scholarly journals Detection of Tibiofemoral Joint Injury in High-Impact Motion Based on Neural Network Reconstruction Algorithm

2021 ◽  
Vol 2021 ◽  
pp. 1-6
Author(s):  
Hongbo Zheng
Keyword(s):  
Neural Network ◽  
3D Model ◽  
Reconstruction Algorithm ◽  
Network Reconstruction ◽  
High Impact ◽  
Joint Injury ◽  
Impact Action ◽  
Tibiofemoral Joint ◽  
Straight Position ◽  
Impact Motion

In order to reduce the damage degree of joint bones, ligaments, and soft tissues caused by the high impact on the tibiofemoral joint during landing, a method for detecting the damage of tibiofemoral joint under high-impact action based on neural network reconstruction algorithm is proposed. Two dimensional X-ray images of knee joints from straightening to bending in 10 healthy volunteers were selected. CT scans were performed on the knee joint on the same side, and the 3D model from the acquired images was reconstructed. The kinematics data of the femur relative to the tibia with full degree of freedom were measured by registering the 3D model with 2D images. The results showed that in the extended position, the femur was rotated inward (5.5° ± 6.3°) relative to the tibia. The range of femoral external rotation is (18.7° ± 5.9°) from flexion to 90° in straight position. However, from 90° to 120°, a small amount of internal rotation occurred (1.4° ± 1.9°), so during the whole flexion process, the femur rotated (17.3° ± 6.9°), among which, from the straight position to 15°, the femur rotated (10.0° ± 5.6°). Damage in different areas is determined by the size of the interlayer displacement sample size method of sample space reduction. It is proved that the detection method of tibiofemoral joint injury in high-impact motion based on neural network reconstruction algorithm has high accuracy and consistency.

Multi-Scale Convolutional Neural Network Reconstruction Algorithm Based on Edge Correction

2018 ◽  
Vol 55 (9) ◽  
pp. 091003
Author(s):  
程德强 Cheng Deqiang ◽  
蔡迎春 Cai Yingchun ◽  
陈亮亮 Chen Liangliang ◽  
宋玉龙 Song Yulong
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Reconstruction Algorithm ◽  
Network Reconstruction ◽  
Multi Scale

scholarly journals Accelerating Quantitative Susceptibility and R2* Mapping using Incoherent Undersampling and Deep Neural Network Reconstruction

NeuroImage ◽  
2021 ◽  
pp. 118404
Author(s):  
Yang Gao ◽  
Martijn Cloos ◽  
Feng Liu ◽  
Stuart Crozier ◽  
G. Bruce Pike ◽  
...  
Keyword(s):  
Neural Network ◽  
Deep Neural Network ◽  
Network Reconstruction

Precise Measurement of Physical Activities and High Impact Motion: Feasibility of Smart Activity Sensor System

2020 ◽  
pp. 1-1
Author(s):  
Hung-Ping Liu ◽  
Yu-Min Chuang ◽  
Chih-Hao Liu ◽  
Phillip C. Yang ◽  
Chiou-Shann Fuh
Keyword(s):  
Precise Measurement ◽  
Physical Activities ◽  
High Impact ◽  
Sensor System ◽  
Impact Motion

scholarly journals An Efficient 3D Model Retrieval Method Based on Convolutional Neural Network

Complexity ◽  
2020 ◽  
Vol 2020 ◽  
pp. 1-14
Author(s):  
Bo Ding ◽  
Lei Tang ◽  
Yong-jun He
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
3D Model ◽  
3D Models ◽  
3D Model Retrieval ◽  
Similarity Matching ◽  
Retrieval Method ◽  
Model Classification ◽  
Model Retrieval ◽  
Index Building

Recently, 3D model retrieval based on views has become a research hotspot. In this method, 3D models are represented as a collection of 2D projective views, which allows deep learning techniques to be used for 3D model classification and retrieval. However, current methods need improvements in both accuracy and efficiency. To solve these problems, we propose a new 3D model retrieval method, which includes index building and model retrieval. In the index building stage, 3D models in library are projected to generate a large number of views, and then representative views are selected and input into a well-learned convolutional neural network (CNN) to extract features. Next, the features are organized according to their labels to build indexes. In this stage, the views used for representing 3D models are reduced substantially on the premise of keeping enough information of 3D models. This method reduces the number of similarity matching by 87.8%. In retrieval, the 2D views of the input model are classified into a category with the CNN and voting algorithm, and then only the features of one category rather than all categories are chosen to perform similarity matching. In this way, the searching space for retrieval is reduced. In addition, the number of used views for retrieval is gradually increased. Once there is enough evidence to determine a 3D model, the retrieval process will be terminated ahead of time. The variable view matching method further reduces the number of similarity matching by 21.4%. Experiments on the rigid 3D model datasets ModelNet10 and ModelNet40 and the nonrigid 3D model dataset McGill10 show that the proposed method has achieved retrieval accuracy rates of 94%, 92%, and 100%, respectively.

Sign in / Sign up

Export Citation Format

Share Document