Algorithm for Predicting Respiratory Motion of Liver Tissue Based on Short-Term Respiratory Monitoring.

In this study, an algorithm for predicting respiratory motion of liver tissue based on the combination of subject-specific external surrogate signals and 2D ultrasound image sequences was investigated to achieve better respiratory monitoring in clinical procedures. To achieve non-invasiveness in clinical procedures, an EM position tracker and a Doppler ultrasound diagnostic system were used as data collectors. Firstly, the mapping relationship between the magnetic sensing surrogate signal and the internal motion of liver tissue was learned by the Ridge regression model to achieve the estimation of the internal motion of liver tissue by the magnetic sensing surrogate signal; then the motion prediction of the estimated internal motion of liver tissue was performed by the artificial neural network (ANN) as the prediction filter; finally, the prediction of the respiratory motion of liver tissue by the magnetic sensing surrogate signal was achieved. Through experimental tests on 16 subject volunteers, the experimental results show that the RMSE of the proposed algorithm for predicting the respiratory motion of liver tissue is 2mm, indicating the potential of this prediction algorithm to achieve the localization of the internal motion position of liver tissue by the human magnetic sensing surrogate signal.

Combination of UNet++ and ResNeSt to classify chronic inflammation of the choledochal cystic wall in patients with pancreaticobiliary maljunction.

OBJECTIVES: The aim of this study was to establish an automatic classification model for chronic inflammation of the choledoch wall using deep learning with CT images in patients with pancreaticobiliary maljunction (PBM). METHODS: CT images were obtained from 76 PBM patients, including 61 cases assigned to the training set and 15 cases assigned to the testing set. The region of interest (ROI) containing the choledochal lesion was extracted and segmented using the UNet++ network. The degree of severity of inflammation in the choledochal wall was initially classified using the ResNeSt network. The final classification result was determined per decision rules. Grad-CAM was used to explain the association between the classification basis of the network and clinical diagnosis. RESULTS: Segmentation of the lesion on the common bile duct wall was roughly obtained with the UNet++ segmentation model and the average value of Dice coefficient of the segmentation model in the testing set was 0.839 ± 0.150, which was verified through fivefold cross-validation. Inflammation was initially classified with ResNeSt18, which resulted in accuracy = 0.756, sensitivity = 0.611, specificity = 0.852, precision = 0.733, and area under curve (AUC) = 0.711. The final classification sensitivity was 0.8. Grad-CAM revealed similar distribution of inflammation of the choledochal wall and verified the inflammation classification. CONCLUSIONS: By combining the UNet++ network and the ResNeSt network, we achieved automatic classification of chronic inflammation of the choledoch in PBM patients and verified the robustness through cross-validation performed five times. This study provided an important basis for classification of inflammation severity of the choledoch in PBM patients. ADVANCES IN KNOWLEDGE: We combined the UNet++ network and the ResNeSt network to achieve automatic classification of chronic inflammation of the choledoch in PBM. These results provided an important basis for classification of choledochal inflammation in PBM and for surgical therapy.

Machine learning-based quantitative analysis of barium enema and clinical features for early diagnosis of short-segment Hirschsprung disease in neonate.

OBJECTIVE: To develop a mathematical model based on a combination of clinical and radiologic features (barium enema) for early diagnosis of short-segment Hirschsprung disease (SHSCR) in neonate. METHODS: The analysis included 54 neonates with biopsy-confirmed SHSCR (the cases) and 59 neonates undergoing barium enema for abdominal symptoms but no Hirschsprung disease (the control). Colon shape features extracted from barium enema images and clinical features were used to develop diagnostic models using support vector machine (SVM) and L2-regularized logistic regression (LR). The training cohort included 32 cases and 37 controls; testing cohort consisted 22 cases and 22 controls. Results were compared to interpretation by 2 radiologists. RESULTS: In the analysis by radiologists, 87 out of 113 cases were correctly classified. Six SHSCR cases were mis-classified into the non-HSCR group. In the remaining 20 cases, radiologists were unable to make a decision. Both the SVM and LR classifiers contained five clinical features and four shape features. The performance of the two classifiers was similar. The best model had 86.36% accuracy, 81.82% sensitivity, and 90.91% specificity. The AUC was 0.9132 for the best-performing SVM classifier and 0.9318 for the best-performing LR classifier. CONCLUSION: A combination of clinical features and colon shape features extracted from barium enemas can be used to improve early diagnosis of SHSCR in neonate.

Retinal image graph-cut segmentation algorithm using multiscale Hessian-enhancement-based nonlocal mean filter.

We propose a new method to enhance and extract the retinal vessels. First, we employ a multiscale Hessian-based filter to compute the maximum response of vessel likeness function for each pixel. By this step, blood vessels of different widths are significantly enhanced. Then, we adopt a nonlocal mean filter to suppress the noise of enhanced image and maintain the vessel information at the same time. After that, a radial gradient symmetry transformation is adopted to suppress the nonvessel structures. Finally, an accurate graph-cut segmentation step is performed using the result of previous symmetry transformation as an initial. We test the proposed approach on the publicly available databases: DRIVE. The experimental results show that our method is quite effective.