Generative adversarial network-based deep learning approach in classification of retinal conditions with optical coherence tomography images.

PURPOSE: To determine whether a deep learning approach using generative adversarial networks (GANs) is beneficial for the classification of retinal conditions with Optical coherence tomography (OCT) images. METHODS: Our study utilized 84,452 retinal OCT images obtained from a publicly available dataset (Kermany Dataset). Employing GAN, synthetic OCT images are produced to balance classes of retinal disorders. A deep learning classification model is constructed using pretrained deep neural networks (DNNs), and outcomes are evaluated using 2082 images collected from patients who visited the Department of Ophthalmology and the Department of Endocrinology and Metabolism at the Tri-service General Hospital in Taipei from January 2017 to December 2021. RESULTS: The highest classification accuracies accomplished by deep learning machines trained on the unbalanced dataset for its training set, validation set, fivefold cross validation (CV), Kermany test set, and TSGH test set were 97.73%, 96.51%, 97.14%, 99.59%, and 81.03%, respectively. The highest classification accuracies accomplished by deep learning machines trained on the synthesis-balanced dataset for its training set, validation set, fivefold CV, Kermany test set, and TSGH test set were 98.60%, 98.41%, 98.52%, 99.38%, and 84.92%, respectively. In comparing the highest accuracies, deep learning machines trained on the synthesis-balanced dataset outperformed deep learning machines trained on the unbalanced dataset for the training set, validation set, fivefold CV, and TSGH test set. CONCLUSIONS: Overall, deep learning machines on a synthesis-balanced dataset demonstrated to be advantageous over deep learning machines trained on an unbalanced dataset for the classification of retinal conditions.

Using machine learning to determine the correlation between physiological and environmental parameters and the induction of acute mountain sickness.

BACKGROUND: Recent studies on acute mountain sickness (AMS) have used fixed-location and fixed-time measurements of environmental and physiological variable to determine the influence of AMS-associated factors in the human body. This study aims to measure, in real time, environmental conditions and physiological variables of participants in high-altitude regions to develop an AMS risk evaluation model to forecast prospective development of AMS so its onset can be prevented. RESULTS: Thirty-two participants were recruited, namely 25 men and 7 women, and they hiked from Cuifeng Mountain Forest Park parking lot (altitude: 2300 m) to Wuling (altitude: 3275 m). Regression and classification machine learning analyses were performed on physiological and environmental data, and Lake Louise Acute Mountain Sickness Scores (LLS) to establish an algorithm for AMS risk analysis. The individual R2 coefficients of determination between the LLS and the measured altitude, ambient temperature, atmospheric pressure, relative humidity, climbing speed, heart rate, blood oxygen saturation (SpO2), heart rate variability (HRV), were 0.1, 0.23, 0, 0.24, 0, 0.24, 0.27, and 0.35 respectively; incorporating all aforementioned variables, the R2 coefficient is 0.62. The bagged trees classifier achieved favorable classification results, yielding a model sensitivity, specificity, accuracy, and area under receiver operating characteristic curve of 0.999, 0.994, 0.998, and 1, respectively. CONCLUSION: The experiment results indicate the use of machine learning multivariate analysis have higher AMS prediction accuracies than analyses utilizing single varieties. The developed AMS evaluation model can serve as a reference for the future development of wearable devices capable of providing timely warnings of AMS risks to hikers.

Research progress on negative pressure wound therapy with instillation in the treatment of orthopaedic wounds.

Negative pressure wound therapy with instillation (NPWTi) has the dual function of negative pressure sealing drainage and irrigation, which overcomes the disadvantages of NPWT, such as tube obstruction, inability to apply topical medicine, and poor anti-infection ability. NPWTi has been researched extensively and widely used in various types of wounds, and certain effects have been achieved. A series of parameters for NPWTi have not been unified at present, including the flushing fluid option, flushing mode, and treatment period. This paper reviews the research progress of these parameters for NPWTi and their application in the treatment of orthopaedic wounds.

General deep learning model for detecting diabetic retinopathy.

BACKGROUND: Doctors can detect symptoms of diabetic retinopathy (DR) early by using retinal ophthalmoscopy, and they can improve diagnostic efficiency with the assistance of deep learning to select treatments and support personnel workflow. Conventionally, most deep learning methods for DR diagnosis categorize retinal ophthalmoscopy images into training and validation data sets according to the 80/20 rule, and they use the synthetic minority oversampling technique (SMOTE) in data processing (e.g., rotating, scaling, and translating training images) to increase the number of training samples. Oversampling training may lead to overfitting of the training model. Therefore, untrained or unverified images can yield erroneous predictions. Although the accuracy of prediction results is 90%-99%, this overfitting of training data may distort training module variables. RESULTS: This study uses a 2-stage training method to solve the overfitting problem. In the training phase, to build the model, the Learning module 1 used to identify the DR and no-DR. The Learning module 2 on SMOTE synthetic datasets to identify the mild-NPDR, moderate NPDR, severe NPDR and proliferative DR classification. These two modules also used early stopping and data dividing methods to reduce overfitting by oversampling. In the test phase, we use the DIARETDB0, DIARETDB1, eOphtha, MESSIDOR, and DRIVE datasets to evaluate the performance of the training network. The prediction accuracy achieved to 85.38%, 84.27%, 85.75%, 86.73%, and 92.5%. CONCLUSIONS: Based on the experiment, a general deep learning model for detecting DR was developed, and it could be used with all DR databases. We provided a simple method of addressing the imbalance of DR databases, and this method can be used with other medical images.

Circular RNA CDR1as disrupts the p53/MDM2 complex to inhibit Gliomagenesis.

BACKGROUND: Inactivation of the tumor suppressor p53 is critical for pathogenesis of glioma, in particular glioblastoma multiforme (GBM). MDM2, the main negative regulator of p53, binds to and forms a stable complex with p53 to regulate its activity. Hitherto, it is unclear whether the stability of the p53/MDM2 complex is affected by lncRNAs, in particular circular RNAs that are usually abundant and conserved, and frequently implicated in different oncogenic processes. METHODS: RIP-seq and RIP-qPCR assays were performed to determine the most enriched lncRNAs (including circular RNAs) bound by p53, followed by bioinformatic assays to estimate the relevance of their expression with p53 signaling and gliomagenesis. Subsequently, the clinical significance of CDR1as was evaluated in the largest cohort of Chinese glioma patients from CGGA (n = 325), and its expression in human glioma tissues was further evaluated by RNA FISH and RT-qPCR, respectively. Assays combining RNA FISH with protein immunofluorescence were performed to determine co-localization of CDR1as and p53, followed by CHIRP assays to confirm RNA-protein interaction. Immunoblot assays were carried out to evaluate protein expression, p53/MDM2 interaction and p53 ubiquitination in cells in which CDR1as expression was manipulated. After AGO2 or Dicer was knocked-down to inhibit miRNA biogenesis, effects of CDR1as on p53 expression, stability and activity were determined by immunoblot, RT-qPCR and luciferase reporter assays. Meanwhile, impacts of CDR1as on DNA damage were evaluated by flow cytometric assays and immunohistochemistry. Tumorigenicity assays were performed to determine the effects of CDR1as on colony formation, cell proliferation, the cell cycle and apoptosis (in vitro), and on tumor volume/weight and survival of nude mice xenografted with GBM cells (in vivo). RESULTS: CDR1as is found to bind to p53 protein. CDR1as expression decreases with increasing glioma grade and it is a reliable independent predictor of overall survival in glioma, particularly in GBM. Through a mechanism independent of acting as a miRNA sponge, CDR1as stabilizes p53 protein by preventing it from ubiquitination. CDR1as directly interacts with the p53 DBD domain that is essential for MDM2 binding, thus disrupting the p53/MDM2 complex formation. Induced upon DNA damage, CDR1as may preserve p53 function and protect cells from DNA damage. Significantly, CDR1as inhibits tumor growth in vitro and in vivo, but has little impact in cells where p53 is absent or mutated. CONCLUSIONS: Rather than acting as a miRNA sponge, CDR1as functions as a tumor suppressor through binding directly to p53 at its DBD region to restrict MDM2 interaction. Thus, CDR1as binding disrupts the p53/MDM2 complex to prevent p53 from ubiquitination and degradation. CDR1as may also sense DNA damage signals and form a protective complex with p53 to preserve p53 function. Therefore, CDR1as depletion may play a potent role in promoting tumorigenesis through down-regulating p53 expression in glioma. Our results broaden further our understanding of the roles and mechanism of action of circular RNAs in general and CDR1as in particular, and can potentially open up novel therapeutic avenues for effective glioma treatment.

Early effect of standalone oblique lateral interbody fusion vs. combined with lateral screw fixation of the vertebral body on single-level lumbar degenerative disc disease: A pilot study.

BACKGROUND: For the treatment of single-level lumbar degenerative disc disease (DDD), oblique lateral interbody fusion (OLIF) has clinical advantages. Whether internal fixation needs to be combined for treatment has been the subject of debate. OBJECTIVE: To compare the early clinical effects of standalone oblique lateral interbody fusion (S-OLIF) versus OLIF combined with lateral screw fixation of the vertebral body (F-OLIF) on single-level lumbar DDD. METHODS: A retrospective analysis was performed on the data of 34 patients for whom the OLIF technique was applied to treat single-level lumbar DDD from August 2018 to May 2021. Patients were divided into the S-OLIF (n= 18) and F-OLIF groups (n= 16). Intraoperative blood loss, operative time, and length of hospital stay were recorded. The pain visual analogue scale (VAS) and Oswestry disability index (ODI) before and after the operation were evaluated. The disc height (DH), foraminal height (FH), fused segment lordosis (FSL), lumbar lordosis (LL), cage subsidence, and fusion by CT examination were measured before and after the operation. RESULTS: The S-OLIF group experienced a shorter operative time and less intraoperative blood loss than the F-OLIF group, and the differences were statistically significant (p< 0.05), but the difference in the length of hospital stay was not statistically significant. The postoperative VAS score and ODI of the two groups were significantly lower than those before the operation, but the postoperative differences between the two groups were not statistically significant. Differences were not statistically significant in postoperative FH, DH, FSL and LL of the two groups. Both groups were followed up for no less than 12 months. In the two groups, fusion was achieved at the last follow-up visit. CONCLUSION: According to short-term follow-up results, both S-OLIF and F-OLIF can achieve reliable and stable fusion and good clinical effect in the treatment of single-level lumbar DDD.

Sp1 induced gene TIMP1 is related to immune cell infiltration in glioblastoma.

Tumor immune microenvironment exerts a profound effect on the population of infiltrating immune cells. Tissue inhibitor of matrix metalloproteinase 1 (TIMP1) is frequently overexpressed in a variety of cells, particularly during inflammation and tissue injury. However, its function in cancer and immunity remains enigmatic. In this study, we find that TIMP1 is substantially up-regulated during tumorigenesis through analyzing cancer bioinformatics databases, which is further confirmed by IHC tissue microarrays of clinical samples. The TIMP1 level is significantly increased in lymphocytes infiltrating the tumors and correlated with cancer progression, particularly in GBM. Notably, we find that the transcriptional factor Sp1 binds to the promoter of TIMP1 and triggers its expression in GBM. Together, our findings suggest that the Sp1-TIMP1 axis can be a potent biomarker for evaluating immune cell infiltration at the tumor sites and therefore, the malignant progression of GBM.

Chronic Inflammation Pathway NF-κB Cooperates with Epigenetic Reprogramming to Drive the Malignant Progression of Glioblastoma.

Without an effective strategy for targeted therapy, glioblastoma is still incurable with a median survival of only 15 months. Both chronic inflammation and epigenetic reprogramming are hallmarks of cancer. However, the mechanisms and consequences of their cooperation in glioblastoma remain unknown. Here, we discover that chronic inflammation governs H3K27me3 reprogramming in glioblastoma through the canonical NF-κB pathway to target EZH2. Being a crucial mediator of chronic inflammation, the canonical NF-κB signalling specifically directs the expression and redistribution of H3K27me3 but not H3K4me3, H3K9me3 and H3K36me3. Using RNA-seq screening to focus on genes encoding methyltransferases and demethylases of histone, we identify EZH2 as a key methyltransferase to control inflammation-triggered epigenetic reprogramming in gliomagenesis. Mechanistically, NF-κB selectively drives the expression of EZH2 by activating its transcription, consequently resulting in a global change in H3K27me3 expression and distribution. Furthermore, we find that co-activation of NF-κB and EZH2 confers the poorest clinical outcome, and that the risk for glioblastoma can be accurately molecularly stratified by NF-κB and EZH2. It is notable that NF-κB can potentially cooperate with EZH2 in more than one way, and most importantly, we demonstrate a Synergistic effect of cancer cells induced by combinatory inhibition of NF-κB and EZH2, which both are frequently over-activated in glioblastoma. In summary, we uncover a functional cooperation between chronic inflammation and epigenetic reprogramming in glioblastoma, combined targeting of which by inhibitors guaranteed in safety and availability furnishes a potent strategy for effective treatment of this fatal disease.

Systematic node management mechanism using ZigBee-based real-time vital sign information monitoring system to manage large numbers of patients.

BACKGROUND: Local hospitals must deal with large numbers of patients during mass casualty incidents, and the wireless sensor networks (WSNs) can help in these situations by monitoring vital signs. Conventional ZigBee nodes can obtain the ID of a device by assigning a unique 16-bit short address or by burning firmware into an IC. These methods tend to complicate node management and lack portability. OBJECTIVE: The study developed a node management mechanism to deal with a large number of patients in real-time, through the wireless monitoring of physiological signals. The mechanism proposed for the ZigBee WSN is based on a three-layer (Coordinator, Control Router, and End Device) tree topology. METHODS: The proposed system includes a node deployment process to formulate a ZigBee WSN as a tree topology, an algorithm to automatically number ZigBee nodes for monitoring and control system (MCS), and an algorithm to automatically obtain the short addresses of nodes for data collection. Specifically, an algorithm automatically collects data from ZigBee nodes for display on a computer graphical user interface (GUI). We also developed a reliable data transmission method capable of resolving the problem of packet loss. RESULTS: The proposed method has been applied in a local hospital. Our research findings provide a valuable reference for the development of ZigBee-based MCS. CONCLUSIONS: The proposed node management mechanism is faster, more reliable, and more intuitive to use, than traditional methods.