The purpose of this study was to determine whether a deep-learning-based assessment system could facilitate preoperative grading of meningioma. This was a retrospective study conducted at two institutions covering 643 patients. The system, designed with a cascade network structure, was developed using deep-learning technology for automatic tumor detection, visual assessment, and grading prediction. Specifically, a modified U-Net convolutional neural network was first established to segment tumor images. Subsequently, the segmentations were introduced into rendering algorithms for spatial reconstruction and another DenseNet convolutional neural network for grading prediction. The trained models were integrated as a system, and the robustness was tested based on its performance on an external dataset from the second institution involving different magnetic resonance imaging platforms. The results showed that the segment model represented a noteworthy performance with dice coefficients of 0.920 ± 0.009 in the validation group. With accurate segmented tumor images, the rendering model delicately reconstructed the tumor body and clearly displayed the important intracranial vessels. The DenseNet model also achieved high accuracy with an area under the curve of 0.918 ± 0.006 and accuracy of 0.901 ± 0.039 when classifying tumors into low-grade and high-grade meningiomas. Moreover, the system exhibited good performance on the external validation dataset.

Background: Associated with increased morbidity and mortality, postoperative pulmonary complications (PPCs) often occur after major abdominal surgery. Diaphragmatic dysfunction is suggested to play an important role in the development of PPCs and diaphragm echodensity can be used as an indicator of diaphragm function. This study aimed to determine whether diaphragm echodensity could predict the occurrence of PPCs in patients after major abdominal surgery.Methods: Diaphragm ultrasound images of patients after major abdominal surgery were collected during spontaneous breathing trials. Echodensity was quantified based on the right-skewed distribution of grayscale values (50th percentile, ED50; 85th percentile, ED85; mean, EDmean). Intra- and inter-analyzer measurement reproducibility was determined. Outcomes including occurrence of PPCs, reintubation rate, duration of ventilation, and length of ICU stay were recorded.Results: Diaphragm echodensity was measured serially in 117 patients. Patients who developed PPCs exhibited a higher ED50 (35.00 vs. 26.00,p < 0.001), higher ED85 (64.00 vs. 55.00,p < 0.001) and higher EDmean (39.32 vs. 33.98,p < 0.001). In ROC curve analysis, the area under the curve of ED50 for predicting PPCs was 0.611. The optimal ED50 cutoff value for predicting the occurrence of PPCs was 36. According to this optimal ED50 cutoff value, patients were further divided into a high-risk group (ED50 > 36, n = 35) and low-risk group (ED50 ≤ 36, n = 82). Compared with the low-risk group, the high-risk group had a higher incidence of PPCs (unadjustedp = 0.003; multivariate-adjustedp < 0.001).Conclusion: Diaphragm echodensity can be feasibly and reproducibly measured in mechanically ventilated patients. The increase in diaphragm echodensity during spontaneous breathing trials was related to an increased risk of PPCs in patients after major abdominal surgery.

(1) Background: Acute muscle inflammation leads to increased sonographic echodensity. We developed a technique to characterize the echodensity of the gastric antrum wall and assess its feasibility in evaluating the severity of acute gastrointestinal injury (AGI); (2) Methods: The B-mode images of the gastric antrum of each enrolled patient were obtained daily by point-of-care ultrasound (POCUS). The 50th percentile, 85th percentile, and mean value of the grayscale distribution according to histogram analysis (ED₅₀, ED₈₅, and ED_mean, respectively) were used to characterize the gastric antrum echodensity. Consistency and correlation analyses were performed to evaluate the feasibility and reproducibility of gastric antrum echodensity measurement. The association of gastric antrum echodensity with the severity of AGI and its ability to predict feeding intolerance (FI) were analyzed; (3) Results: In total, 206 POCUS images of 43 patients were analyzed. The gastric antrum echodensity measurements had sufficient intra- and inter-investigator reliabilities (intraclass correlation coefficient >0.9 for all parameters). The ED₅₀ showed a significant upward trend as AGI severity increased, as well as ED₈₅ and ED_mean (p for trend <0.001, respectively). Patients who experienced FI had a higher ED₅₀ (67.8 vs. 56.1, p = 0.02), ED₈₅ (85.6 vs. 71.2, p = 0.01), and ED_mean (70.3 vs. 57.6, p = 0.01) upon enteral feeding initiation; (4) Conclusions: Measurement of gastric antrum echodensity was technically feasible and reproducible in ventilated patients. Increased gastric antrum echodensity was associated with greater severity of AGI. Patients with higher gastric antrum echodensity upon enteral nutrition initiation via a nasogastric tube were more likely to develop FI.

Hospital acquired thrombocytopenia (HAT) is a common hematological complication after surgery. This research aimed to develop and compare the performance of seven machine learning (ML) algorithms for predicting patients that are at risk of HAT after surgery. We conducted a retrospective cohort study which enrolled adult patients transferred to the intensive care unit (ICU) after surgery in West China Hospital of Sichuan University from January 2016 to December 2018. All subjects were randomly divided into a derivation set (70%) and test set (30%). ten-fold cross-validation was used to estimate the hyperparameters of ML algorithms during the training process in the derivation set. After ML models were developed, the sensitivity, specificity, area under the curve (AUC), and net benefit (decision analysis curve, DCA) were calculated to evaluate the performances of ML models in the test set. A total of 10,369 patients were included and in 1354 (13.1%) HAT occurred. The AUC of all seven ML models exceeded 0.7, the two highest were Gradient Boosting (GB) (0.834, 0.814–0.853, p < 0.001) and Random Forest (RF) (0.828, 0.807–0.848, p < 0.001). There was no difference between GB and RF (0.834 vs. 0.828, p = 0.293); however, these two were better than the remaining five models (p < 0.001). The DCA revealed that all ML models had high net benefits with a threshold probability approximately less than 0.6. In conclusion, we found that ML models constructed by multiple preoperative variables can predict HAT in patients transferred to ICU after surgery, which can improve risk stratification and guide management in clinical practice.

One nutritional challenge in critically ill patients is enteral feeding intolerance (EFI), but current prokinetic agents have uncertain efficacy and safety profiles. We conducted a longitudinal, single-center, retrospective study to evaluate the efficacy and safety of domperidone administered via the feeding tube versus intravenous (IV) metoclopramide among adult patients with EFI. The primary outcome was feeding success, defined as the proportion of patients with average percentage of daily protein prescription >80% of the target dose. The secondary outcomes were safety endpoints. Among 28,814 intensive care unit (ICU) admissions, 552 patients with EFI were included, 38 receiving IV metoclopramide and 514 receiving tube feeding domperidone. The proportion of feeding success in patients receiving tube feeding domperidone and IV metoclopramide was 42.02% and 21.05%, respectively. After 1:2 matching (IV metoclopramide to tube feeding domperidone), the proportion of feeding success was 40.79% in patients receiving tube feeding domperidone. Basically, after matching, there were no differences in any safety endpoints (mortality and length of stay during ICU and hospitalization, organ-support-treatment free days) or adverse events (recurrence of EFI, electrolyte disturbance, abdominal and other symptoms) between the two groups (p > 0.05). A logistic regression analysis in the matched cohort indicated that domperidone administered via the feeding tube was independently associated with feeding success. We found that tube feeding domperidone was efficient in increasing enteral nutrition delivery performance among critically ill adult patients with EFI.