Application of artificial neural network technologies to vertebral segmentation according on CT data

摘要

The method of 2D segmentation of vertebrae based on computer tomography data (DICOM files) in sagittal projectionusing artificial neural network Mask-RCNNN is considered in this study. The effectiveness of network recognition wascompared with manual segmentation performed by a professional physician. The comparison of accuracy between theneural network and manual segmentation was evaluated using the S?rens coefficient. The result of automatic 2Dsegmentation has been tested by professional physicians. The application of the method makes it possible to significantlyspeed up the process of modeling bone structures of the spine in 2D mode to solve the problems of biomechanics.