A novel deep-learning based weighted feature fusion architecture for precise classification of pressure injury.

Introduction: Precise classification has an important role in treatment of pressure injury (PI), while current machine-learning or deeplearning based methods of PI classification remain low accuracy. Methods: In this study, we developed a deeplearning based weighted feature fusion architecture for fine-grained classification, which combines a top-down and bottom-up pathway to fuse high-level semantic information and low-level detail representation. We validated it in our established database that consist of 1,519 images from multi-center clinical cohorts. ResNeXt was set as the backbone network. Results: We increased the accuracy of stage 3 PI from 60.3% to 76.2% by adding weighted feature pyramid network (wFPN). The accuracy for stage 1, 2, 4 PI were 0.870, 0.788, and 0.845 respectively. We found the overall accuracy, precision, recall, and F1-score of our network were 0.815, 0.808, 0.816, and 0.811 respectively. The area under the receiver operating characteristic curve was 0.940. Conclusions: Compared with current reported study, our network significantly increased the overall accuracy from 75% to 81.5% and showed great performance in predicting each stage. Upon further validation, our study will pave the path to the clinical application of our network in PI management.

RND1 inhibits epithelial-mesenchymal transition and temozolomide resistance of glioblastoma via AKT/GSK3-ß pathway.

GBM is one of the most malignant tumor in central nervous system. The resistance to temozolomide (TMZ) is inevitable in GBM and the characterization of TMZ resistance seriously hinders clinical treatment. It is worthwhile exploring the underlying mechanism of aggressive invasion and TMZ resistance in GBM treatment. Bioinformatic analysis was used to analyze the association between RND1 and a series of EMT-related genes. Colony formation assay and cell viability assay were used to assess the growth of U87 and U251 cells. The cell invasion status was evaluated based on transwell and wound-healing assays. Western blot was used to detect the protein expression in GBM cells. Treatment targeted RND1 combined with TMZ therapy was conducted in nude mice to evaluate the potential application of RND1 as a clinical target for GBM. The overexpression of RND1 suppressed the progression and migration of U87 and U251 cells. RND1 knockdown facilitated the growth and invasion of GBM cells. RND1 regulated the EMT of GBM cells via inhibiting the phosphorylation of AKT and GSK3-ß. The promoted effects of RND1 on TMZ sensitivity was identified both in vitro and in vivo. This research demonstrated that the overexpression of RND1 suppressed the migration and EMT status by downregulating AKT/GSK3-ß pathway in GBM. RND1 enhanced the TMZ sensitivity of GBM cells both in vitro and in vivo. Our findings may contribute to the targeted therapy for GBM and the understanding of mechanisms of TMZ resistance in GBM.

Tumor Immune Microenvironment Landscape in Glioma Identifies a Prognostic and Immunotherapeutic Signature.

The tumor immune microenvironment (TIME) has been recognized to be associated with sensitivity to immunotherapy and patient prognosis. Recent research demonstrates that assessing the TIME patterns on large-scale samples will expand insights into TIME and will provide guidance to formulate immunotherapy strategies for tumors. However, until now, thorough research has not yet been reported on the immune infiltration landscape of glioma. Herein, the CIBERSORT algorithm was used to unveil the TIME landscape of 1,975 glioma observations. Three TIME subtypes were established, and the TIMEscore was calculated by least absolute shrinkage and selection operator (LASSO)-Cox analysis. The high TIMEscore was distinguished by an elevated tumor mutation burden (TMB) and activation of immune-related biological process, such as IL6-JAK-STAT3 signaling and interferon gamma (IFN-γ) response, which may demonstrate that the patients with high TIMEscore were more sensitive to immunotherapy. Multivariate analysis revealed that the TIMEscore could strongly and independently predict the prognosis of gliomas [Chinese Glioma Genome Atlas (CGGA) cohort: hazard ratio (HR): 2.134, p < 0.001; Gravendeel cohort: HR: 1.872, p < 0.001; Kamoun cohort: HR: 1.705, p < 0.001; The Cancer Genome Atlas (TCGA) cohort: HR: 2.033, p < 0.001; the combined cohort: HR: 1.626, p < 0.001], and survival advantage was evident among those who received chemotherapy. Finally, we validated the performance of the signature in human tissues from Wuhan University (WHU) dataset (HR: 15.090, p = 0.008). Our research suggested that the TIMEscore could be applied as an effective predictor for adjuvant therapy and prognosis assessment.

[Research on algorithms of uterine contraction curve analysis and its real-time status identification].

Identification of real-time uterine contraction status is very significant to labor analgesia, but the traditional uterine contraction analysis algorithms and systems cannot meet the requirement. According to the situations mentioned above, this paper designs a set of algorithms for the real-time analysis of uterine contraction status. The algorithms include uterine contraction signal preprocessing, uterine contraction baseline extraction based on histogram and linear iteration and an algorithm for the real-time analysis of uterine contraction status based on finite state machines theory. It uses the last uterine status and a series of state transfer conditions to identify the current uterine contraction status, as well as a buffer mechanism to avoid false status transitions. To evaluate the performance of the algorithm, we compare it with an existing uterine contraction analysis algorithm used in the electronic fetal monitor. The experiments show that our algorithm can analyze the uterine contraction status while monitoring the uterine contraction signal in a real-time. Its sensitivity reaches 0.939 9 and its positive predictive value is 0.869 3, suggesting that the algorithm has high accuracy and meets the need of clinical monitoring.