A Multimodal Image Registration Method Combining Gradient Orientation Mutual Information with Multi-resolution Hybrid Optimization Algorithm

2010 ◽  
Vol 39 (8) ◽  
pp. 1359-1366
Author(s):  
凌志刚 LING Zhi-gang ◽  
潘泉 PAN quan ◽  
程咏梅 CHENG Yong-mei ◽  
张绍武 ZHANG Shao-wu ◽  
李耀军 LI Yao-jun
2020 ◽  
Author(s):  
Nailong Jia ◽  
Long Fan ◽  
Chuanzi Li ◽  
Zhongshi Nie ◽  
Suihuang Wang ◽  
...  

BACKGROUND Background: At present, the incidence of diabetes is on the rise. When doctors diagnose and treat patients' diseases, they often need to image patients to provide complementary information on patient anatomy and functional metabolism. OBJECTIVE Objective: The aim was to understand the morphological features of peripheral blood vessels of diabetes more accurately and explore its Risk factors for the occurrence of lesions for early diagnosis and early prevention. METHODS Methods: The paper selected subclinical diabetes patients admitted to our hospital from October 2013 to October 2018 as a research object. After performing colour Doppler ultrasonography on peripheral blood vessels, images of ultrasound images were taken. Then the paper proposes a multi-mode medical image registration method based on hybrid optimization algorithm for the multi-extreme problem of mutual information function. Mutual information is used as the similarity measure. The hybrid optimization algorithm is used to search for the best registration exchange parameters. The quasi-colour super images are exchanged for registration purposes. RESULTS Results: The experimental results show that the hybrid optimization algorithm can accurately analyse the colour ultrasound image of the peripheral blood vessels of subclinical diabetes, avoiding falling into the local optimal value, and the accuracy of the registration result reaches the sub-pixel level. CONCLUSIONS Conclusion: With the rapid development of imaging technology, the increasing image resolution, and the increasing amount of image data, parallel performance is high. The quasi-method has a very important significance for multi-modal medical image registration. The parameters in this algorithm can be further optimized. CLINICALTRIAL


2002 ◽  
Vol 124 (2) ◽  
pp. 278-285 ◽  
Author(s):  
Gang Liu ◽  
Zhongqin Lin ◽  
Youxia Bao

In the tooling design of autobody cover panels, design of drawbead will affect the distribution of drawing restraining force along mouth of dies and the relative flowing velocity of the blank, and consequently, will affect the distributions of strain and thickness in a formed part. Therefore, reasonable design of drawbead is the key point of cover panels’ forming quality. An optimization design method of drawbead, using one improved hybrid optimization algorithm combined with FEM software, is proposed in this paper. First, we used this method to design the distribution of drawbead restraining force along the mouth of a die, then the actual type and geometrical parameters of drawbead could be obtained according to an improved drawbead restraining force model and the improved hybrid optimization algorithm. This optimization method of drawbead was used in designing drawing tools of an actual autobody cover panel, and an optimized drawbead design plan has been obtained, by which deformation redundancy was increased from 0% under uniform drawbead control to 10%. Plastic strain of all area of formed part was larger than 2% and the minimum flange width was larger than 10 mm. Therefore, not only better formability and high dent resistance were obtained, but also fine cutting contour line and high assembly quality could be obtained. An actual drawing part has been formed using the optimized drawbead, and the experimental results were compared with the simulating results in order to verify the validity of the optimized design plan. Good agreement of thickness on critical areas between experimental results and simulation results proves that the optimization design method of drawbead could be successfully applied in designing actual tools of autobody cover panels.


Sign in / Sign up

Export Citation Format

Share Document