Search

With the rapid growth of the Internet of Things (IoT), smart systems and applications are equipped with an increasing number of wearable sensors and mobile devices. These sensors are used not only to collect data but, more importantly, to assist in tracking and analyzing the daily human activities. Sensor-based human activity recognition is a hotspot and starts to employ deep learning approaches to supersede traditional shallow learning that rely on hand-crafted features. Although many successful methods have been proposed, there are three challenges to overcome: (1) deep model’s performance overly depends on the data size; (2) deep model cannot explicitly capture abundant sample distribution characteristics; (3) deep model cannot jointly consider sample features, sample distribution characteristics, and the relationship between the two. To address these issues, we propose a meta-learning-based graph prototypical model with priority attention mechanism for sensor-based human activity recognition. This approach learns not only sample features and sample distribution characteristics via meta-learning-based graph prototypical model, but also the embeddings derived from priority attention mechanism that mines and utilizes relations between sample features and sample distribution characteristics. What is more, the knowledge learned through our approach can be seen as a priori applicable to improve the performance for other general reasoning tasks. Experimental results on fourteen datasets demonstrate that the proposed approach significantly outperforms other state-of-the-art methods. On the other hand, experiments of applying our model to two other tasks show that our model effectively supports other recognition tasks related to human activity and improves performance on the datasets of these tasks.

Objective To analyze the duration of gonadotropin hyperstimulation's impacts on oocyte quality and clinical outcomes in aged in vitro fertilization (IVF) patients. Materials and methods This retrospective study was carried out using IVF records of the Chang Gung Memorial Hospital IVF center from January 2017 to December 2019. A total of 308 IVF cycles with patients aged 40–44 years were included. Clinical characteristics of patients who received a short controlled ovarian hyperstimulation (COH) (i.e., 6–7 days; s-COH group) or a long COH treatment (i.e., 9–10 days; l-COH group) were compared. In addition, analysis was conducted using data within two age subgroups: 40–42 years and 42–44 years subgroups. Results The s-COH group received significantly lower total doses of gonadotropin and had smaller leading follicles at the time of ovulation trigger when compared to the l-COH group. The s-COH group also produced a significantly lower number of oocytes, mature metaphase II (MII) oocytes, and 2 PN zygotes compared to the l-COH group. However, there was no significant difference in the number of transferable and good-quality embryos between the two treatment groups. Likewise, the pregnancy rate and live birth rate were comparable in the s-COH and l-COH groups. Similar results were obtained when the analysis was limited to select age subgroups (i.e., 40–42 and 42–44 years subgroups). Conclusion While a long COH generates more oocytes per cycle, a 6–7 days COH treatment, which is at the lower end of the recommended window of stimulation, could achieve a pregnancy outcome comparable to that applied 9–10 days of COH in aged patients.

The sudden increase in coronavirus disease 2019 (COVID-19) cases puts high pressure on healthcare services worldwide. At this stage, fast, accurate, and early clinical assessment of the disease severity is vital. In general, there are two issues to overcome: (1) Current deep learning-based works suffer from multimodal data adequacy issues; (2) In this scenario, multimodal (e.g., text, image) information should be taken into account together to make accurate inferences. To address these challenges, we propose a multi-modal knowledge graph attention embedding for COVID-19 diagnosis. Our method not only learns the relational embedding from nodes in a constituted knowledge graph but also has access to medical knowledge, aiming at improving the performance of the classifier through the mechanism of medical knowledge attention. The experimental results show that our approach significantly improves classification performance compared to other state-of-the-art techniques and possesses robustness for each modality from multi-modal data. Moreover, we construct a new COVID-19 multi-modal dataset based on text mining, consisting of 1393 doctor–patient dialogues and their 3706 images (347 X-ray + 2598 CT + 761 ultrasound) about COVID-19 patients and 607 non-COVID-19 patient dialogues and their 10754 images (9658 X-ray + 494 CT + 761 ultrasound), and the fine-grained labels of all. We hope this work can provide insights to the researchers working in this area to shift the attention from only medical images to the doctor–patient dialogue and its corresponding medical images.

Mortality rates due to Cryptococcus neoformans var. grubii fungemia remain significant despite treatment with antifungal drugs. The predictive function of antifungal susceptibility and its correlation with treatment outcome remains controversial. A retrospective study was conducted from January 1, 2009, to December 31, 2016, on 85 patients with C. neoformans var. grubii fungemia confirmed by matrix-assisted laser desorption ionization-time-of-flight mass spectrometry. Antifungal drug susceptibility was determined using the YeastONE™ colorimetric broth microdilution method coupled with Vizion™ System following the Clinical and Laboratory Standards Institute guidelines. Six antifungal agents-amphotericin B, fluconazole, flucytosine, itraconazole, posaconazole, and voriconazole-were tested. The patients' demographic data and clinical information were abstracted for further analyses. Antifungal regimens consisting of amphotericin B with or without fluconazole or flucytosine were administered for induction treatment of these patients, followed with intravenous or oral fluconazole for maintenance therapy. Clinical outcomes were defined by 14- and 30-day mortality rates. Risk factors associated with outcomes were fitted in a logistic regression model by univariate or multivariate method. Eighty-five patients with C. neoformans var. grubii fungemia were enrolled in the study. The Sequential Organ Failure Assessment Score, Glasgow Coma Scale, Charlson comorbidity score, and adequate duration of therapy for amphotericin B were predictors for mortality in univariate analysis. Antifungal susceptibility testing with YeastONE™ does not predict clinical outcomes of C. neoformans var. grubii fungemia. Greater disease severity, high comorbidities, poor consciousness level, and inappropriate treatment were associated with increased mortality in cryptococcemia cases. Lay abstract Cryptococcus neoformans is an encapsulated yeast living in both plants and animals that is composed of three main serotypes: C. neoformans var. grubii, C. neoformans var. gattii, and C. neoformans var. neoformans. C. neoformans var. grubii is the most common disease-causing Cryptococcus species worldwide. C. neoformans var. gattii is more prevalent than C. neoformans var. neoformans in both tropical and subtropical regions of Asia. C. neoformans causes severe, even fatal, diseases such as pulmonary infection, bloodstream infection, skin and soft tissue infection, bone and joint infection, central nervous system infection, and disseminated infection, regardless of host immunocompetence. We conducted a retrospective study on 85 patients who contracted cryptococcemia from January 1, 2009, to December 31, 2016. This work conducted both microbiological and clinical studies involving in vitro susceptibility testing, demographic data, comorbidities, treatment modalities, and treatment outcomes. We utilized a modern medical technique-based instrument, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-ToF MS; Biotyper, Bruker Daltonics, Inc.), which determines the unique proteomic fingerprint of an organism, to identify the C. neoformans serotype. We utilized Thermo Fisher Scientific™ Sensititre™ YeastONE™ colorimetric broth microdilution plates coupled with a Vizion™ Digital MIC Viewing System (a computer-assisted optical reading machine) to determine the in vitro susceptibility of amphotericin B, flucytosine, fluconazole, itraconazole, posaconazole, and voriconazole against 85 C. neoformans var. grubii blood isolates. In conclusion, the susceptibility patterns of these antifungal agents did not correlate significantly with treatment outcomes. However, a lower disease severity score, a lower Glasgow Coma Scale score, fewer comorbidities, and adequate amphotericin B treatment duration were predictors for treatment success in univariate analysis.

The goal of diagnosing the coronavirus disease 2019 (COVID‐19) from suspected pneumonia cases, that is, recognizing COVID‐19 from chest X‐ray or computed tomography (CT) images, is to improve diagnostic accuracy, leading to faster intervention. The most important and challenging problem here is to design an effective and robust diagnosis model. To this end, there are three challenges to overcome: (1) The lack of training samples limits the success of existing deep‐learning‐based methods. (2) Many public COVID‐19 data sets contain only a few images without fine‐grained labels. (3) Due to the explosive growth of suspected cases, it is urgent and important to diagnose not only COVID‐19 cases but also the cases of other types of pneumonia that are similar to the symptoms of COVID‐19. To address these issues, we propose a novel framework called Unsupervised Meta‐Learning with Self‐Knowledge Distillation to address the problem of differentiating COVID‐19 from pneumonia cases. During training, our model cannot use any true labels and aims to gain the ability of learning to learn by itself. In particular, we first present a deep diagnosis model based on a relation network to capture and memorize the relation among different images. Second, to enhance the performance of our model, we design a self‐knowledge distillation mechanism that distills knowledge within our model itself. Our network is divided into several parts, and the knowledge in the deeper parts is squeezed into the shallow ones. The final results are derived from our model by learning to compare the features of images. Experimental results demonstrate that our approach achieves significantly higher performance than other state‐of‐the‐art methods. Moreover, we construct a new COVID‐19 pneumonia data set based on text mining, consisting of 2696 COVID‐19 images (347 X‐ray + 2349 CT), 10,155 images (9661 X‐ray + 494 CT) about other types of pneumonia, and the fine‐grained labels of all. Our data set considers not only a bacterial infection or viral infection which causes pneumonia but also a viral infection derived from the influenza virus or coronavirus.

Visual reasoning is a special kind of visual question answering, which is essentially multi-step and compositional, and also requires intensive text-visual interaction. The most important and challenging problem of visual reasoning is to design an effective and robust visual reasoning model. To this end, there are two challenges to overcome. The first is that textual and visual information must be jointly considered to make accurate inferences about reasoning. The second is that existing deep learning-based works are often too specific to a particular task. To address these issues, we propose a knowledge memory embedding model with mutual modulation for visual reasoning. This approach learns not only knowledge-based embeddings derived from key–value memory network to make the full and joint of textual and visual information, but also exploits the prior knowledge to improve the performance with knowledge-based representation learning for applying other general reasoning tasks. Experimental results on four benchmarks show that the proposed approach significantly improves performance compared with other state-of-the-art methods, guarantees the robustness with our model. Most importantly, we apply our model to four reasoning tasks, and experimentally show that our model effectively supports relational reasoning and improves performance in several tasks and datasets.

Scene understanding comes in many flavors, two of the most popular being object detection and semantic segmentation, which act as two important aspects for scene understanding, and are applied to many areas, such as autonomous driving and intelligent surveillance. Although much progress has already been made, the two tasks of object detection and semantic segmentation are often investigated independently. In practice, scene understanding is complicated, and comprises many sub-tasks, so that research of learning multiple tasks simultaneously with a single model is feasible. With the interrelated goals of these two tasks, there is a strong motivation to improve the object detection accuracy with the help of semantic segmentation, and vice versa. In this paper, we propose a loss-balanced multi-task model for simultaneous object detection and semantic segmentation. We explore multi-task learning with sharing parameters based on deep learning to realize improved object detection and segmentation, and propose a single-stage deep architecture based on multi-task learning, jointly performing object detection and semantic segmentation to boost each other. With no more computation load in the inference compared with the baselines of SSD and FCN, we show that these two tasks, object detection and semantic segmentation, benefit from each other. Experimental results on Pascal VOC and COCO show that our method improves much in object detection and semantic segmentation compared with the corresponding baselines of both tasks.

Face image super-resolution refers to recovering a high-resolution face image from a low-resolution one. In recent years, due to the breakthrough progress of deep representation learning for super-resolution, the study of face super-resolution has become one of the hot topics in the field of super-resolution. However, the performance of these deep learning-based approaches highly relies on the scale of training samples and is limited in efficiency in real-time applications. To address these issues, in this work, we introduce a novel method based on the parallel imaging theory and OpenVINO. In particular, inspired by the methodology of learning-by-synthesis in parallel imaging, we propose to learn from the combination of virtual and real face images. In addition, we introduce a center loss function borrowed from the deep model to enhance the robustness of our model and propose to apply OpenVINO to speed up the inference. To the best of our knowledge, it is the first time to tackle the problem of face super-resolution based on parallel imaging methodology and OpenVINO. Extensive experimental results and comparisons on the publicly available LFW, WebCaricature, and FERET datasets demonstrate the effectiveness and efficiency of the proposed method.

BACKGROUND Abnormal microangiogenesis and microenvironmental disturbances within the Nasopharyngeal carcinoma (NPC) can exacerbate tumor hypoxia, which may increase radiotherapy resistance and thus lead to poor prognosis in NPC patients. A non-invasive imaging technique, dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), which can reflect the tumor blood perfusion and angiogenesis status, was used to investigate the relationships of DCE-MRI parameters with hypoxia-inducible factor 1 alpha (HIF-1α) expression and tumor grades in NPC patients. METHODS 42 treatment-naive patients with pathologically confirmed NPC were enrolled in this analysis. Plain magnetic resonance scans and DCE-MRI scans were performed before treatment, and post-processing was performed to calculate the relative enhancement (RE), maximum relative enhancement (MRE), maximum enhancement (ME), wash-in rate (WIR), wash-out rate (WOR), time to peak (TTP), and area under the curve (AUC). Immunohistochemistry was used to detect HIF-1α expression in electronasopharyngeal fiberoscopic specimens. The clinical grade/stage of NPC was jointly assessed by an experienced radiologist and a radiotherapist. The potential correlations of the DCE-MRI parameters with HIF-1α expression and clinical grades were analyzed. The statistical analysis was performed using SPSS 17.0 software package. RESULTS Among DCE-MRI parameters, RE, ME, and MRE were associated with the positive expression of HIF-lα in NPC and could reflect the hypoxic status in the local microenvironment of the cancer foci in vivo. RE, ME, and MRE were significantly higher in the positive HIF-1α expression group than in the negative HIF-1α expression group (F=5.281, P=0.027; F=11.923, P=0.001; F=6.228, P=0.017). RE, ME, and MRE were significantly correlated with clinical grade (F=3.646, P=0.021; F=3.204, P=0.034, F=3.050, P=0.040) and T stage (F=6.578, P=0.001; F=3.540, P=0.023; F=4.384, P=0.010). The values of RE, MRE, and ME gradually increased as the clinical grade and T stage increased. CONCLUSIONS DCE-MRI is a valuable imaging technique for the noninvasive evaluation of hypoxia in NPC, the development of individualized treatment protocols, and the prediction of efficacy.

Background/Purpose We sought to compare the diagnostic performances of 68Ga-PSMA-11 PET/CT and prostate/whole-abdomen multiparametric magnetic resonance imaging (PWAmpMRI) in Taiwanese patients with biochemically recurrent prostate cancer following robot-assisted radical prostatectomy. Methods Between June 2017 and December 2018, we prospectively enrolled 34 patients. Upon review of all available clinical and imaging data, a best valuable comparator (BVC) was defined on an individual basis in the light of a consensus reached by a multidisciplinary tumor board. Diagnostic positivity was investigated in relation to the different lesion types. Results On a patient-based analysis, 68Ga-PSMA-11 PET/CT and PWAmpMRI showed a moderate agreement (kappa coefficient = 0.62). 68Ga-PSMA-11 PET/CT identified local recurrences, regional, and non-regional lymph node metastases, and bone metastases in 15, 10, 1, and 5 patients, respectively. Conversely, PWAmpMRI detected these lesions in 26, 8, 1, and 4 patients, respectively. When the BVC was used as reference standard, the positive diagnostic rates for local recurrences, regional lymph node metastases, non-regional lymph node metastases, and bone metastases were 57.7%, 90.9%, 100%, and 100%, respectively for 68Ga-PSMA-11 PET/CT, and 100%, 72.7%, 100%, and 80% for PWAmpMRI, respectively. The use of both PWAmpMRI and 68Ga-PSMA-11 PET/CT showed a complete diagnostic yield for detecting both local recurrence and systemic failure when PSA levels reached 0.5 ng/mL. Conclusion Due to urine radioactivity, 68Ga-PSMA-11 PET/CT performs less than PWAmpMRI on local recurrences. However, it can have a complementary diagnostic role in the detection of lymph node metastases and in identifying non-axial bone metastases beyond the PWAmpMRI scanning field.

Background To evaluate the use of a human fibrin glue (Tisseel) for minor bleeding control and approximation of ovarian defect during transvaginal natural orifice ovarian cystectomy (TNOOC) of benign and non-endometriotic ovarian tumors. Methods A total of 125 women with benign and non-endometriotic ovarian tumors who underwent TNOOC between May 2011 and January 2020: 54 with the aid of Tisseel and 71 with traditional suture for hemostasis and approximation of ovarian defect. Surgical outcomes such as length of surgery, operative blood loss, postoperative pain score, and postoperative hospital stay were recorded. Before and immediately (10 days) and at 6 months after the procedure, serum anti-Müllerian hormone (AMH) levels were also determined. Results Complete hemostasis and approximation of ovarian defect were achieved in all cases. No significant difference was noted in the operating time, operative blood loss, postoperative pain scores after 12, 24 and 48 h, length of postoperative stay, and baseline AMH levels between the two groups. The operation did not have a negative effect on the immediate and 6-month postoperative AMH levels in the suture group. However, the decline in the AMH levels was significant immediately after surgery in the Tisseel group, nevertheless, no significant difference was noted in the AMH levels at 6 months (3.3 vs. 1.7 mg/mL; p = 0.042, adjusted p = 0.210). Conclusion The use of Tisseel in TNOOC of benign and non-endometriotic ovarian tumors without suturing the ovarian tissue is clinically safe and feasible.

Human papillomavirus (HPV) is the well‐established etiologic factor for cervical neoplasia. Cervical conization constitutes an effective treatment for high‐grade cervical intraepithelial neoplasia (HG‐CIN). We conducted an observational study for long‐term outcomes and HPV genotype changes after conization for HG‐CIN. Between 2008 and 2014, patients with newly diagnosed HG‐CIN before conization (surveillance new [SN] group) and those who had undergone conization without hysterectomy (surveillance previous [SP] group) were enrolled. HPV testing and Pap smear were performed periodically for the SN and SP (collectively S) groups. All other patients receiving conization for HG‐CIN during the study period were identified from our hospital database. Those eligible but not enrolled into our study were assigned to the non‐surveillance (non‐S) group. For the S group (n = 493), the median follow‐up period was 74.3 months. Eighty‐four cases had recurrent CIN Grade 2 or worse (CIN2+) (5‐year cumulative rate: 14.8%), of which six had invasive cancer. Among the 84 patients, 65 (77.4%) exhibited type‐specific persistence in the paired HPV results, whereas only 7 (8.3%) harbored new HPV types that belonged to the 9‐valent vaccine types. Among the 7397 non‐S patients, 789 demonstrated recurrent CIN2+, of which 57 had invasive cancer. The stages distribution of those progressed to invasive cancer in the non‐S group were more advanced than the S group (P = .033). Active surveillance might reduce the severity of those progressed to cancer. Because a majority of the patients with recurrent CIN2+ had persistent type‐specific HPV infections, effective therapeutic vaccines are an unmet medical need., What's new? High‐grade cervical intraepithelial neoplasia (HG‐CIN), a cervical carcinoma precursor, frequently is caused by high‐risk human papillomavirus (hr‐HPV) infection. HG‐CIN can be treated by cervical conization, though the procedure does not eradicate hr‐HPV, potentially enabling CIN recurrence. Here, in Taiwanese patients, investigation of HPV genotype changes in relation to HG‐CIN status after conization shows that 77.4 percent of patients with recurrent CIN 2 grade or worse (CIN2+) after conization had persistent type‐specific HPV infections. Vaccination against the remaining high‐risk HPVs prevented recurrent CIN2+ in only 8.3 percent of patients, revealing an unmet need for effective therapeutic vaccines.

Background Zheng Qing Feng Tong Ning (ZQFTN) is a sinomenine (SIN) preparation that has been used in clinical practice. Our study aimed to assess the methodological and reporting quality of meta-analyses on the Chinese herbal formula ZQFTN for the treatment of rheumatoid arthritis (RA). Methods Systematic searches were carried out with the 5 following electronic databases from inception to July 2019: China National Knowledge Infrastructure (CNKI), Wanfang, VIP database for Chinese technical periodicals (VIP), Cochrane Library and PubMed. The quality of the methodology and reporting was measured with the assessment of multiple systematic reviews 2 (AMSTAR 2) scale, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and the Grading of Recommendations, Assessment, Development and Evaluation (GRADE). Results Eight studies were identified. Among the 16 items of the AMSTAR 2 scale, four items were optimally reported (“Y” =100% of the items), and another four items were poorly reported (“Y” =0% of the items). Only 2 studies received a good overall score (“Y” ≥50% of the items). Regarding the PRISMA statement, the scores of 5 studies were lower than the average score (17.69), indicating that the quality of the reports was very low. In terms of the GRADE, none of the 61 results were of high quality (0.0%). Fifteen results were of medium quality (25%), 34 were of low quality (55%), and 12 were of very low quality (20%). Among the five downgrading factors, deviation risk (n = 61, 100%) was the most common downgrading factor, followed by inconsistency (n = 30, 50%), publication bias (n = 17, 28%), inaccuracy (n = 11, 18%) and indirectness (n = 0, 0%). Conclusions The methodological and reporting quality of the meta-analyses and systematic reviews in the included studies are less than optimal, and researchers should undergo additional training and follow the AMSTAR 2 scale, PRISMA statement and GRADE to design high-quality studies in the future.

BackgroundPrevious studies have found that elevated copper levels induce oxidation, which correlates with the occurrence of major depressive disorder (MDD). However, the mechanism of abnormal cerebral metabolism of MDD patients remains ambiguous. The main function of the enzyme ATPase copper-transporting alpha (ATP7A) is to transport copper across the membrane to retain copper homeostasis, which is closely associated with the onset of mental disorders and cognitive impairment. However, less is known regarding the association of ATP7A expression in MDD patients.MethodsA total of 31 MDD patients and 21 healthy controls were recruited in the present study. Proton magnetic resonance spectroscopy was used to assess the concentration levels of N-acetylaspartate, choline (Cho), and creatine (Cr) in brain regions of interest, including prefrontal white matter (PWM), anterior cingulate cortex (ACC), thalamus, lentiform nucleus, and cerebellum. The mRNA expression levels of ATP7A were measured using polymerase chain reaction (SYBR Green method). The correlations between mRNA expression levels of ATP7A and/or ceruloplasmin levels and neuronal biochemical metabolite ratio in the brain regions of interest were evaluated.ResultsThe decline in the mRNA expression levels of ATP7A and the increase in ceruloplasmin levels exhibited a significant correlation in MDD patients. In addition, negative correlations were noted between the decline in mRNA expression levels of ATP7A and the increased Cho/Cr ratios of the left PWM, right PWM, and right ACC in MDD patients. A positive correlation between elevated ceruloplasmin levels and increased Cho/Cr ratio of the left PWM was noted in MDD patients.ConclusionsThe findings suggested that the decline in the mRNA expression levels of ATP7A and the elevated ceruloplasmin levels induced oxidation that led to the disturbance of neuronal metabolism in the brain, which played important roles in the pathophysiology of MDD. The decline in the mRNA expression levels of ATP7A and the elevated ceruloplasmin levels affected neuronal membrane metabolic impairment in the left PWM, right PWM, and right ACC of MDD patients.

Background/purpose: Ovarian clear cell carcinoma (OCCC) with recurrence/progression after treatment has dismal prognosis. We aimed to investigate the management and outcomes of such patients. Methods: OCCC patients who were treated between 2000 and 2013 with cancer recurrence or progression after primary treatment were analyzed. Univariate and multivariate analyses were used to identify the independent predictors of survival after recurrence (SAR) and cancer-specific survival (CSS). Results: A total of 64 patients experienced treatment failure (49 recurred after remission and 15 progressed without remission). The 5-year CSS rates of recurrent/progressive OCCC patients were 22.9% (progression group: median CSS 5.9 months [range, 0.8–25.2] vs recurrence group: 43.6 months [range, 7.1–217.8]; p 12 months (HR: 0.22–0.40, p = 0.001 and p = 0.023), CA125

Study Objective This study was designed to evaluate the learning curve of transvaginal natural orifice surgery (NOS) for adnexal tumors based on the type of procedure (adnexectomy or cystectomy). Design Retrospective cohort study. Setting Tertiary-care university hospital. Patients One hundred thirty-six women with adnexal tumors. Interventions A series of 136 consecutive transvaginal NOS for adnexal tumor were performed between April 2011 and June 2016. Eighty-five patients (62.5%) had undergone cystectomy, and 51 patients (37.5%) had undergone adnexectomy. Measurements and Main Results The transvaginal NOS procedures included vaginal, endoscopic, single-port access techniques and were divided into 2 categories: adnexectomy and cystectomy. Operating time was electronically recorded and was defined as the time from the creation of a posterior colpotomy incision until the complete closure of the posterior colpotomy opening. The average operating time for the cystectomy and adnexectomy groups was 42.7 and 37.7 minutes, respectively (p = .015). The learning curve was analyzed using the cumulative summation method and showed that 36 cases were needed to achieve proficiency in transvaginal NOS for ovarian cystectomy. However, there was no significant cut-off point to determine the number of patients who had undergone adnexectomy. Conclusion A well-trained gynecologic endoscopist can achieve surgical proficiency in transvaginal NOS cystectomy after 36 cases. It is suggested that the surgical transvaginal NOS procedure should begin with adnexectomy rather than with cystectomy to avoid initial technical challenges.

Background To evaluate the incidence and outcomes of septostomy in twin-to-twin transfusion syndrome (TTTS) after fetoscopic laser therapy. Methods A retrospective analysis of TTTS postlaser septostomy between 2005 and 2018 was performed. Postlaser septostomy was diagnosed using both (1) a free-floating intertwin membrane flap visible on ultrasound examination and (2) the rapid equalization of amniotic fluid maximum vertical pocket in the donor and recipient amniotic sacs observed after laser therapy. Perinatal survival, neonatal brain image anomaly, gestational age at operation and birth, incidence of premature rupture of membranes (PROM) within 3 weeks after operation, pseudoamniotic band syndrome, and cord entanglement were evaluated. Results In the 159 TTTS cases included, 12 had postlaser septostomy. Relative to the group without septostomy, the septostomy group had a lower total fetal survival rate (54.2% vs 73.6%, p = 0.041), an earlier mean gestational age at delivery (27.8 vs 34.4 weeks, p = 0.009), a higher risk of PROMs within 3 weeks after operation (33.3% vs 5.4%, p = 0.004), a higher cord entanglement rate (16.7% vs 0%, p = 0.005), and a higher brain image anomaly rate (23.0% [3/13] vs 5.0% [11/218], p = 0.035). After considering the severe Quintero stages (stage III and IV), postlaser septostomy was the only variable [p = 0.003, odds ratio = 5.1] to predict neonatal brain image anomaly. Postlaser septostomy combined with severe Quintero stages could predict PROMs within 3 weeks after laser therapy [p = 0.001, odds ratio = 14.1 and p = 0.03, odds ratio = 5.4, respectively] and delivery before the gestational age of 28 weeks [p = 0.017, odds ratio = 4.5 and p = 0.034, odds ratio = 2.3, respectively]. The risk of pseudoamniotic band syndrome was not increased by postlaser septostomy in this case series. Conclusions Postlaser septostomy in TTTS was associated with poorer fetal survival and more adverse perinatal outcomes even after considering severe Quintero stages before laser therapy. Efforts should be made to prevent septostomy during laser therapy, and septostomy as the primary method to treat TTTS is not advisable.

Public concerns on image encryption grow significantly as the development and application of edge computing and the Internet of Things intensified recently. However, most existing image cryptosystems are not sophisticated enough to resist the two major attack strategies available currently, that is: 1) differential attacks and 2) chosen-plaintext attacks, which are famous for their destructive power, especially their capability of exploiting cryptosystems' features to recover the secret key. In this article, we propose an artificial image, computational experiment, and parallel execution (ACP)-based color image encryption approach using redundant blocks. First, a redundant blocks strategy with redundant spaces is proposed to prevent differential attacks and accelerate operating speed while guaranteeing the security of image cryptosystems. Second, real-world chaotic data (e.g., stock data) are obtained to generate artificial images and conduct computational experiments. Furthermore, artificial images are encrypted via real-world chaos, while the original images are encrypted via simulated chaos (such as Chen's hyperchaos). Finally, we design the process of parallel execution for image encryption and use DNA XOR to merge two groups of encrypted subimages to fuse the effect of the chaotic characteristics in both the real world and the simulation. The final encrypted image is realized through the recovery of redundant blocks. The ACP mechanism of color image encryption achieves the goal of improving the sophistication of chaos-based cryptosystems and resists both the differential and chosen-plaintext attacks. Experimental results and security analysis show that our approach provides not only excellent encryption but also security sufficient to prevent known attacks.

The α7-human papillomavirus (HPV)-related cervical squamous cell carcinoma (SCC) is associated with poor prognosis. We compared the genomic profiles of this disease in a cohort corresponding to the 2001–2014 period with various responses to radiotherapy or concurrent chemoradiation through microRNA (miR) profiling involving miR 4.0 array and human transcriptome array 2.0 analyses. A real-time quantitative polymerase chain reaction was then conducted to identify the predictive biomarkers. A significantly lower expression of miR143-3p in recurrent tumors (p = 0.0309) relative to that in nonrecurrent tumors was observed. The miR143-3p targeted the mRNA expression of the baculoviral inhibitor of the apoptosis protein (IAP) repeat-containing 2 (BIRC2, p = 0.0261). The BIRC2 protein levels (p = 0.0023) were significantly higher in recurrent tumors than in nonrecurrent tumors. Moreover, the miR-143-3p sensitized the response of α7-HPV-related cervical SCC to chemotherapy by targeting BIRC2. A combination of BIRC2-inhibitor LCL161 and topotecan exerted synergistic effects on cancer cells and animal tumor models. In a pooled cohort of α7-HPV-related cervical SCC (including mixed infections with non-α7-HPV) treated between 1993 and 2014, high BIRC2 expression was associated with significantly worse outcomes (cancer-specific survival, hazard ratio (HR) = 1.42, p = 0.008, progression-free survival, HR = 1.64, p = 0.005). Summarily, BIRC2 constitutes a novel prognostic factor and therapeutic target for α7-HPV-related cervical SCC.

Person search often requires a query photo of the target person. However, in many practical scenarios, there is no guarantee that such a photo is always available. In this paper, we define the problem of sketch based person search, which uses a sketch instead of a photo as the probe for retrieving. We tackle this problem in a weak supervision setting and propose a clustering and feature attention based weakly supervised learning framework, which contains two stages of pedestrian detection and sketch based person re-identification. Specially, we introduce multiple detectors, followed by fuzzy c-means clustering to achieve weakly supervised pedestrian detection. Moreover, we design an attention module to learn discriminative features in subsequent re-identification network. Extensive experiments show the superiority of our method.

This paper deals with the challenging problem of building visual reasoning models for answering questions related to artworks in artistic domains. The nature of abstract styles and cultural contexts within an artistic image makes the corresponding learning tasks extremely difficult. We propose a novel framework termed as Hierarchical-Knowledge Embedded Meta-Learning to address the critical issues of visual reasoning in artistic domains. In particular, we firstly present a deep relational model to capture and memorize the relations among different samples. Then, we provide the hierarchical-knowledge embedding that mines the implicit relationship between question-answer pairs for knowledge representation as the guidance of our meta-learner. This is a case of "knowledge is power" in the sense that the hierarchical knowledge representation is incorporated into our meta-learning based model. The final classification is derived from our model by learning to compare the features of samples. Experimental results show that our approach achieves significantly higher performance compared with other state-of-the-arts.

Expansive soil is a kind of soil with special properties, widely distributed, difficult treatment technology, and affecting the engineering safety of banks and dams. It is particularly necessary to summarize the development status and trend of the monitoring and warning system of banks and dams for expansive soil based on the characteristics of expansion soil. Based on extensive literatures, the progress of the hardware and software were systematically analyzed for the monitoring and warning system of banks and dams for expansive soil. Specially, the hardware mainly includes: sensors, data acquisition, data transmission, and postprocessing platform and so on; The software mainly includes: multi-sensor integration, data sharing, post-processing, human-computer interface, and route planning of unmanned aerial vehicle and so on. Accordingly, the research directions of the hardware and software of monitoring and early warning system are put forward, and the results will be used for references in the development of monitoring and warning system, engineering construction, engineering management, and emergency management and so on for banks and dams of expansive soil.

Purpose. To evaluate Humphrey visual field (VF) progression rate before and after cataract surgery in patients with open-angle glaucoma (OAG) and angle-closure glaucoma (ACG). Methods. After reviewing all records in our Humphrey VF database from July 1, 2001, to December 31, 2018, eyes of OAG and ACG that had received uncomplicated phacoemulsification with intraocular lens implantation during the period and had been followed up regularly for more than one year with ≥3 reliable VF tests before and after cataract surgery, respectively, were enrolled. The VF progression rate, visual acuity, and intraocular pressure (IOP) before and after cataract surgery were compared. Results. Seventy-seven eyes (OAG: 51; ACG: 26) from 60 patients were enrolled. The mean preoperative and postoperative follow-up durations were 4.89 ± 2.70 and 5.48 ± 2.30 years in the OAG group and 5.14 ± 3.31 and 5.97 ± 2.35 years in the ACG group. IOP and visual acuity significantly improved postoperatively in both groups. In the OAG group, there was no significant change in the VF progression rate (pre-op: −0.39 ± 0.94 dB/year; post-op: −0.47 ± 0.72 dB/year) ( P = 0.619 ), whereas, in the ACG group, the rate significantly flattened after surgery from −1.39 ± 1.31 dB/year to −0.34 ± 0.64 dB/year ( P = 0.003 ). Conclusions. The VF progression rate did not differ between preoperative and postoperative eyes with OAG, but those with ACG exhibited rapid VF decline preoperatively, which was significantly flattened postoperatively.

Mask wearing has been considered as an effective measure to prevent the spread of COVID-19 during the current pandemic. However, most advanced face recognition approaches are not adequate for masked face recognition, particularly in dealing with the issue of training through the datasets covering only a limited number of images with ground-truth labels. In this work, we propose to learn from the large scale of web images and corresponding tags without any manual annotations along with limited fully annotated datasets. In particular, inspired by the recent success of webly supervised learning in deep neural networks, we capitalize on readily-available web images with noisy annotations to learn a robust representation for masked faces. Besides, except for the conventional spatial representation learning, we propose to leverage the power of frequency domain to capture the local representative information of un-occluded facial parts. This approach learns robust feature embeddings derived from our feature fusion architecture to make joint and full use of information from both spatial and frequency domains. Experimental results on seven benchmarks show that the proposed approach significantly improves the performance compared with other state-of-the-art methods.

Human can naturally understand scenes in depth with the help of various knowledge accumulated and by a comprehensive visual concept organization including category labels and different-level attributes. This inspires us to unify professional knowledge at different levels with deep neural network architectures progressively for scene understanding. Different from the general embedding approaches, we construct different knowledge graphs for different levels of vision tasks by organizing the rich visual concepts accordingly. We employ a gated graph neural network and relational graph convolutional networks to propagate node messages for different levels of tasks and generate progressively different levels of knowledge representation through the graph. Compared with existing methods, our framework has a main appealing property leading to a novel progressive knowledge-embedded representation learning framework that incorporates different level knowledge graphs into the learning of networks at corresponding level. Extensive experiments on the widely used Broden+ dataset demonstrate the superiority of the proposed framework over other existing state-of-the-art methods.

Simple Summary Uterine leiomyosarcoma is an aggressive tumor and the current staging system cannot differentiate the patients into different prognostic groups. This leads to difficulty in predicting the patients’ outcomes and planning for adjuvant therapy. We aimed to develop a prediction model that can predict the chance of survival by the third year. In this article, we had used different statistical tests to identify five readily available clinicopathologic parameters to build the prediction model. Internal validation was performed with satisfactory accuracy. Such a prediction model might help to predict survival outcome, and guide future research on the treatment modality. Abstract Background: The existing staging systems of uterine leiomyosarcoma (uLMS) cannot classify the patients into four non-overlapping prognostic groups. This study aimed to develop a prediction model to predict the three-year survival status of uLMS. Methods: In total, 201 patients with uLMS who had been treated between June 1993 and January 2014, were analyzed. Potential prognostic indicators were identified by univariate models followed by multivariate analyses. Prediction models were constructed by binomial regression with 3-year survival status as a binary outcome, and the final model was validated by internal cross-validation. Results: Nine potential parameters, including age, log tumor diameter, log mitotic count, cervical involvement, parametrial involvement, lymph node metastasis, distant metastasis, tumor circumscription and lymphovascular space invasion were identified. 110 patients had complete data to build the prediction models. Age, log tumor diameter, log mitotic count, distant metastasis, and circumscription were significantly correlated with the 3-year survival status. The final model with the lowest Akaike’s Information Criterion (117.56) was chosen and the cross validation estimated prediction accuracy was 0.745. Conclusion: We developed a prediction model for uLMS based on five readily available clinicopathologic parameters. This might provide a personalized prediction of the 3-year survival status and guide the use of adjuvant therapy, a cancer surveillance program, and future studies.

Virtual simulation testing is becoming indispensable for the intelligence testing of intelligent vehicles. However, even the most advanced simulation software provides rather limited test conditions. In the long run, intelligent vehicles are expected to work at SAE (Society of Automotive Engineers) level 4 or level 5. Researchers should make full use of virtual simulation scenarios to test the visual intelligence algorithms of intelligent vehicles under various imaging conditions. In this paper, we create realistic artificial scenes to simulate the self-driving scenarios, and collect a dataset of synthetic images from the virtual driving scenes, named “ParallelEye-CS”. In the artificial scenes, we can flexibly change environmental conditions and automatically acquire accurate and diverse ground-truth labels. As a result, ParallelEye-CS has six ground-truth labels and includes twenty types of tests, which are divided into normal, environmental, and difficult tasks. Furthermore, we utilize ParallelEye-CS in combination with other publicly available datasets to conduct experiments for visual object detection. The experimental results indicate that: 1) object detection algorithms of intelligent vehicles can be tested under various scenario challenges; 2) mixed dataset can improve the accuracy of object detection algorithms, but domain shift is a serious issue worthy of attention.

Background Several heart failure (HF) risk models exist, however, most of them perform poorly when applied to real-world situations. This study aimed to develop a convenient and efficient risk model to identify patients with high readmission risk within 90 days of HF. Methods A multivariate logistic regression model was used to predict the risk of 90-day readmission. Data were extracted from electronic medical records from January 1, 2017 to December 31, 2017 and follow-up records of patients with HF within 3 months after discharge. Model performance was evaluated using a receiver operating characteristic curve. All statistical analysis was done using R version 3.5.0. Results A total of 350 patients met the inclusion criterion of being readmitted within in 90 days. All data sets were randomly divided into derivation and validation cohorts at a 7/3 ratio. The baseline data were fairly consistent among the derivation and validation cohorts. The variables most clearly related to readmission were logarithm of serum N-terminal pro b-type natriuretic peptide (NT-proBNP) level, red cell volume distribution width (RDW-CV), and Charlson comorbidity index (CCI). The model had good discriminatory ability (C-statistic = 0.73). Conclusions We developed and validated a multivariate logistic regression model to predict the 90-day readmission risk for Chinese patients with HF. The predictors included in the model are derived from electronic medical record (EMR) admission data, making it easier for physicians and pharmacists to identify high-risk patients and tailor more intensive precautionary strategies.

According to the AJCC third to seventh edition staging manuals (1988-2010), the presence of through cortex and/or skin invasion in oral cavity squamous cell carcinoma (OCSCC) identifies T4a tumors. The AJCC eighth edition (2018) introduced a depth of invasion (DOI) 20 mm as a criterion for pT4a. Subsequently, a revision maintained that tumors 4 cm with a DOI 10 mm should be classified as pT4a. We sought to analyze the prognostic impact of the three distinct criteria identifying pT4a disease.We examined 667 consecutive patients with pT3-4 buccal/gum/hard palate/retromolar SCC who underwent surgery between 1996 and 2016. pT1/pT2 (n = 108/359) disease were included for comparison purposes.The 5-year outcomes of patients with pT1/pT2/without (n = 406)/with tumor 4 cm/DOI 10 mm (n = 261), pT1/pT2/DOI ≤ 20 mm (n = 510)/ 20 mm (n = 157), and pT1/pT2/without (n = 305)/with through cortex/skin invasion (n = 362) were as follows: disease-specific survival (DSS), 98%/89%/79%/65%, p 0.001, 98%/89%/78%/59%, p 0.001, and 98%/89%79%/69%, p 0.001; overall survival (OS), 90%/79%/63%/51%, p 0.001, 90%/79%/63%/42%, p 0.001, and 90%/79%/65%/52%, p 0.001. In pT3-4 disease, a tumor 4 cm/DOI 10 mm was an independent adverse prognosticator for 5-year DSS rate, DOI 20 mm was an independent adverse prognosticator for 5-year DSS and OS rates, whereas through cortex/skin invasion independently predicted 5-year OS rates.All of the three criteria (tumor 4 cm/DOI 10 mm, DOI 20 mm, and through cortex/skin invasion) identify high-risk patients, which should be reflected in further revisions of pT4a classification in OCSCC.

In Singapore, officers of the border security agency are required to maintain 24 h, 7 days a week operation across various immigration checkpoints. As the day-to-day workload of a Singapore border security officer is physically demanding and operationally challenging, ensuring the highest level of operational vigilance is vital to operation success in border security operations. In this paper, ‘operational vigilance’ is defined as exercising cognitive agility and sustained attention for the duration of a work shift in a border security operational setting. To enhance and maintain operations in border security checkpoints, this paper explores the drivers of operational vigilance in border security from the Singapore experience. Twelve veteran border security officers participated in this study as subject matter experts from which a thematic analysis was used to extract insightful themes central to operational vigilance in a border security context. Findings from this study are used to provide a deeper understanding of drivers that affect operational vigilance in border security in the Singapore context and used as reference for enhancement of border security operations. Limitations and future directions were discussed.

Dataset plays an essential role in the training and testing of traffic vision algorithms. However, the collection and annotation of images from the real world is time-consuming, labor-intensive, and error-prone. Therefore, more and more researchers have begun to explore the virtual dataset, to overcome the disadvantages of real datasets. In this paper, we propose a systematic method to construct large-scale artificial scenes and collect a new virtual dataset (named “ParallelEye”) for the traffic vision research. The Unity3D rendering software is used to simulate environmental changes in the artificial scenes and generate ground-truth labels automatically, including semantic/instance segmentation, object bounding boxes, and so on. In addition, we utilize ParallelEye in combination with real datasets to conduct experiments. The experimental results show the inclusion of virtual data helps to enhance the per-class accuracy in object detection and semantic segmentation. Meanwhile, it is also illustrated that the virtual data with controllable imaging conditions can be used to design evaluation experiments flexibly.

BACKGROUND The female genital tract is an uncommon site of involvement for extra-genital malignancies. Ovarian metastases have been described as disseminations of lung adenocarcinoma; rare cases of secondary localizations in the cervix, adnexa, and vagina have also been reported in the literature. Here, we report two cases of advanced lung adenocarcinoma with female genital tract metastasis. CASE SUMMARY The first case was a 41-year-old woman with stage IV lung adenocarcinoma metastasizing to the cervix. Immunohistochemistry of the cervical biopsy specimen revealed thyroid transcription factor (TTF)-1(+), cytokeratin (CK)-7(+), and (CK)-20(-). Gene mutational analysis showed epidermal growth factor receptor (EGFR) L858R mutation in exon 21. She had a positive response to gefitinib, for both the pulmonary mass and cervical neoplasm. The second case was a 29-year-old woman who was diagnosed with stage IV lung adenocarcinoma with EGFR mutation. After 12 mo of treatment with icotinib, ovarian biopsy showed adenocarcinoma with CDX2(-), TTF-1(+++), PAX8(-), CK-7(+++), CK-20(++), and Ki67(15%+), accompanied with EGFR 19-del mutation and T790M mutation. CONCLUSION Immunohistochemistry and gene mutational testing have greatly helped in locating the initial tumor site when both pulmonary and female genital tract neoplasms exist.

Background Prosthetic joint infections may lead to failures of total joint arthroplasty. Radionuclide imaging can play a diagnostic role in identifying such infections, which require two-stage exchange arthroplasty (instead of simple revision surgery performed in non-infected cases). Although 18F-FDG PET/CT has emerged as a novel diagnostic tool in this setting, the clinical usefulness of 68Ga-citrate PET/CT has not been previously investigated. This single-center prospective study was designed to address this issue. Methods Between January 2016 and October 2017, we examined 34 patients with clinically proven or suspected prosthetic hip/knee joint infections scheduled to undergo surgery. All patients underwent 68Ga-citrate PET/CT scans and sequential 18F-FDG PET/CT imaging for comparative purposes. Intraoperative findings and the results of microbiological analyses of surgical specimens served as gold standard. The diagnostic results were examined according to (1) image interpretation based on radiotracer uptake patterns and (2) quantitative analysis using volumes of interest (VOIs) to calculate standard uptake values (SUVs) and metabolic volumes (MVs). Results A total of 26 (76%) patients were diagnosed as having infections. Based on radiotracer uptake pattern criteria, the sensitivity, specificity, and accuracy of 68Ga-citrate PET/CT and 18F-FDG PET/CT were 92%, 88%, and 91% and 100%, 38%, and 85%, respectively. MV was significantly higher in the infected group when 68Ga-citrate PET/CT was used (422.45 vs. 303.65 cm3, p = 0.027), whereas no significant differences were observed on 18F-FDG PET/CT. According to receiver operating characteristic (ROC) curve analysis, a cut-off value of 370.86 for MV resulted in a sensitivity of 61.5% and a specificity of 87.5% (area under curve: 0.75, 95% confidence interval: 0.57–0.88, p = 0.035). Conclusions Subject to future confirmation, our data provide preliminary evidence that 68Ga-citrate PET/CT may have a complimentary role to 18F-FDG PET/CT in detecting prosthetic joint infections, being characterized by a higher specificity and the possibility to discriminate between an infectious condition and sterile inflammation. Trial registration This prospective study was registered at clinicaltrials.gov (registration number: NCT02855190).

Background/purpose: Bacteremia portends high rates of morbidity and mortality. Although 18F-fluorodeoxyglucose positron emission tomography and computed tomography (18F-FDG PET/CT) imaging has clinical value in assessing fever of unknown origin, its usefulness in bacteremia has not been entirely elucidated. We therefore designed the current single-center retrospective study to investigate 1) the clinical value of 18F-FDG PET/CT imaging in assessing bacteremia and 2) the association between laboratory data and imaging findings. Methods: We examined 102 patients with bacteremia who had undergone 18F-FDG PET/CT imaging. The patients’ clinical and laboratory data were reviewed and analyzed in relation to 18F-FDG PET/CT findings. Patients showing positive results underwent quantitative measurements of 18F-FDG uptake. Results: Positive 18F-FDG PET/CT findings were identified in 74 (72.5%) patients, and 40 (54.1%) underwent modified treatment or management because of the imaging results (p = 0.003). Positive 18F-FDG PET/CT findings were significantly associated with higher white blood cell (WBC) counts and C-reactive protein (CRP) levels (p = 0.012 and 54.025 mg/L were accurate predictors of positive 18F-FDG PET/CT results. Keywords: Bacteremia, C-reactive protein, 18F-fluorodeoxyglucose positron emission tomography and computed tomography

The subcutaneous implantable cardioverter-defibrillator (S-ICD) may provide comparable protection while avoiding the disadvantage of transvenous lead, but the abnormal features of the hypertrophic cardiomyopathy (HCM) electrocardiogram (ECG) make it a challenge for S-ICD template screening. We aimed to investigate S-ICD eligibility according to the S-ICD manufacturer's surface ECG screening template in China, and further analyze its corresponding ineligible predicting factors in 12-lead suface ECG. A total of 179 HCM patients (114 males; mean age: 45 ± 14 years) underwent S-ICD screening at rest and on exercise, among which 91 patients (50.8%) were eligible for S-ICD. Among the patients who passed screening, 43 (47.3%) had 3 vectors eligibility; 64 (70.3%) screening qualified on both sides; 10 patients (11.0%) passed the screening while the electrodes located only on the left parasternal line versus 17 patients (18.7%) moved to the right line. The secondary sensing vector (Lead III) was mostly appropriate (53.6%), followed by the primary sensing vector (lead II, 53.1%) and the alternate sensing vector (Lead I, 46.9%). Higher R wave was the major cause, accounted for 70.5%, for screening failure. There existed significant difference in T wave in lead II, aVF, V5 and V6, adds R/T ratio in lead V5 and V6, between the screening success group (group A) and screening failure group (group B) at rest and on exercise. A multivariable logistic regression analysis was performed to identify that R/T ≤ 3.5 in lead V5 was the independent factor to predict the screening ineligibility, with odds ratio 3.648. S-ICD screening success is 50.8% in HCM patients, which is much lower than that in other studies. R/T ≤ 3.5 in lead V5 in 12-lead surface ECG was an independent predicting factor for screening failure.

Background: Despite wide application of high frequency oscillatory ventilation (HFOV) in neonates with respiratory distress, little has been reported about its rescue use in preterm infants. We aimed to evaluate the therapeutic effects of HFOV in preterm neonates with refractory respiratory failure and investigate the independent risk factors of in-hospital mortality.Methods: We retrospectively analyzed data collected prospectively (January 2011-December 2018) in four neonatal intensive care units of two tertiary-level medical centers in Taiwan. All premature infants (gestational age 24-34 weeks) receiving HFOV as rescue therapy for refractory respiratory failure were included.Results: A total of 668 preterm neonates with refractory respiratory failure were enrolled. The median (IQR) gestational age and birth weight were 27.3 (25.3-31.0) weeks and 915.0 (710.0-1380.0) g, respectively. Pre-HFOV use of cardiac inotropic agents and inhaled nitric oxide were 70.5% and 23.4%, respectively. The oxygenation index (OI), FiO2, and AaDO2 were markedly increased after HFOV initiation (all p < 0.001), and can be decreased within 24-48 hours (all p < 0.001) after use of HFOV. 375 (56.1%) patients had a good response to HFOV within 3 days. The final in-hospital mortality rate was 34.7%. No association was found between specific primary pulmonary disease and survival in multivariate analysis. We found preterm neonates with gestational age < 28 weeks, occurrences of sepsis, severe hypotension, multiple organ dysfunctions, initial higher severity of respiratory failure and response to HFOV within the first 72 hours were independently associated with final in-hospital mortality.Conclusions: The mortality rate of preterm neonates with severe respiratory failure remains high after rescue HFOV treatment. Aggressive therapeutic interventions to treat sepsis and prevent organ dysfunctions are the suggested strategies to optimize outcomes.

pStage IVB oral cavity squamous cell carcinoma (OCSCC) is defined as either pT4b or pN3 disease. We sought to devise an improved prognostic stratification of this patient group.Between December 2003 and January 2018, we retrospectively reviewed the clinical records of 1331 consecutive patients with OCSCC who received tumor excision and neck dissection. The number of patients with pT4a/pT4b, pT1N3b/pT2N3b/pT3N3b/pT4N3b, and pStage IVA/IVB was 370/83, 3/49/42/142, and 332/295, respectively.The 5-year rates of disease-free survival (DFS) and disease-specific survival (DSS) for patients with pT4a/pT4b disease were 64%/63% (p = 0.973) and 72%/69% (p = 0.672), respectively. The 5-year DFS and DSS rates for patients with pT1N3b/pT2N3b/pT3N3b/pT4N3b disease were 67%/65%/40%/42% (p 0.001; pT1-2N3b versus pT3-4N3b, p = 0.002) and 100%/68%/45%/49% (p 0.001; pT1-2N3b versus pT3-4N3b, p = 0.002), respectively. We devised a new definition for pStage IV by considering patients with pT4bN0-2 and pT1-2N3b diseases as pStage-IVA. The number of patients with pStage IVA/IVB (pT3-4N3b) was 443/184. The 5-year rates of AJCC pStage IVA/IVB and the newly proposed pStage IVA/IVB (pT3-4N3b) were as follows: DFS, 74%/52% and 72%/42%; DSS, 83%/58% and 81%/47%; respectively, all p value 0.001.The clinical outcomes of pT4b and pT4a OCSCC are similar. However, patients with pT3-4N3b disease have a less favorable 5-year prognosis compared with cases with pT1-2N3b. In light of the unfavorable outcomes, pT3-4N3b disease should continue to be classified as pStage IVB. Conversely, pT4bN0-2 and pT1-2N3b diseases portend a less adverse prognosis and should therefore be downstaged to pStage IVA.

Precise risk stratification in lymphadenectomy is important for patients with endometrial cancer (EC), to balance the therapeutic benefit against the operation-related morbidity and mortality. We aimed to investigate added values of computer-aided segmentation and machine learning based on clinical parameters and diffusion-weighted imaging radiomics for predicting lymph node (LN) metastasis in EC. This prospective observational study included 236 women with EC (mean age ± standard deviation, 51.2 ± 11.6 years) who underwent magnetic resonance (MR) imaging before surgery during July 2010–July 2018, randomly split into training (n = 165) and test sets (n = 71). A decision-tree model was constructed based on mean apparent diffusion coefficient (ADC) value of the tumor (cutoff, 1.1 × 10−3 mm2/s), skewness of the relative ADC value (cutoff, 1.2), short-axis diameter of LN (cutoff, 1.7 mm) and skewness ADC value of the LN (cutoff, 7.2 × 10−2), as well as tumor grade (1 vs. 2 and 3), and clinical tumor size (cutoff, 20 mm). The sensitivity and specificity of the model were 94% and 80% for the training set and 86%, 78% for the independent testing set, respectively. The areas under the receiver operating characteristics curve (AUCs) of the decision-tree was 0.85—significantly higher than the mean ADC model (AUC = 0.54) and LN short-axis diameter criteria (AUC = 0.62) (both p &lt, 0.0001). We concluded that a combination of clinical and MR radiomics generates a prediction model for LN metastasis in EC, with diagnostic performance surpassing the conventional ADC and size criteria.

Objective The first Poly (adenosine diphosphate [ADP]-ribose) polymerase inhibitor, olaparib, was approved by Taiwan Food and Drug Administration in June 2018, which was available under a compassionate use program since 2015. This study aims to report the early experience of the effectiveness and adverse effects of olaparib in recurrent ovarian cancer patients in Taiwan. Materials and methods This retrospective study enrolled patients with recurrent epithelial ovarian and peritoneal cancer who received olaparib as maintenance therapy or salvage therapy between December, 2015 and October, 2019. We observed response rates in the salvage therapy group, and progression-free interval (PFI) in both groups. Results A total of 20 patients (10 in maintenance and 10 in salvage groups) were enrolled. BRCA status was checked in 18 patients by blood or tumor samples, and 83.3% were mutated (n = 15), including pathological/probable pathological variants in BRCA1 (n = 11), BRCA2 (n = 2), or both BRCA1/BRCA2 (n = 2). In the salvage group, there were two partial responses and two stable diseases, adding up to a clinical benefit rate at 40%. In the maintenance group, median PFI was 20.1 months (range, 1.0–33.1). The median PFI of those with chemotherapy-free interval >12 months was not reached, which was significantly better than those ≤12 months, with median PFI 3.1 months (p = 0.022). The most common grade 3/4 adverse effects in patient with olaparib as monotherapy were neutropenia (30.8%) and fatigue (7.7%). Anemia of grade 1/2 was noted in 76.9%. Conclusion This real-world experience of olaparib for recurrent ovarian cancer in Taiwan showed efficacy and safety similar to the results of previous clinical trial.

Drug exposure impairs cortical plasticity and motor learning, which underlies the reduced behavioral flexibility in drug addiction. Physical exercise has been used to prevent relapse in drug rehabilitation program. However, the potential benefits and molecular mechanisms of physical exercise on drug-evoked motor-cortical dysfunctions are unknown. Here we report that 1-week treadmill training restores cocaine-induced synaptic deficits, in the form of improved in vivo spine formation, synaptic transmission, and spontaneous activities of cortical pyramidal neurons, as well as motor-learning ability. The synaptic and behavioral benefits relied on de novo protein synthesis, which are directed by the activation of the mechanistic target of rapamycin (mTOR)-ribosomal protein S6 pathway. These findings establish synaptic functional restoration and mTOR signaling as the critical mechanism supporting physical exercise training in rehabilitating the addicted brain.

PurposeTo elucidate the usefulness of intravoxel incoherent motion (IVIM)/apparent diffusion coefficient (ADC) parameters in preoperative risk stratification using International Society of Urological Pathology (ISUP) grades.Materials and MethodsForty-five prostate cancer (PCa) patients undergoing radical prostatectomy (RP) after prostate multiparametric magnetic resonance imaging (mpMRI) were included. The ISUP grades were categorized into low-risk (I-II) and high-risk (III-V) groups, and the concordance between the preoperative and postoperative grades was analyzed. The largest region of interest (ROI) of the dominant tumor on each IVIM/ADC image was delineated to obtain its histogram values (i.e., minimum, mean, and kurtosis) of diffusivity (D), pseudodiffusivity (D*), perfusion fraction (PF), and ADC. Multivariable logistic regression analysis of the IVIM/ADC parameters without and with preoperative ISUP grades were performed to identify predictors for the postoperative high-risk group.ResultsThirty-two (71.1%) of 45 patients had concordant preoperative and postoperative ISUP grades. Dmean, D*kurtosis, PFkurtosis, ADCmin, and ADCmean were significantly associated with the postoperative ISUP risk group (all p < 0.05). Dmean and D*kurtosis (model I, both p < 0.05) could predict the postoperative ISUP high-risk group with an area under the curve (AUC) of 0.842 and a 95% confidence interval (CI) of 0.726–0.958. The addition of D*kurtosis to the preoperative ISUP grade (model II) may enhance prediction performance, with an AUC of 0.907 (95% CI 0.822–0.992).ConclusionsThe postoperative ISUP risk group could be predicted by Dmean and D*kurtosis from mpMRI, especially D*kurtosis. Obtaining the biexponential IVIM parameters is important for better risk stratification for PCa.

Rhizoma Coptidis is a popular phytomedicine for the treatment of type 2 diabetes in Asia, but its effective dose for diabetes treatment remains confused because of diverse origins. This study aimed to investigate the dose-response effects of Rhizoma Coptidis extract granules (RCEG), produced with standardized quality control, on hypoglycemic effects in patients with type 2 diabetes. We conducted a retrospective analysis of Chang Gung Research Database from January 01, 2008 to November 30, 2017. Outpatients visiting traditional Chinese medicine clinics and receiving RCEG for type 2 diabetes treatment were included. Plasma glucose, lipid, and other parameters were analyzed from 93 patients with a total of 737 visits within 60 weeks. Scatter plots with the LOESS analysis were used to explore the association between RCEG dose and hypoglycemic effect. The minimal effective dose was chosen to divide the study population into the high-dose and low-dose RCEG groups. Non-parametric tests were used for between-group and within-group comparisons. The multivariate nonlinear mixed-effects model was applied to access the effect of treatment length and groups simultaneously on the change of HbA1c and fasting plasma glucose. The “arule” package in R was used to present the network diagram of RCEG and other co-prescriptions. We first discovered a significant relationship between RCEG dose and HbA1c reduction when the dose reached 0.08 g/kg/day or higher. We thus defined 0.08 g/kg/day of RCEG as the minimum effective dose, and a threshold to separate patients into the high-dose (≥0.08 g/kg/d) and low-dose (

Adversarial images are commonly viewed negatively for neural network training. Here we present an opposite perspective: adversarial images can be used to investigate the problem of classifying adversarial images themselves. To this end, we propose a novel framework termed as Deep Negative Correlation Meta-Learning. In particular, we present a deep relation network to capture and memorize the relation among different samples. We further formulate a deep negative correlation learning, and design a novel meta-learning-based classifier to prevent overfitting and learn discriminative features. The final classification is derived from our relation network by learning to compare the features of samples. Experimental results demonstrate that our approach achieves significantly higher performance compared with other state-of-the-arts.

At present, the online multi-module educational administration system is poorly targeted, resulting in low retrieval accuracy. In order to solve the above problems, a new online multi-module educational administration system is designed, and the hardware and software of the system are designed. The hardware of the system consists of a computer carrier and its supporting peripherals, an information collector, a central data processor and communication equipment. The internal storage recording instrument of the hardware is equipped with an FTP server, and the network can view, download and delete data at any time. The remote control can also control the operation status of the collection and storage tasks. Mainly through Tomcat 7.0 server, using the Java language to compile software program code, build a hierarchical framework system to achieve software work. The experimental results show that the designed online multi-module educational administration system has stronger information retrieval ability and higher retrieval precision.

To investigate outcomes and morbidity of patients undergoing secondary cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) in recurrent ovarian cancer.Between April 2014 and January 2019, a total of 51 recurrent ovarian cancer patients receiving secondary CRS and HIPEC were retrospectively reviewed.Among the 51 patients, median peritoneal cancer index score was 13 (range 3-34), and completeness of cytoreduction (CC) score of 0/1 was achieved in 41 patients (78.8%). Regimen of HIPEC included cisplatin and paclitaxel in 39 (75%) cases. The median follow-up duration of survivors was 20.2 months. Sixteen (30.8%) patients remained free of recurrence after HIPEC. The median progression-free survival (PFS) and overall survival (OS) were 11.8 months and 34.5 months respectively. Multivariate analysis showed previous chemotherapylt;2 lines (HR 0.24, 0.11-0.52; p = 0.001), chemotherapy-free interval ≥6 months (HR 0.19, 0.09-0.37; p lt; 0.001) and CA125 lt; 35 U/mL before HIPEC (HR 0.133, 0.021-0.0832; p = 0.031) were good prognostic factors for PFS. CC0/1 was not significant in multivariate analysis. The most common grade 3/4 toxicity was anemia (17.3%), pleural effusion (11.5%) and renal insufficiency (5.7%). Patients with age ≥50, peritoneal carcinomatosis index (PCI) ≥ 11, operation time ≥10 h and diaphragm surgery had significantly higher incidence of pleural effusion.The current study showed adding HIPEC to secondary CRS might prolong PFS especially in patients with previous chemotherapylt;2 lines, chemotherapy-free interval ≥6 months and CA125 lt; 35 U/mL before HIPEC.

ObjectivesTo determine whether there is serum vitamin D deficiency and the low levels of serum vitamin D are correlated with serological and immunological indexes in patients with idiopathic inflammatory myopathy (IIM).MethodsA total of 63 newly diagnosed patients with IIM, and 55 age- and sex- matched healthy controls were enrolled. Serum levels of 25-(OH)-D were measured by enzyme-linked immunosorbent assay. The correlations of 25-(OH)-D levels with disease indicators and T cell subsets were analyzed.ResultThe levels of serum 25-(OH)-D in IIM were significantly lower than those in healthy controls (9.36 ± 5.56 vs 26.56 ± 5.37 ng/ml, pConclusionsVitamin D deficiency existed in IIM patients, which was significantly correlated with muscle enzyme, presence of anti-Jo-1 and anti-Mi-2 antibody, and the absolute numbers of total T and Treg cells in IIM. It is suggested that vitamin D may play an important role in the immunological pathogenesis of IIM.