Three-dimensional spine reconstruction from biplane radiographs using convolutional neural networks

被引：0

作者：

Li, Bo ^{[1
]}

Zhang, Junhua ^{[1
]}

Wang, Qian ^{[1
]}

Li, Hongjian ^{[2
]}

Wang, Qiyang ^{[2
]}

机构：

[1] Yunnan Univ, Dept Elect Engn, Kunming, Peoples R China

[2] First Peoples Hosp Yunnan Prov, Kunming, Peoples R China

来源：

MEDICAL ENGINEERING & PHYSICS | 2024年 / 123卷

关键词：

Three-dimensional reconstruction; Scoliosis; Feature extraction; Deep learning; Convolutional neural networks; STATISTICAL SHAPE MODEL; HIP OSTEOARTHRITIS; APPEARANCE MODELS; FEMUR; FRAMEWORK;

D O I：

10.1016/j.medengphy.2023.104088

中图分类号：

R318 [生物医学工程];

学科分类号：

0831 ;

摘要：

Purpose: The purpose of this study was to develop and evaluate a deep learning network for three-dimensional reconstruction of the spine from biplanar radiographs.Methods: The proposed approach focused on extracting similar features and multiscale features of bone tissue in biplanar radiographs. Bone tissue features were reconstructed for feature representation across dimensions to generate three-dimensional volumes. The number of feature mappings was gradually reduced in the recon-struction to transform the high-dimensional features into the three-dimensional image domain. We produced and made eight public datasets to train and test the proposed network. Two evaluation metrics were proposed and combined with four classical evaluation metrics to measure the performance of the method. Results: In comparative experiments, the reconstruction results of this method achieved a Hausdorff distance of 1.85 mm, a surface overlap of 0.2 mm, a volume overlap of 0.9664, and an offset distance of only 0.21 mm from the vertebral body centroid. The results of this study indicate that the proposed method is reliable.

引用

页数：11

共 60 条

[1] Bone surface reconstruction and clinical features estimation from sparse landmarks and Statistical Shape Models: a feasibility study on the femur [J].

Asvadi, Alireza ;

Dardenne, Guillaume ;

Troccaz, Jocelyne ;

Burdin, Valerie .

MEDICAL ENGINEERING & PHYSICS, 2021, 95 :30-38

[2] Toward Automated 3D Spine Reconstruction from Biplanar Radiographs Using CNN for Statistical Spine Model Fitting [J].

Aubert, B. ;

Vazquez, C. ;

Cresson, T. ;

Parent, S. ;

de Guise, J. A. .

IEEE TRANSACTIONS ON MEDICAL IMAGING, 2019, 38 (12) :2796-2806

[3] Statistical Shape Model-Based Femur Kinematics From Biplane Fluoroscopy [J].

Baka, N. ;

de Bruijne, M. ;

van Walsum, T. ;

Kaptein, B. L. ;

Giphart, J. E. ;

Schaap, M. ;

Niessen, W. J. ;

Lelieveldt, B. P. F. .

IEEE TRANSACTIONS ON MEDICAL IMAGING, 2012, 31 (08) :1573-1583

[4] 2D-3D shape reconstruction of the distal femur from stereo X-ray imaging using statistical shape models [J].