Computer-assisted diagnosis is key for scaling up cervical cancer screening. However, current recognition algorithms perform poorly on whole slide image (WSI) analysis, fail to generalize for diverse staining and imaging, and show sub-optimal clinical-level verification. Here, we develop a progressive lesion cell recognition method combining low- and high-resolution WSIs to recommend lesion cells and a recurrent neural network-based WSI classification model to evaluate the lesion degree of WSIs. We train and validate our WSI analysis system on 3,545 patient-wise WSIs with 79,911 annotations from multiple hospitals and several imaging instruments. On multi-center independent test sets of 1,170 patient-wise WSIs, we achieve 93.5% Specificity and 95.1% Sensitivity for classifying slides, comparing favourably to the average performance of three independent cytopathologists, and obtain 88.5% true positive rate for highlighting the top 10 lesion cells on 447 positive slides. After deployment, our system recognizes a one giga-pixel WSI in about 1.5 min.
The purpose of this study was to quantitatively analyze the relationship between three dimensional arterial spin labeling (3D-ASL) and dynamic susceptibility contrast-enhanced perfusion weighted imaging (DSC-PWI) in ischemic stroke patients. Thirty patients with ischemic stroke were included in this study. All subjects underwent routine magnetic resonance imaging scanning, diffusion weighted imaging (DWI), magnetic resonance angiography (MRA), 3D-ASL and DSC-PWI on a 3.0T MR scanner. Regions of interest (ROIs) were drawn on the cerebral blood flow (CBF) maps (derived from ASL) and multi-parametric DSC perfusion maps, and then, the absolute and relative values of ASL-CBF, DSC-derived CBF, and DSC-derived mean transit time (MTT) were calculated. The relationships between ASL and DSC parameters were analyzed using Pearson's correlation analysis. Receiver operative characteristic (ROC) curves were performed to define the thresholds of relative value of ASL-CBF (rASL) that could best predict DSC-CBF reduction and MTT prolongation. Relative ASL better correlated with CBF and MTT in the anterior circulation with the Pearson correlation coefficients (R) values being 0.611 (P<0.001) and-0.610 (P<0.001) respectively. ROC curves demonstrated that when rASL ≤0.585, the sensitivity, specificity and accuracy for predicting ROIs with rCBF<0.9 were 92.3%, 63.6% and 76.6% respectively. When rASL ≤0.952, the sensitivity, specificity and accuracy for predicting ROIs rMTT>1.0 were 75.7%, 89.2% and 87.8% respectively. ASL-CBF map has better linear correlations with DSC-derived parameters (DSC-CBF and MTT) in anterior circulation in ischemic stroke patients. Additionally, when rASL is lower than 0.585, it could predict DSC-CBF decrease with moderate accuracy. If rASL values range from 0.585 to 0.952, we just speculate the prolonged MTT.
This article proposes a model predictive control (MPC) with the discrete space vector modulation (DSVM) to reduce input current ripple of three-level Vienna rectifier. The finite control set MPC (FCS-MPC) applied in power converters has variable switching frequency and needs high sampling frequency for better input current quality. With the proposed approach, more feasible vectors can be used to improve the input current's performance with the DSVM using the virtual vectors. The DSVM-MPC features fixed switching frequency and should implement large numbers of virtual vectors with the calculation burden. The improved DSVM-MPC is proposed to optimise the MPC algorithm by selecting the adjacent feasible vectors according to the reference voltage vector trajectory. The effectiveness of the proposed method is verified by the experimental results in a Vienna rectifier platform, which shows that the proposed method possesses better input current performance compared with the FCS-MPC method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.