Learning U-Net Based Multi-Scale Features in Encoding-Decoding for MR Image Brain Tissue Segmentation.

Accurate brain tissue segmentation of MRI is vital to diagnosis aiding, treatment planning, and neurologic condition monitoring. As an excellent convolutional neural network (CNN), U-Net is widely used in MR image segmentation as it usually generates high-precision features. However, the performance of U-Net is considerably restricted due to the variable shapes of the segmented targets in MRI and the information loss of down-sampling and up-sampling operations. Therefore, we propose a novel network by introducing spatial and channel dimensions-based multi-scale feature information extractors into its encoding-decoding framework, which is helpful in extracting rich multi-scale features while highlighting the details of higher-level features in the encoding part, and recovering the corresponding localization to a higher resolution layer in the decoding part. Concretely, we propose two information extractors, multi-branch pooling, called MP, in the encoding part, and multi-branch dense prediction, called MDP, in the decoding part, to extract multi-scale features. Additionally, we designed a new multi-branch output structure with MDP in the decoding part to form more accurate edge-preserving predicting maps by integrating the dense adjacent prediction features at different scales. Finally, the proposed method is tested on datasets MRbrainS13, IBSR18, and ISeg2017. We find that the proposed network performs higher accuracy in segmenting MRI brain tissues and it is better than the leading method of 2018 at the segmentation of GM and CSF. Therefore, it can be a useful tool for diagnostic applications, such as brain MRI segmentation and diagnosing.

[Preliminary construction of three-dimensional visual educational system for clinical dentistry based on world wide web webpage].

OBJECTIVE: To establish a new visual educational system of virtual reality for clinical dentistry based on world wide web (WWW) webpage in order to provide more three-dimensional multimedia resources to dental students and an online three-dimensional consulting system for patients. METHODS: Based on computer graphics and three-dimensional webpage technologies, the software of 3Dsmax and Webmax were adopted in the system development. In the Windows environment, the architecture of whole system was established step by step, including three-dimensional model construction, three-dimensional scene setup, transplanting three-dimensional scene into webpage, reediting the virtual scene, realization of interactions within the webpage, initial test, and necessary adjustment. RESULTS: Five cases of three-dimensional interactive webpage for clinical dentistry were completed. The three-dimensional interactive webpage could be accessible through web browser on personal computer, and users could interact with the webpage through rotating, panning and zooming the virtual scene. CONCLUSIONS: It is technically feasible to implement the visual educational system of virtual reality for clinical dentistry based on WWW webpage. Information related to clinical dentistry can be transmitted properly, visually and interactively through three-dimensional webpage.

[The application of artificial neural networks to predict individual risk of essential hypertension].

OBJECTIVE: To establish models to predict individual risk of essential hypertension and to evaluate and explore new forecasting methods. METHODS: To select data of 3054 community residents from an epidemiological survey and divided them into 4:1 (2438 cases and 616 cases) ratio in accordance with the balance of age and sex to filter variables, and to establish, test and evaluate the prediction models. Using artificial neural network (ANN) and logistic regression analysis to establish models while applying ROC to evaluate the prediction models. RESULTS: Forecast results of the models applying to the test set proved that ANN had lower specificity but better veracity and sensitivity than logistic regression. In particular, the Youden's index of the ANN2 came up to 0.8399 which was distinctly higher than the other two models. When the area was under the ROC curve of logistic regression, the ANN, and ANN2 models equaled to 0.732 +/- 0.026, 0.900 +/- 0.014 and 0.918 +/- 0.013 respectively, which proved that the ANN model was better in the prediction about individual health risk of essential hypertension. CONCLUSION: Our results showed that ANN method seemed better than logistic regression in terms of predicting the individual risk from hypertension thus supplied a new method to solve the forecast of individual risk.