Multiparametric MRI-Based Interpretable Radiomics Machine Learning Model Differentiates Medulloblastoma and Ependymoma in Children: A Two-Center Study

RATIONALE AND OBJECTIVES

Medulloblastoma (MB) and Ependymoma (EM) in children, share similarities in age group, tumor location, and clinical presentation. Distinguishing between them through clinical diagnosis is challenging. This study aims to explore the effectiveness of using radiomics and machine learning on multiparametric magnetic resonance imaging (MRI) to differentiate between MB and EM and validate its diagnostic ability with an external set.

MATERIALS AND METHODS

Axial T2 weighted image (T2WI) and contrast-enhanced T1weighted image (CE-T1WI) MRI sequences of 135 patients from two centers were collected as train/test sets. Volume of interest (VOI) was manually delineated by an experienced neuroradiologist, supervised by a senior. Feature selection analysis and the least absolute shrinkage and selection operator (LASSO) algorithm identified valuable features, and Shapley additive explanations (SHAP) evaluated their significance. Five machine-learning classifiers-extreme gradient boosting (XGBoost), Bernoulli naive Bayes (Bernoulli NB), Logistic Regression (LR), support vector machine (SVM), linear support vector machine (Linear SVC) classifiers were built based on T2WI (T2 model), CE-T1WI (T1 model), and T1 + T2WI (T1 + T2 model). A human expert diagnosis was developed and corrected by senior radiologists. External validation was performed at Sun Yat-Sen University Cancer Center.

RESULTS

31 valuable features were extracted from T2WI and CE-T1WI. XGBoost demonstrated the highest performance with an area under the curve (AUC) of 0.92 on the test set and maintained an AUC of 0.80 during external validation. For the T1 model, XGBoost achieved the highest AUC of 0.85 on the test set and the highest accuracy of 0.71 on the external validation set. In the T2 model, XGBoost achieved the highest AUC of 0.86 on the test set and the highest accuracy of 0.82 on the external validation set. The human expert diagnosis had an AUC of 0.66 on the test set and 0.69 on the external validation set. The integrated T1 + T2 model achieved an AUC of 0.92 on the test set, 0.80 on the external validation set, achieved the best performance. Overall, XGBoost consistently outperformed in different classification models.

CONCLUSION

The combination of radiomics and machine learning on multiparametric MRI effectively distinguishes between MB and EM in childhood, surpassing human expert diagnosis in training and testing sets.

[The cases of tracing the source of patients infected with Omicron variant of SARS-CoV-2 based on wastewater-based epidemiology in Shenzhen]

Wastewater-based epidemiology (WBE) is an emerging discipline, which has been applied to drug abuse tracking and infectious disease pathogen surveillance. During the COVID-19 epidemic, WBE has been applied to monitor the epidemic trend and SARS-CoV-2 variants etc. In order to detect hidden COVID-19 cases and prevent transmission in the community, wastewater surveillance system for monitoring SARS-CoV-2 RNA was developed in Shenzhen. The sewage sampling sites were set up in key places such as the port areas, urban villages and residential communities of Futian, Nanshan, Luohu and Yantian districts. From July 26 to November 30, 2022, a total of 369 sewage sampling sites were set up, covering 1.93 million people. Continuous sampling was carried out for 3 hours in the peak period of water use every day. Sewage virus enrichment and SARS-CoV-2 nucleic acid detection were carried out by polyethylene glycol precipitation method and RT-qPCR, and a positive water sample disposal process was molded. This article aims to introduce the case of source tracing of COVID-19 infected patients based on urban sewage in Shenzhen. The sewage monitoring of Honghu water treatment plant in Luohu District played an early warning role, and the source of infection was traced. In the disposal of positive water samples in Futian South Road, Futian District, the important experience of monitoring point layout was obtained. In the sewage monitoring of Nanshan village, Nanshan District, the existence of occult infection was revealed. Sharing the experience of tracing the source of COVID-19 patients to avoid the spread of COVID-19 in the community based on wastewater surveillance of SARS-CoV-2 RNA in Shenzhen, and summarizing the advantages and application prospects of sewage surveillance can provide new ideas for monitoring emerging or re-emerging pathogens that are known to exhibit gastrointestinal excretion in the future.

Deep learning and radiomics of longitudinal CT scans for early prediction of tuberculosis treatment outcomes

BACKGROUND

To predict tuberculosis (TB) treatment outcomes at an early stage, prevent poor outcomes ofdrug-resistant tuberculosis(DR-TB) and interrupt transmission.

METHODS

An internal cohort for model development consists of 204 bacteriologically-confirmed TB patients who completed anti-tuberculosis treatment, with one pretreatment and two follow-up CT images (612 scans). Three radiomics feature-based models (RM) with multiple classifiers of Bagging, Random forest and Gradient boosting and two deep-learning-based models (i.e., supervised deep-learning model, SDLM; weakly supervised deep-learning model, WSDLM) are developed independently. Prediction scores of RM and deep-learning models with respectively highest performance are fused to create new fusion models under different fusion strategies. An additional independent validation was conducted on the external cohort comprising 80 patients (160 scans).

RESULTS

For RM scheme, 16 optimal radiomics features are finally selected using longitudinal scans. The AUCs of RM for Bagging, Random forest and Gradient boosting were 0.789, 0.773 and 0.764 in the internal cohort and 0.840, 0.834 and 0.816 in the external cohort, respectively. For deep learning-based scheme, AUCs of SDLM and WSDLM were 0.767 and 0.661 in the internal cohort, and 0.823 and 0.651 in the external. The fusion model yields AUCs from 0.767 to 0.802 in the internal cohort, and from 0.831 to 0.857 in the external cohort.

CONCLUSIONS

Fusion of radiomics features and deep-learning model may have the potential to predict early failure outcome of DR-TB, which may be combined to help prevent poor TB treatment outcomes.

Deep learning on longitudinal CT scans: automated prediction of treatment outcomes in hospitalized tuberculosis patients

Three deep learning (DL)-based prediction models (PMs) using longitudinal CT images were developed to predict tuberculosis (TB) treatment outcomes. The internal dataset consists of 493 bacteriologically confirmed TB patients who completed the anti-tuberculosis treatment with three-time CT scans, including a pretreatment CT scan and two follow-up CT scans. PM1 was trained using only pretreatment CT scans, and PM2 and PM3 were developed by adding follow-up scans. An independent testing was performed on external dataset comprising 86 TB patients. The area under the curve for classifying success and drug-resistant (DR)-TB was improved on both internal (0.609 vs. 0.625 vs. 0.815) and external (0.627 vs. 0.705 vs. 0.735) dataset by adding follow-up scans. The accuracy and F1-score also showed an increasing tendency in the external test. Regular follow-up CT scans can aid in the treatment prediction, and special attention should be given to early intensive phase of treatment to identify high-risk DR-TB patients.

[Analysis on the secondary attack rates of SARS-CoV-2 Omicron variant and the associated factors]

Objective: To evaluate the secondary attack rates of the SARS-CoV-2 Omicron variant and the associated factors. Methods: A total of 328 primary cases and 40 146 close contacts of the SARS-CoV-2 Omicron variant routinely detected in local areas of Jiangsu Province from February to April 2022 were selected in this study, and those with positive nucleic acid test results during 7 days of centralized isolation medical observation were defined as secondary cases. The demographic information and clinical characteristics were collected, and the secondary attack rate (SAR) and the associated factors were analyzed by using a multivariate logistic regression model. Results: A total of 1 285 secondary cases of close contacts were reported from 328 primary cases, with a SAR of 3.2% (95%CI: 3.0%-3.4%). Among the 328 primary cases, males accounted for 61.9% (203 cases), with the median age (Q1, Q3) of 38.5 (27, 51) years old. Among the 1 285 secondary cases, males accounted for 59.1% (759 cases), with the median age (Q1, Q3) of 34 (17, 52) years old. The multivariate logistic regression model showed that the higher SAR was observed in the primary male cases (OR=1.632, 95%CI: 1.418-1.877), younger than 20 years old (OR=1.766, 95%CI: 1.506-2.072),≥60 years old (OR=1.869, 95%CI: 1.476-2.365), infected with the BA.2 strain branch (OR=2.906, 95%CI: 2.388-3.537), the confirmed common cases (OR=2.572, 95%CI: 2.036-3.249), and confirmed mild cases (OR=1.717, 95%CI: 1.486-1.985). Meanwhile, the higher SAR was observed in the close contacts younger than 20 years old (OR=2.604, 95%CI: 2.250-3.015),≥60 years old (OR=1.287, 95%CI: 1.052-1.573) and exposure for co-residence (OR=27.854, 95%CI: 23.470-33.057). Conclusion: The sex and age of the primary case of the Omicron variant, the branch of the infected strain, case severity of the primary case, as well as the age and contact mode of close contacts are the associated factors of SAR.

Outdoor environmental exposome and the burden of tuberculosis: Findings from nearly two million adults in northwestern China

We simultaneously assessed the associations for a range of outdoor environmental exposures with prevalent tuberculosis (TB) cases in a population-based health program with 1940,622 participants ≥ 15 years of age. TB status was confirmed through bacteriological and clinical assessment. We measured 14 outdoor environmental exposures at residential addresses. An exposome-wide association study (ExWAS) approach was used to estimate cross-sectional associations between environmental exposures and prevalent TB, an adaptive elastic net model (AENET) was implemented to select important exposure(s), and the Extreme Gradient Boosting algorithm was subsequently applied to assess their relative importance. In ExWAS analysis, 12 exposures were significantly associated with prevalent TB. Eight of the exposures were selected as predictors by the AENET model: particulate matter ≤ 2.5 µm (odds ratio [OR]=1.01, p = 0.3295), nitrogen dioxide (OR=1.09, p < 0.0001), carbon monoxide (OR=1.19, p < 0.0001), and wind speed (OR=1.08, p < 0.0001) were positively associated with the odds of prevalent TB while sulfur dioxide (OR=0.95, p = 0.0017), altitude (OR=0.97, p < 0.0001), artificial light at night (OR=0.98, p = 0.0001), and proportion of forests, shrublands, and grasslands (OR=0.95, p < 0.0001) were negatively associated with the odds of prevalent TB. Air pollutants had higher relative importance than meteorological and geographical factors, and the outdoor environment collectively explained 11% of TB prevalence.

Short-Term Clinical Response and Changes in the Fecal Microbiota and Metabolite Levels in Patients with Crohn's Disease After Stem Cell Infusions

Recent studies have shown a close relationship between the gut microbiota and Crohn's disease (CD). This study aimed to determine whether mesenchymal stem cell (MSC) treatment alters the gut microbiota and fecal metabolite pathways and to establish the relationship between the gut microbiota and fecal metabolites. Patients with refractory CD were enrolled and received 8 intravenous infusions of MSCs at a dose of 1.0 × 106 cells/kg. The MSC efficacy and safety were evaluated. Fecal samples were collected, and their microbiomes were analyzed by 16S rDNA sequencing. The fecal metabolites at baseline and after 4 and 8 MSC infusions were identified by liquid chromatography-mass spectrometry (LC--MS). A bioinformatics analysis was conducted using the sequencing data. No serious adverse effects were observed. The clinical symptoms and signs of patients with CD were substantially relieved after 8 MSC infusions, as revealed by changes in weight, the CD activity index (CDAI) score, C-reactive protein (CRP) level, and erythrocyte sedimentation rate (ESR). Endoscopic improvement was observed in 2 patients. A comparison of the gut microbiome after 8 MSC treatments with that at baseline showed that the genus Cetobacterium was significantly enriched. Linoleic acid was depleted after 8 MSC treatments. A possible link between the altered Cetobacterium abundance and linoleic acid metabolite levels was observed in patients with CD who received MSCs. This study enabled an understanding of both the gut microbiota response and bacterial metabolites to obtain more information about host-gut microbiota metabolic interactions in the short-term response to MSC treatment.

[Prediction of 1p/19q codeletion status in diffuse lower-grade glioma using multimodal MRI radiomics]

OBJECTIVE

To develop a noninvasive method for prediction of 1p/19q codeletion in diffuse lower-grade glioma (DLGG) based on multimodal magnetic resonance imaging (MRI) radiomics.

METHODS

We collected MRI data from 104 patients with pathologically confirmed DLGG between October, 2015 and September, 2022. A total of 535 radiomics features were extracted from T₂WI, T₁WI, FLAIR, CE-T₁WI and DWI, including 70 morphological features, 90 first order features, and 375 texture features. We constructed logistic regression (LR), logistic regression least absolute shrinkage and selection operator (LRlasso), support vector machine (SVM) and Linear Discriminant Analysis (LDA) radiomics models and compared their predictive performance after 10-fold cross validation. The MRI images were reviewed by two radiologists independently for predicting the 1p/19q status. Receiver operating characteristic curves were used to evaluate classification performance of the radiomics models and the radiologists.

RESULTS

The 4 radiomics models (LR, LRlasso, SVM and LDA) achieved similar area under the curve (AUC) in the validation dataset (0.833, 0.819, 0.824 and 0.819, respectively; P>0.1), and their predictive performance was all superior to that of resident physicians of radiology (AUC=0.645, P=0.011, 0.022, 0.016, 0.030, respectively) and similar to that of attending physicians of radiology (AUC=0.838, P>0.05).

CONCLUSION

Multiparametric MRI radiomics models show good performance for noninvasive prediction of 1p/19q codeletion status in patients with in diffuse lower-grade glioma.

[Preliminary experience of gasless transoral vestibular robotic thyroidectomy]

Objective: To explore the feasibility and safety of the gasless transoral vestibular robotic thyroidectomy using skin suspension. Methods: The clinical data of 20 patients underwent gasless transoral vestibular robotic thyroidectomy in the Department of Otorhinolaryngology, Sun Yat-sen Memorial Hospital, Sun Yat-sen University from February 2022 to May 2022 were retrospectively analyzed. Among them, 18 were females and 2 were males, aged (38.7±8.0) years old. The intraoperative blood loss, operation time, postoperative hospital stay, postoperative drainage volume, postoperative pain visual analogue scale (VAS) score, postoperative swallowing function swallowing impairment score-6 (SIS-6), postoperative aesthetic VAS score, postoperative voice handicap index-10 (VHI-10) voice quality, postoperative pathology and complications were recorded. SPSS 25.0 was used for statistical analysis of the data. Results: The operations were successfully completed without conversion to open surgery in all patients. Pathological examination showed papillary thyroid carcinoma in 18 cases, retrosternal nodular goiter in 1 case, and cystic change in goiter in 1 case. The operative time for thyroid cancer was 161.50 (152.75, 182.50) min [M (P₂₅, P₇₅), the same below] and the average operative time for benign thyroid diseases was 166.50 minutes. The intraoperative blood loss 25.00 (21.25, 30.00) ml. In 18 cases of thyroid cancer, the mean diameter of the tumors was (7.22±2.02) mm, and lymph nodes (6.56±2.14) were dissected in the central region, with a lymph node metastasis rate of 61.11%. The postoperative pain VAS score was 3.00 (2.25, 4.00) points at 24 hours, the mean postoperative drainage volume was (118.35±24.32) ml, the postoperative hospital stay was 3.00 (3.00, 3.75) days, the postoperative SIS-6 score was (4.90±1.58) points at 3 months, and the postoperative VHI-10 score was 7.50 (2.00, 11.00) points at 3 months. Seven patients had mild mandibular numbness, 10 patients had mild cervical numbness, and 3 patients had temporary hypothyroidism three months after surgery and 1 patient had skin flap burn, but recovered one month after surgery. All patients were satisfied with the postoperative aesthetic effects, and the postoperative aesthetic VAS score was 10.00 (10.00, 10.00). Conclusion: Gasless transoral vestibular robotic thyroidectomy using skin suspension is a safe and feasible option with good postoperative aesthetic effect, which can provide a new treatment option for some selected patients with thyroid tumors.

[Epidemiological characteristics of a 2019-nCoV outbreak caused by Omicron variant BF.7 in Shenzhen]

Objective: To explore the epidemiological characteristic of a COVID-19 outbreak caused by 2019-nCoV Omicron variant BF.7 and other provinces imported in Shenzhen and analyze transmission chains and characteristics. Methods: Field epidemiological survey was conducted to identify the transmission chain, analyze the generation relationship among the cases. The 2019-nCoV nucleic acid positive samples were used for gene sequencing. Results: From 8 to 23 October, 2022, a total of 196 cases of COVID-19 were reported in Shenzhen, all the cases had epidemiological links. In the cases, 100 were men and 96 were women, with a median of age, M (Q₁, Q₃) was 33(25, 46) years. The outbreak was caused by traverlers initial cases infected with 2019-nCoV who returned to Shenzhen after traveling outside of Guangdong Province.There were four transmission chains, including the transmission in place of residence and neighbourhood, affecting 8 persons, transmission in social activity in the evening on 7 October, affecting 65 persons, transmission in work place on 8 October, affecting 48 persons, and transmission in a building near the work place, affecting 74 persons. The median of the incubation period of the infection, M (Q₁, Q₃) was 1.44 (1.11, 2.17) days. The incubation period of indoor exposure less than that of the outdoor exposure, M (Q₁, Q₃) was 1.38 (1.06, 1.84) and 1.95 (1.22, 2.99) days, respcetively (Wald χ²=10.27, P=0.001). With the increase of case generation, the number and probability of gene mutation increased. In the same transmission chain, the proportion of having 1-3 mutation sites was high in the cases in the first generation. Conclusions: The transmission chains were clear in this epidemic. The incubation period of Omicron variant BF.7 infection was shorter, the transmission speed was faster, and the gene mutation rate was higher. It is necessary to conduct prompt response and strict disease control when epidemic occurs.

Radiomics signature based on CECT for non-invasive prediction of response to anti-PD-1 therapy in patients with hepatocellular carcinoma

PURPOSE

This study aimed to develop a radiomics signature (RS) based on contrast-enhanced computed tomography (CECT) and evaluate its potential predictive value in hepatocellular carcinoma (HCC) patients receiving anti-PD-1 therapy.

METHOD

CECT scans of 76 HCC patients who received anti-PD-1 therapy were obtained in this study (training group = 53 and validation group = 23). The least absolute shrinkage and selection operator (LASSO) regression was applied to select radiomics features of primary and metastatic lesions and establish a RS to predict lesion-level response. Then, a nomogram combined the mean RS (MRS) and clinical variables with patient-level response as the end point.

RESULTS

In the lesion-level analysis, the area under the curves (AUCs) of RS in the training and validation groups were 0.751 (95% CI, 0.668-0.835) and 0.734 (95% CI, 0.604-0.864), respectively. In the patient-level analysis, the AUCs of the nomogram in the training and validation groups were 0.897 (95% CI, 0.798-0.996) and 0.889 (95% CI, 0.748-1.000), respectively. The nomogram stratified patients into low- and high-risk groups, which showed a significant difference in progression-free survival (PFS) (p<0.05).

CONCLUSIONS

The RS is a noninvasive biomarker for predicting anti-PD-1 therapy response in patients with HCC. The nomogram may be of clinical use for identifying high-risk patients and formulating individualised treatments.

Stent pretreatment for internal carotid artery exposed to necrotic lesions in nasopharyngeal carcinoma

BACKGROUND

Post radiation nasopharyngeal necrosis (PRNN) invading the internal carotid artery (ICA) contributes to the death of 69.2-72.7% of PRNN patients. ICA occlusion is an effective treatment to avoid fatal bleeding, while some patients are intolerant. We present a novel method that allows for these patients without interrupting blood flow through the ICA.

METHODOLOGY

This study enrolled patients with PRNN-invaded ICA who were not suitable for ICA occlusion from April 2020 to November 2022. ICA stent pretreatment was performed in the 36 patients and followed the endoscopic nasopharyngectomy (ENPG) or conservative treatment for PRNN. We report the survival outcome and incidence of complications after stent implantation and compare the survival outcomes of ENPG and conservative treatment for PRNN followed by stent implantation.

RESULTS

ICA stent pretreatment was performed in the 36 enrolled patients, among which 14 underwent ENPG, and 22 received conservative treatment. 27.8% patients died after a median follow-up of 15 months. The Kaplan-Meier estimates of overall survival were higher in the ENPG group than in the conservative treatment group. Karnofsky performance status (KPS) was significantly higher in the ENPG group than in the non-ENPG group.

CONCLUSIONS

The innovative application of ICA stents is a promising treatment to improve outcomes in patients with PRNN invading the ICA who are unsuitable for ICA embolization, especially when followed by endoscopic surgery. However, methods to avoid postoperative cerebral ischemia and nasopharyngeal hemorrhage still require further study.

Protective effect of astaxanthin on tuberculosis-associated inflammatory lung injury

Mycobacterium tuberculosis (MTB) invades the lungs and is the key cause of tuberculosis (TB). MTB induces immune overreaction and inflammatory damage to lung tissue. There is a lack of protective drugs against pulmonary inflammatory damage. Herein, the protective roles and mechanisms of Astaxanthin (ASTA), a natural compound, in inflammatory injured lung epithelial cells were investigated. Lipopolysaccharide (LPS) was used to establish inflammatory injury model in the murine lung epithelial (MLE)-12 cells. Cell counting kit-8 was used for screening of compound concentrations. Cell proliferation was observed real-time with a high content analysis system. Flow cytometry assessed apoptosis. The changes of apoptotic proteins and key proteins in nuclear factor kappa-B (NF-κB) pathway were measured with the western blot. LPS was used to establish an animal model of pulmonary injury. The pathological changes and degree of inflammatory injury in lung tissue were observed with hematoxylin and eosin (HE) staining. The levels of inflammatory mediators were detected with enzyme-linked immunosorbent assay. The results showed that ASTA reduced lung inflammation and attenuated inflammatory damage in lung tissues. ASTA reduced apoptosis stimulated by LPS through suppressing the NF-κB pathway in MLE-12 cells. We believe that ASTA may have great potential for protection against inflammatory damage to lung tissue.

Key genes associated with non-alcoholic fatty liver disease and hepatocellular carcinoma with metabolic risk factors

Background: Hepatocellular carcinoma (HCC) has become the world's primary cause of cancer death. Obesity, hyperglycemia, and dyslipidemia are all illnesses that are part of the metabolic syndrome. In recent years, this risk factor has become increasingly recognized as a contributing factor to HCC. Around the world, non-alcoholic fatty liver disease (NAFLD) is on the rise, especially in western countries. In the past, the exact pathogenesis of NAFLD that progressed to metabolic risk factors (MFRs)-associated HCC has not been fully understood. Methods: Two groups of the GEO dataset (including normal/NAFLD and HCC with MFRs) were used to analyze differential expression. Differentially expressed genes of HCC were verified by overlapping in TCGA. In addition, functional enrichment analysis, modular analysis, Receiver Operating Characteristic (ROC) analysis, LASSO analysis, and Genes with key survival characteristics were analyzed. Results: We identified six hub genes (FABP5, SCD, CCL20, AGPAT9(GPAT3), PLIN1, and IL1RN) that may be closely related to NAFLD and HCC with MFRs. We constructed survival and prognosis gene markers based on FABP5, CCL20, AGPAT9(GPAT3), PLIN1, and IL1RN.This gene signature has shown good diagnostic accuracy in both NAFLD and HCC and in predicting HCC overall survival rates. Conclusion: As a result of the findings of this study, there is some guiding significance for the diagnosis and treatment of liver disease associated with NAFLD progression.

Dietary Total Antioxidant Capacity and Cognitive Function in Older Adults in the United States: The NHANES 2011-2014

OBJECTIVES

Oxidative stress level takes part in the development of cognitive decline. However, the association between total antioxidant capacity (TAC) from diet and cognitive function is controversial. The aim of this study was to investigate the relationship between TAC and the cognitive function of older adults in the U.S.

DESIGN

A cross-sectional study.

SETTING

National Health and Nutrition Examination Surveys database.

PARTICIPANTS

2712 older adults aged over 60 years.

MEASUREMENTS

TAC was calculated from 8 antioxidative vitamins based on the reference values for vitamin C equivalent antioxidant capacity obtained from individuals' 24 h dietary recall. Four memory-related assessments were employed [Immediate Recall test (IRT), Delayed Recall test (DRT), Animal Fluency test (AFT), and Digit Symbol Substitution test (DSST)].

RESULTS

Among the 2712 participants, the median age was 68 years, and 50.4% were women. Participants in the group with higher TAC levels had relatively higher IRT, AFT and DSST scores (P=0.025, P=0.008, P<0.001, respectively). In adjusted weighted linear regression, log-transformed TAC was positively associated with AFT (β=1.10, 95%CI: 0.51, 1.70) and DSST (β=2.81, 95%CI: 1.16, 4.45). Compared with the first quartile, the participants in the second (Q2 vs. Q1, OR=0.66, 95%CI: 0.43,1.02) and fourth quartile (Q4 vs. Q1, OR=0.47, 95%CI:0.28, 0.78) of log-transformed TAC showed a decreased risk of impaired cognitive function (ICF) after adjusting for confounders. The dose-response analysis indicated a gradual descent in the risk of ICF as TAC increases. Diabetes mellitus (DM) mediated part of the effect of TAC on ICF. The relationship between TAC and ICF was more pronounced in subjects with DM (Q4 vs Q1, OR=0.36, 95%CI:0.17, 0.74).

CONCLUSION

Our findings support that higher dietary antioxidant potential was related to a decreased risk of cognitive dysfunction, particularly in the subjects with DM who may have oxidative injury. DM was one of the factors mediating the effect of TAC on ICF.

Deep learning based CT images automatic analysis model for active/non-active pulmonary tuberculosis differential diagnosis

Active pulmonary tuberculosis (ATB), which is more infectious and has a higher mortality rate compared with non-active pulmonary tuberculosis (non-ATB), needs to be diagnosed accurately and timely to prevent the tuberculosis from spreading and causing deaths. However, traditional differential diagnosis methods of active pulmonary tuberculosis involve bacteriological testing, sputum culturing and radiological images reading, which is time consuming and labour intensive. Therefore, an artificial intelligence model for ATB differential diagnosis would offer great assistance in clinical practice. In this study, computer tomography (CT) scans images and corresponding clinical information of 1160 ATB patients and 1131 patients with non-ATB were collected and divided into training, validation, and testing sets. A 3-dimension (3D) Nested UNet model was utilized to delineate lung field regions in the CT images, and three different pre-trained deep learning models including 3D VGG-16, 3D EfficientNet and 3D ResNet-50 were used for classification and differential diagnosis task. We also collected an external testing set with 100 ATB cases and 100 Non-ATB cases for further validation of the model. In the internal and external testing set, the 3D ResNet-50 model outperformed other models, reaching an AUC of 0.961 and 0.946, respectively. The 3D ResNet-50 model reached even higher levels of diagnostic accuracy than experienced radiologists, while the CT images reading and diagnosing speed was 10 times faster than human experts. The model was also capable of visualizing clinician interpretable lung lesion regions important for differential diagnosis, making it a powerful tool assisting ATB diagnosis. In conclusion, we developed an auxiliary tool to differentiate active and non-active pulmonary tuberculosis, which would have broad prospects in the bedside.

A symmetric prior knowledge based deep learning model for intracerebral hemorrhage lesion segmentation

Background: Accurate localization and classification of intracerebral hemorrhage (ICH) lesions are of great significance for the treatment and prognosis of patients with ICH. The purpose of this study is to develop a symmetric prior knowledge based deep learning model to segment ICH lesions in computed tomography (CT). Methods: A novel symmetric Transformer network (Sym-TransNet) is designed to segment ICH lesions in CT images. A cohort of 1,157 patients diagnosed with ICH is established to train (n = 857), validate (n = 100), and test (n = 200) the Sym-TransNet. A healthy cohort of 200 subjects is added, establishing a test set with balanced positive and negative cases (n = 400), to further evaluate the accuracy, sensitivity, and specificity of the diagnosis of ICH. The segmentation results are obtained after data pre-processing and Sym-TransNet. The DICE coefficient is used to evaluate the similarity between the segmentation results and the segmentation gold standard. Furthermore, some recent deep learning methods are reproduced to compare with Sym-TransNet, and statistical analysis is performed to prove the statistical significance of the proposed method. Ablation experiments are conducted to prove that each component in Sym-TransNet could effectively improve the DICE coefficient of ICH lesions. Results: For the segmentation of ICH lesions, the DICE coefficient of Sym-TransNet is 0.716 ± 0.031 in the test set which contains 200 CT images of ICH. The DICE coefficients of five subtypes of ICH, including intraparenchymal hemorrhage (IPH), intraventricular hemorrhage (IVH), extradural hemorrhage (EDH), subdural hemorrhage (SDH), and subarachnoid hemorrhage (SAH), are 0.784 ± 0.039, 0.680 ± 0.049, 0.359 ± 0.186, 0.534 ± 0.455, and 0.337 ± 0.044, respectively. Statistical results show that the proposed Sym-TransNet can significantly improve the DICE coefficient of ICH lesions in most cases. In addition, the accuracy, sensitivity, and specificity of Sym-TransNet in the diagnosis of ICH in 400 CT images are 91.25%, 98.50%, and 84.00%, respectively. Conclusion: Compared with recent mainstream deep learning methods, the proposed Sym-TransNet can segment and identify different types of lesions from CT images of ICH patients more effectively. Moreover, the Sym-TransNet can diagnose ICH more stably and efficiently, which has clinical application prospects.

Development of a wide-range soft sensor for predicting wastewater BOD5 using an eXtreme gradient boosting (XGBoost) machine

In wastewater monitoring, detecting extremely high pollutant concentrations is necessary to properly calibrate the treatment process. However, existing hardware sensors have a limited linear range which may fail to measure extremely high levels of pollutants; and likewise, the conventional "soft" model sensors are not suitable for the highly-skewed data distributions either. This study developed a new soft sensor by using eXtreme Gradient Boosting (XGBoost) machine learning to 'measure' the wastewater organics (in terms of 5-day biochemical oxygen demand (BOD5)). The soft sensor was tested on influent and effluent BOD5 of two different wastewater treatment plants to validate the results. The model results showed that XGBoost can detect these extreme values better than conventional soft sensors. This new soft sensor can function using a sparse input matrix via XGBoost's sparsity awareness algorithm - which can address the limitation of the conventional soft sensor with the fallibility of supporting hardware sensors even.

Extracellular vesicles from bone mesenchymal stem cells transport microRNA-206 into osteosarcoma cells and target NRSN2 to block the ERK1/2-Bcl-xL signaling pathway

Osteosarcoma (OS) is a kind of malignant tumor originating from mesenchymal tissues. Bone mesenchymal stem cells-derived extracellular vesicles (BMSCs-EVs) can play important roles in OS. This study investigated the mechanism of BMSCs-EVs on OS. BMSC surface antigens and adipogenic and osteogenic differentiation were detected by flow cytometry, and oil red O and alizarin red staining. EVs were isolated from BMSCs by differential centrifugation and identified by transmission electron microscopy, nanoparticle tracking analysis, and Western blot (WB). miR-206 and neurensin-2 (NRSN2) levels in human osteoblast hFOB 1.19 or OS cells (143B, MG-63, Saos2, HOS) were detected by RT-qPCR. Human OS cells with lower miR-206 levels were selected and treated with BMSCs-EVs or pSUPER-NRSN2. The uptake of EVs by 143B cells, cell proliferation, apoptosis, invasion, and migration were detected by immunofluorescence, 5-ethynyl-2’-deoxyuridine (EdU) and colony formation assays, flow cytometry, scratch test, and transwell assays. The binding sites between miR-206 and NRSN2 were predicted by Starbase database and verified by dual-luciferase assay. The OS xenograft model was established and treated with BMSCs-EVs. Tumor growth rate and volume, cell proliferation, and p-ERK1/2, ERK1/2, and Bcl-xL levels were detected by vernier caliper, immunohistochemistry, and WB. BMSCs-EVs were successfully extracted. miR-206 was diminished and NRSN2 was promoted in OS cells. BMSCs-EVs inhibited proliferation, migration, and invasion, and promoted apoptosis of OS cells. BMSCs-EVs carried miR-206 into OS cells. Inhibition of miR-206 in EVs partially reversed the inhibitory effect of EVs on malignant behaviors of OS cells. miR-206 targeted NRSN2. Overexpression of NRSN2 reversed the inhibitory effect of EVs on OS cells. NRSN2 activated the ERK1/2-Bcl-xL pathway. BMSC-EVs inhibited OS growth in vivo. In summary, BMSC-EVs targeted NRSN2 and inhibited the ERK1/2-Bcl-xL pathway by carrying miR-206 into OS cells, thus inhibiting OS progression.

Bioartificial livers: a review of their design and manufacture

Acute liver failure (ALF) is a rapidly progressive disease with high morbidity and mortality rates. Liver transplantation and artificial liver support systems, such as artificial livers (ALs) and bioartificial livers (BALs), are the two major therapies for ALF. Compared to ALs, BALs are composed of functional hepatocytes that provide essential liver functions, including detoxification, metabolite synthesis, and biotransformation. Furthermore, BALs can potentially provide effective support as a form of bridging therapy to liver transplantation or spontaneous recovery for patients with ALF. In this review, we systematically discussed the currently available state-of-the-art designs and manufacturing processes for BAL support systems. Specifically, we classified the cell sources and bioreactors that are applied in BALs, highlighted the advanced technologies of hepatocyte culturing and bioreactor fabrication, and discussed the current challenges and future trends in developing next generation BALs for large scale clinical applications.

Prediction and Diagnosis of Respiratory Disease by Combining Convolutional Neural Network and Bi-directional Long Short-Term Memory Methods

Objective

Based on the respiratory disease big data platform in southern Xinjiang, we established a model that predicted and diagnosed chronic obstructive pulmonary disease, bronchiectasis, pulmonary embolism and pulmonary tuberculosis, and provided assistance for primary physicians.

Methods

The method combined convolutional neural network (CNN) and long-short-term memory network (LSTM) for prediction and diagnosis of respiratory diseases. We collected the medical records of inpatients in the respiratory department, including: chief complaint, history of present illness, and chest computed tomography. Pre-processing of clinical records with “jieba” word segmentation module, and the Bidirectional Encoder Representation from Transformers (BERT) model was used to perform word vectorization on the text. The partial and total information of the fused feature set was encoded by convolutional layers, while LSTM layers decoded the encoded information.

Results

The precisions of traditional machine-learning, deep-learning methods and our proposed method were 0.6, 0.81, 0.89, and F1 scores were 0.6, 0.81, 0.88, respectively.

Conclusion

Compared with traditional machine learning and deep-learning methods that our proposed method had a significantly higher performance, and provided precise identification of respiratory disease.

Artificial Intelligence Assisting the Early Detection of Active Pulmonary Tuberculosis From Chest X-Rays: A Population-Based Study

As a major infectious disease, tuberculosis (TB) still poses a threat to people’s health in China. As a triage test for TB, reading chest radiography with traditional approach ends up with high inter-radiologist and intra-radiologist variability, moderate specificity and a waste of time and medical resources. Thus, this study established a deep convolutional neural network (DCNN) based artificial intelligence (AI) algorithm, aiming at diagnosing TB on posteroanterior chest X-ray photographs in an effective and accurate way. Altogether, 5,000 patients with TB and 4,628 patients without TB were included in the study, totaling to 9,628 chest X-ray photographs analyzed. Splitting the radiographs into a training set (80.4%) and a testing set (19.6%), three different DCNN algorithms, including ResNet, VGG, and AlexNet, were trained to classify the chest radiographs as images of pulmonary TB or without TB. Both the diagnostic accuracy and the area under the receiver operating characteristic curve were used to evaluate the performance of the three AI diagnosis models. Reaching an accuracy of 96.73% and marking the precise TB regions on the radiographs, ResNet algorithm-based AI outperformed the rest models and showed excellent diagnostic ability in different clinical subgroups in the stratification analysis. In summary, the ResNet algorithm-based AI diagnosis system provided accurate TB diagnosis, which could have broad prospects in clinical application for TB diagnosis, especially in poor regions with high TB incidence.

Whole-Exome Sequencing Implicates the USP34 rs777591A > G Intron Variant in Chronic Obstructive Pulmonary Disease in a Kashi Cohort

Genetic factors are important factors in chronic obstructive pulmonary disease (COPD) onset. Plenty of risk and new causative genes for COPD have been identified in patients of the Chinese Han population. In contrast, we know considerably little concerning the genetics in the Kashi COPD population (Uyghur). This study aims at clarifying the genetic maps regarding COPD susceptibility in Kashi (China). Whole-exome sequencing (WES) was used to analyze three Uyghur families with COPD in Kashi (eight patients and one healthy control). Sanger sequencing was also used to verify the WES results in 541 unrelated Uyghur COPD patients and 534 Uyghur healthy controls. WES showed 72 single nucleotide variants (SNVs), two deletions, and small insertions (InDels), 26 copy number variants (CNVs), and 34 structural variants (SVs), including g.71230620T > A (rs12449210T > A, NC_000,016.10) in the HYDIN axonemal central pair apparatus protein (HYDIN) gene and g.61190482A > G (rs777591A > G, NC_000002.12) in the ubiquitin-specific protease 34 (USP34) gene. After Sanger sequencing, we found that rs777591“AA” under different genetic models except for the dominant model (adjusted OR = 0.8559, 95%CI 0.6568–1.115, p > .05), could significantly reduce COPD risk, but rs12449210T > A was not related to COPD. In stratified analysis of smoking status, rs777591“AA” reduced COPD risk significantly among the nonsmoker group. Protein and mRNA expression of USP34 in cigarette smoke extract-treated BEAS-2b cells increased significantly compared with those in the control group. Our findings associate the USP34 rs777591“AA” genotype as a protector factor in COPD.

Dynamic Changes of Endogenic or Exogenic β-Carboline Alkaloid Harmine in Different Mammals and Human in vivo at Developmental and Physiological States

OBJECTIVE

Several β-carboline alkaloids (βCBs), such as harmine, harmaline, harmane, and nor-harmane, are effective for Alzheimer's disease mouse models. They can be found in some plants, common foodstuffs, and blank plasma of various mammals. However, whether these compounds in mammals are exogenous or endogenous remain unclear.

METHODS

The exposure levels of βCBs and of neurotransmitters in plasma and tissues of pup rats, aging rats, mice of different physiological states, and healthy volunteers were detected by using UPLC-MS/MS. Plasma and tissue samples from 110 newborn rats up to 29 days old at 11 sampling points were collected and were analyzed to determine the concentration variation of βCBs in the developmental phase of newborn rats. The plasma of rats aged 2 to 18 months was used to detect the variation trend of βCBs and with some neurotransmitters. The plasma samples of normal C57BL/6 mice, APP/PS1 double transgenic mice, and scopolamine-induced memory impairment mice were collected and were analyzed to compare the difference of βCBs in different physiological states. The exposure levels of βCBs such as harmine, harmaline, and harmane in plasma of 550 healthy volunteers were also detected and analyzed on the basis of gender, race, and age.

RESULTS

Results showed that harmine was the main compound found in rats, mice, and human, which can be detected in a newborn rat plasma (0.16 ± 0.03 ng/ml) and brain (0.33 ± 0.14 ng/g) without any exogenous consumption. The concentration of harmine in rat plasma showed a decreasing trend similar to the exposure levels of neurotransmitters such as 5-hydroxytryptamine, acetylcholine chloride, glutamic acid, tyrosine, and phenylalanine during the growth period of 18 months. The harmine exposure in rats and human indicates high dependence on the physiological and pathological status such as aging, gender, and race.

CONCLUSION

The dynamic changes of harmine exposure in different animals and human, in vivo, at developmental and physiological states indicate that harmine is a naturally and widely distributed endogenous substance in different mammals and human. In addition to exogenous ingestion, spontaneous synthesis might be another important source of harmine in mammals, which should be verified by further experiment.

Deep-learning models for the detection and incidence prediction of chronic kidney disease and type 2 diabetes from retinal fundus images

Regular screening for the early detection of common chronic diseases might benefit from the use of deep-learning approaches, particularly in resource-poor or remote settings. Here we show that deep-learning models can be used to identify chronic kidney disease and type 2 diabetes solely from fundus images or in combination with clinical metadata (age, sex, height, weight, body-mass index and blood pressure) with areas under the receiver operating characteristic curve of 0.85-0.93. The models were trained and validated with a total of 115,344 retinal fundus photographs from 57,672 patients and can also be used to predict estimated glomerulal filtration rates and blood-glucose levels, with mean absolute errors of 11.1-13.4 ml min^-1 per 1.73 m² and 0.65-1.1 mmol l^-1, and to stratify patients according to disease-progression risk. We evaluated the generalizability of the models for the identification of chronic kidney disease and type 2 diabetes with population-based external validation cohorts and via a prospective study with fundus images captured with smartphones, and assessed the feasibility of predicting disease progression in a longitudinal cohort.

A multicenter study to develop a non-invasive radiomic model to identify urinary infection stone in vivo using machine-learning

Urolithiasis is a common urological disease, and treatment strategy options vary between different stone types. However, accurate detection of stone composition can only be performed in vitro. The management of infection stones is particularly challenging with yet no effective approach to pre-operatively identify infection stones from non-infection stones. Therefore, we aimed to develop a radiomic model for preoperatively identifying infection stones with multicenter validation. In total, 1198 eligible patients with urolithiasis from three centers were divided into a training set, an internal validation set, and two external validation sets. Stone composition was determined by Fourier transform infrared spectroscopy. A total of 1316 radiomic features were extracted from the pre-treatment Computer Tomography images of each patient. Using the least absolute shrinkage and selection operator algorithm, we identified a radiomic signature that achieved favorable discrimination in the training set, which was confirmed in the validation sets. Moreover, we then developed a radiomic model incorporating the radiomic signature, urease-producing bacteria in urine, and urine pH based on multivariate logistic regression analysis. The nomogram showed favorable calibration and discrimination in the training and three validation sets (area under the curve [95% confidence interval], 0.898 [0.840-0.956], 0.832 [0.742-0.923], 0.825 [0.783-0.866], and 0.812 [0.710-0.914], respectively). Decision curve analysis demonstrated the clinical utility of the radiomic model. Thus, our proposed radiomic model can serve as a non-invasive tool to identify urinary infection stones in vivo, which may optimize disease management in urolithiasis and improve patient prognosis.

A deep-learning pipeline for the diagnosis and discrimination of viral, non-viral and COVID-19 pneumonia from chest X-ray images

Common lung diseases are first diagnosed using chest X-rays. Here, we show that a fully automated deep-learning pipeline for the standardization of chest X-ray images, for the visualization of lesions and for disease diagnosis can identify viral pneumonia caused by coronavirus disease 2019 (COVID-19) and assess its severity, and can also discriminate between viral pneumonia caused by COVID-19 and other types of pneumonia. The deep-learning system was developed using a heterogeneous multicentre dataset of 145,202 images, and tested retrospectively and prospectively with thousands of additional images across four patient cohorts and multiple countries. The system generalized across settings, discriminating between viral pneumonia, other types of pneumonia and the absence of disease with areas under the receiver operating characteristic curve (AUCs) of 0.94-0.98; between severe and non-severe COVID-19 with an AUC of 0.87; and between COVID-19 pneumonia and other viral or non-viral pneumonia with AUCs of 0.87-0.97. In an independent set of 440 chest X-rays, the system performed comparably to senior radiologists and improved the performance of junior radiologists. Automated deep-learning systems for the assessment of pneumonia could facilitate early intervention and provide support for clinical decision-making.

[Clinical efficacy of anlotinib plus S-1 as a second-line therapy for recurrent or metastatic esophageal squamous cell carcinoma]

OBJECTIVE

To investigate the efficacy of anlotinib plus S-1 for treatment of patients with recurrent or metastatic esophageal squamous cell carcinoma with failed first-line chemotherapy.

OBJECTIVE

Twenty-six patients with recurrent or metastatic esophageal squamous cell carcinoma patients who experienced progression after first-line paclitaxel plus platinum chemotherapy in our hospital between July, 2018 and February, 2020 were enrolled in this study. The patients received oral anlotinib along with S-1 treatment (anlotinib at 12 mg once daily and S-1 at 50 mg twice daily for two weeks; 3 weeks per cycle). The objective response rate (ORR), disease control rate (DCR), progression-free survival (PFS) and adverse effects were evaluated for all the patients.

OBJECTIVE

No complete remission (CR) was observed in the 26 patients. Partial remission (PR) was achieved in 6 cases, stable disease (SD) in 12 cases, and progressive disease (PD) occurred in 8 cases, with an ORR of 23.1% and a DCR of 69.2% in these patients. The median PFS was 4.5 months (95%CI: 2.7-6.4 months). Univariate analysis showed that the patients with moderate or high tumor differentiation had significantly longer PFS than those with low tumor differentiation (6.1 months vs 1.9 months, P < 0.05). Multivariate analysis suggested that pathological differentiation grade (HR=6.778, 95%CI: 1.997-23.012) was an independent factor for a prolonged PFS. The adverse effects in the patients included mainly fatigue, hypertension and hand-foot syndrome, mostly of grade 1 to 2.

OBJECTIVE

Patients with recurrent or metastatic esophageal squamous cell carcinoma can benefit from a second-line anlotinib plus S-1 treatment, which has relatively mild adverse effects with a good safety profile.