China 
Africa 
Explore chapters and articles related to this topic
Emergence of Groupwise Registration in MR Brain Study
Published in Hualou Liang, Joseph D. Bronzino, Donald R. Peterson, Biosignal Processing, 2012
Guorong Wu, Hongjun Jia, Qian Wang, Feng Shi, Pew-Thian Yap, Dinggang Shen
It is generally difficult to achieve good registration by directly registering each image to a fixed population center, especially when cross-subject anatomical variation is large. Some algorithms have been proposed to warp the individual images with the help of intermediate template [47–50]. These intermediate templates are produced to pave the path for connecting an individual image to the population center. The final registration result can be obtained by deforming each individual image along its respective path to the population center. This idea is applied to the groupwise registration by building a minimum spanning tree (MST) [51] where each node corresponds to one image and each edge indicates the distance between two connected nodes. The root node of the MST, namely the population center, can be determined by selecting a node that has the minimal edge length to all other nodes or that has the maximal number of children. In Hamm et al. [27], after learning the intrinsic manifold of the whole dataset, the population center is determined as the pseudo-geodesic median image since it minimizes the total path length from each image to the template. The corresponding geodesic paths between individual images and the population center are computed to construct a tree based on the learned manifold. The large deformation between the subject and the population center is thus decomposed into several small ones, and the accuracy of registration is improved. Nevertheless, since the population center is approximated by a fixed image (i.e., the root image) from the dataset, the bias is unavoidable in this scenario.
A three-dimensional measurement based on CT for the posterior tilt with ideal inter-and intra-observer reliability in non-displaced femoral neck fractures
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2021
Shenghui Wu, Wei Wang, Binbin Zhang, Haowei Zhang, Xinsheng Xu, Guangyi Li, Huipeng Shi, Jiong Mei
The 2D method is less cumbersome and time-consuming to use in clinical practice than the 3D way. Nonetheless, our proposed method is more realistic and accurate, thus it can be relied upon to describe the displacement of the femoral head in non-displaced FNFs. Although the 2D measurement of posterior tilt of the femoral head is a simple tool, it has been suggested that the angle based on X-ray deviates from the real value due to different quality of imaging or anatomical variation (Hoelsbrekken and Dolatowski 2017). Our results showed that there were only five cases (10%) with similar angles and discrepancies within 1°. Moreover, there were 30 cases (60%) with a lower angle in the 2D method than in 3D measurement (Figure 6). Therefore, in most cases, the precise degree of displacement measured by 3D method might be blurred by the 2D method. This difference may be the reason for the discrepancy of correlation between the posterior tilt and prognosis for non-displaced FNFs in previous studies. The 3D method, as proposed by our study, is relatively complicated and requires trained skills to operate. In regards to this, all the study observers individually practiced this novel measurement on ten separate CTs in two weeks before starting the study. The almost perfect inter-and intra-reliability in the 3D method as revealed in our study demonstrates the method's feasibility. Of note, there is enough time for orthopedic surgeons to assess the displacement using CT, during preoperative examinations like electrocardiograms and echocardiographic.