printing technology to make a patch for the damaged part, which fits perfectly after embedding, with a perfect shape and good tissue compatibility; The mechanical properties are close ...

AI Technical Summary

Problems solved by technology

[0003] The current mirror image method commonly used in the design of prostheses is only applicable to the case where the defect is on the side. If the defect is located on the front or top, it is necessary to use the Bezier curve as a reference to manually draw the approximate curved surface of the defective skull.

That is, existing methods are not suitable for all situations, and the design process still requires human intervention

Images