Limits...
The Architecture of an Automatic eHealth Platform With Mobile Client for Cerebrovascular Disease Detection.

Wang X, Bie R, Sun Y, Wu Z, Zhou M, Cao R, Xie L, Zhang D - JMIR Mhealth Uhealth (2013)

Bottom Line: The anisotropic diffusion model was used to reduce the noise.The application results have also been validated by our neurosurgeon and radiologist.The long-term benefits and additional applications of this technology warrant further study.

View Article: PubMed Central - HTML - PubMed

Affiliation: College of Information Science and Technology, Beijing Normal University, Beijing, China.

ABSTRACT

Background: In recent years, cerebrovascular disease has been the leading cause of death and adult disability in the world. This study describes an efficient approach to detect cerebrovascular disease.

Objective: In order to improve cerebrovascular treatment, prevention, and care, an automatic cerebrovascular disease detection eHealth platform is designed and studied.

Methods: We designed an automatic eHealth platform for cerebrovascular disease detection with a four-level architecture: object control layer, data transmission layer, service supporting layer, and application service layer. The platform has eight main functions: cerebrovascular database management, preprocessing of cerebral image data, image viewing and adjustment model, image cropping compression and measurement, cerebrovascular segmentation, 3-dimensional cerebrovascular reconstruction, cerebrovascular rendering, cerebrovascular virtual endoscope, and automatic detection. Several key technologies were employed for the implementation of the platform. The anisotropic diffusion model was used to reduce the noise. Statistics segmentation with Gaussian-Markov random field model (G-MRF) and Stochastic Estimation Maximization (SEM) parameter estimation method were used to realize the cerebrovascular segmentation. Ball B-Spline curve was proposed to model the cerebral blood vessels. Compute unified device architecture (CUDA) based on ray-casting volume rendering presented by curvature enhancement and boundary enhancement were used to realize the volume rendering model. We implemented the platform with a network client and mobile phone client to fit different users.

Results: The implemented platform is running on a common personal computer. Experiments on 32 patients' brain computed tomography data or brain magnetic resonance imaging data stored in the system verified the feasibility and validity of each model we proposed. The platform is partly used in the cranial nerve surgery of the First Hospital Affiliated to the General Hospital of People's Liberation Army and radiology of Beijing Navy General Hospital. At the same time it also gets some applications in medical imaging specialty teaching of Tianjin Medical University. The application results have also been validated by our neurosurgeon and radiologist.

Conclusions: The platform appears beneficial in diagnosis of the cerebrovascular disease. The long-term benefits and additional applications of this technology warrant further study. The research built a diagnosis and treatment platform of the human tissue with complex geometry and topology such as brain vessel based on the Internet of things.

No MeSH data available.


Related in: MedlinePlus

Comparison of G - MRF and double Gaussian model segmentation result. Left: G - MRF segmentation result; right: double Gaussian model segmentation result.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4114469&req=5

figure7: Comparison of G - MRF and double Gaussian model segmentation result. Left: G - MRF segmentation result; right: double Gaussian model segmentation result.

Mentions: An automatic statistical approach based on the Gaussian-Markov Random Field Model (G-MRF) is employed. The voxels are classified as either blood vessels or background noise by a finite mixture of two Gaussian distributions. 3D MRF is employed to improve precision and those parameters are estimated by the Stochastic Estimation Maximization (SEM) algorithm, which converges to the true likelihood under a large lattice. The MRF field embeds the spatial neighborhood information to the parameters statistical model and increases regional morphology information in the gray statistic information. With this method, more three-level brain vessels can be achieved. The cerebral vascular segmentation algorithm based on SEM hybrid model solves the traditional slow convergence and local minimum problem of expectation maximum (EM) algorithm. From Figures 7 and 8, compared with double Gauss model [22], our algorithm can effectively segment the main branch and the surrounding smaller branches of brain vessel, and its convergence speed improves greatly than the traditional EM algorithm.


The Architecture of an Automatic eHealth Platform With Mobile Client for Cerebrovascular Disease Detection.

Wang X, Bie R, Sun Y, Wu Z, Zhou M, Cao R, Xie L, Zhang D - JMIR Mhealth Uhealth (2013)

Comparison of G - MRF and double Gaussian model segmentation result. Left: G - MRF segmentation result; right: double Gaussian model segmentation result.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4114469&req=5

figure7: Comparison of G - MRF and double Gaussian model segmentation result. Left: G - MRF segmentation result; right: double Gaussian model segmentation result.
Mentions: An automatic statistical approach based on the Gaussian-Markov Random Field Model (G-MRF) is employed. The voxels are classified as either blood vessels or background noise by a finite mixture of two Gaussian distributions. 3D MRF is employed to improve precision and those parameters are estimated by the Stochastic Estimation Maximization (SEM) algorithm, which converges to the true likelihood under a large lattice. The MRF field embeds the spatial neighborhood information to the parameters statistical model and increases regional morphology information in the gray statistic information. With this method, more three-level brain vessels can be achieved. The cerebral vascular segmentation algorithm based on SEM hybrid model solves the traditional slow convergence and local minimum problem of expectation maximum (EM) algorithm. From Figures 7 and 8, compared with double Gauss model [22], our algorithm can effectively segment the main branch and the surrounding smaller branches of brain vessel, and its convergence speed improves greatly than the traditional EM algorithm.

Bottom Line: The anisotropic diffusion model was used to reduce the noise.The application results have also been validated by our neurosurgeon and radiologist.The long-term benefits and additional applications of this technology warrant further study.

View Article: PubMed Central - HTML - PubMed

Affiliation: College of Information Science and Technology, Beijing Normal University, Beijing, China.

ABSTRACT

Background: In recent years, cerebrovascular disease has been the leading cause of death and adult disability in the world. This study describes an efficient approach to detect cerebrovascular disease.

Objective: In order to improve cerebrovascular treatment, prevention, and care, an automatic cerebrovascular disease detection eHealth platform is designed and studied.

Methods: We designed an automatic eHealth platform for cerebrovascular disease detection with a four-level architecture: object control layer, data transmission layer, service supporting layer, and application service layer. The platform has eight main functions: cerebrovascular database management, preprocessing of cerebral image data, image viewing and adjustment model, image cropping compression and measurement, cerebrovascular segmentation, 3-dimensional cerebrovascular reconstruction, cerebrovascular rendering, cerebrovascular virtual endoscope, and automatic detection. Several key technologies were employed for the implementation of the platform. The anisotropic diffusion model was used to reduce the noise. Statistics segmentation with Gaussian-Markov random field model (G-MRF) and Stochastic Estimation Maximization (SEM) parameter estimation method were used to realize the cerebrovascular segmentation. Ball B-Spline curve was proposed to model the cerebral blood vessels. Compute unified device architecture (CUDA) based on ray-casting volume rendering presented by curvature enhancement and boundary enhancement were used to realize the volume rendering model. We implemented the platform with a network client and mobile phone client to fit different users.

Results: The implemented platform is running on a common personal computer. Experiments on 32 patients' brain computed tomography data or brain magnetic resonance imaging data stored in the system verified the feasibility and validity of each model we proposed. The platform is partly used in the cranial nerve surgery of the First Hospital Affiliated to the General Hospital of People's Liberation Army and radiology of Beijing Navy General Hospital. At the same time it also gets some applications in medical imaging specialty teaching of Tianjin Medical University. The application results have also been validated by our neurosurgeon and radiologist.

Conclusions: The platform appears beneficial in diagnosis of the cerebrovascular disease. The long-term benefits and additional applications of this technology warrant further study. The research built a diagnosis and treatment platform of the human tissue with complex geometry and topology such as brain vessel based on the Internet of things.

No MeSH data available.


Related in: MedlinePlus