Self-Supervised Joint Learning Fault Diagnosis Method Based on Three-Channel Vibration Images.

The accuracy of bearing fault diagnosis is of great significance for the reliable operation of rotating machinery. In recent years, increasing attention has been paid to intelligent fault diagnosis techniques based on deep learning. However, most of these methods are based on supervised learning with a large amount of labeled data, which is a challenge for industrial applications. To reduce the dependence on labeled data, a self-supervised joint learning (SSJL) fault diagnosis method based on three-channel vibration images is proposed. The method combines self-supervised learning with supervised learning, makes full use of unlabeled data to learn fault features, and further improves the feature recognition rate by transforming the data into three-channel vibration images. The validity of the method was verified using two typical data sets from a motor bearing. Experimental results show that this method has higher diagnostic accuracy for small quantities of labeled data and is superior to the existing methods.

IL-7 and CCL19-secreting CAR-T cell therapy for tumors with positive glypican-3 or mesothelin.

Although chimeric antigen receptor (CAR)-engineered T cells have shown great success in the treatment of B cell malignancies, this strategy has limited efficacy in patients with solid tumors. In mouse CAR-T cells, IL-7 and CCL19 expression have been demonstrated to improve T cell infiltration and CAR-T cell survival in mouse tumors. Therefore, in the current study, we engineered human CAR-T cells to secrete human IL-7 and CCL19 (7 × 19) and found that these 7 × 19 CAR-T cells showed enhanced capacities of expansion and migration in vitro. Furthermore, 7 × 19 CAR-T cells showed superior tumor suppression ability compared to conventional CAR-T cells in xenografts of hepatocellular carcinoma (HCC) cell lines, primary HCC tissue samples and pancreatic carcinoma (PC) cell lines. We then initiated a phase 1 clinical trial in advanced HCC/PC/ovarian carcinoma (OC) patients with glypican-3 (GPC3) or mesothelin (MSLN) expression. In a patient with advanced HCC, anti-GPC3-7 × 19 CAR-T treatment resulted in complete tumor disappearance 30 days post intratumor injection. In a patient with advanced PC, anti-MSLN-7 × 19 CAR-T treatment resulted in almost complete tumor disappearance 240 days post-intravenous infusion. Our results demonstrated that the incorporation of 7 × 19 into CAR-T cells significantly enhanced the antitumor activity against human solid tumor. Trial registration: NCT03198546. Registered 26 June 2017, https://clinicaltrials.gov/ct2/show/NCT03198546?term=NCT03198546&draw=2&rank=1.

Siamese neural networks for the classification of high-dimensional radiomic features.

This study demonstrates that a variant of a Siamese neural network architecture is more effective at classifying high-dimensional radiomic features (extracted from T2 MRI images) than traditional models, such as a Support Vector Machine or Discriminant Analysis. Ninety-nine female patients, between the ages of 20 and 48, were imaged with T2 MRI. Using biopsy pathology, the patients were separated into two groups: those with breast cancer (N=55) and those with GLM (N=44). Lesions were segmented by a trained radiologist and the ROIs were used for radiomic feature extraction. The radiomic features include 536 published features from Aerts et al., along with 20 features recurrent quantification analysis features. A Student T-Test was used to select features found to be statistically significant between the two patient groups. These features were then used to train a Siamese neural network. The label given to test features was the label of whichever class the test features with the highest percentile similarity within the training group. Within the two highest-dimensional feature sets, the Siamese network produced an AUC of 0.853 and 0.894, respectively. This is compared to best non-Siamese model, Discriminant Analysis, which produced an AUC of 0.823 and 0.836 for the two respective feature sets. However, when it came to the lower-dimensional recurrent features and the top-20 most significant features from Aerts et al., the Siamese network performed on-par or worse than the competing models. The proposed Siamese neural network architecture can outperform competing other models in high-dimensional, low-sample size spaces with regards to tabular data.

Radiomics Analysis of MRI for Predicting Molecular Subtypes of Breast Cancer in Young Women.

Breast cancer in young women is commonly aggressive, in part because the proportion of high-grade, triple-negative (TN) tumor is too high. There are certain limitations in the detection of biopsies or surgical specimens which only select part of tumor sample tissue and ignore the possible heterogeneity of tumors. In clinical practice, MRI is used for the diagnosis of breast cancer. MRI-based radiomics is a developing approach that may provide not only the diagnostic value for breast cancer but also the predictive or prognostic associations between the images and biological characteristics. In this work, we used radiomics methods to analyze MR images of breast cancer in 53 young women, and correlated the radiomics data with molecular subtypes. The results indicated a significant difference between TN type and non-TN type of breast cancer in young women on the radiomics features based on T2-weighted MR images. This may be helpful for the identification of TN type and guiding the therapeutic strategies.

Artesunate-Loaded and Near-Infrared Dye-Conjugated Albumin Nanoparticles as High-Efficiency Tumor-Targeted Photo-Chemo Theranostic Agent.

Herein, a tumor-targeted multifunctional theranostic agent was synthetized using a facile method, combining four clinically approved materials: artesunate (Arte), human serum albumin (HSA), folic acid (FA), and indocyanine green (ICG). The obtained nanocomposites (FA-IHA NPs) showed an excellent photo- and physiological stability. The ICG in the FA-IHA NPs was used not only for near infrared (NIR) fluorescence imaging, but also for photothermal and photodynamic (PTT-PDT) therapy under a single NIR irradiation. In addition, the NIR irradiation (808 nm, 1 W/cm2) could trigger Arte release that showed enhanced chemotherapeutic effect. Through fluorescence imaging, the cell uptake and tumor accumulation of FA-IHA NPs were observed in vitro and in vivo, analyzed by confocal microscopy and NIR fluorescence imaging in tumor xenograft mice. Based on the diagnostic results, FA-IHA NPs at 24 h post injection and combined with NIR irradiation (808 nm, 1 W/cm2) could efficiently suppress tumor growth through a photo-chemo combination therapy, with no tumor recurrence in vitro and in vivo. The obtained results suggested that FA-IHA NPs are promising photo-chemo theranostic agents for future clinical translation.

Magnetic resonance features of meningeal solitary fibrous tumors.

The aim of the present study was to investigate magnetic resonance (MR) features of meningeal solitary fibrous tumors (SFTs) in order to improve their recognition. The study retrospectively analyzed MR manifestations in 12 cases of meningeal SFTs confirmed by surgery. The lesions were analyzed in terms of the site of their growth, growth mode, morphology, size, changes in the MR signal and the edge of the focus. The tumors were malignant in 3 cases, benign in 7 cases and borderline in 2 cases. A total of 3 cases spanned the tentorium cerebelli, 3 spanned the falx and another case was close to the falx. Overall, 4 superficial tumors were accompanied by adjacent bony destruction. One case was located at the triangular area created by the right lateral ventricles. The tumors were all lobular, with a large diameter and more cystic areas in the center of the focus. The signals were often heterogeneous. The parenchyma in the tumors was significantly enhanced. For certain tumors, the signal was usually inhomogeneous and not uniform. The MR features of a SFT were often similar to other intracranial tumors and easily misdiagnosed. There were certain innate characteristics, for example, the tumor was often a solitary large soft-tissue mass with an irregular edge and a clear boundary, with a lobulated contour, and was widely involved with other tissues. The tumors often grew across either the falx or tentorium cerebelli. A significantly inhomogeneous signal was found and either a low or low-high mixed signal on T2-weighted images in particular. These features are useful for the differential diagnosis of SFTs and other tumors.

Quantitative Visualization of Dynamic Tracer Transportation in the Extracellular Space of Deep Brain Regions Using Tracer-Based Magnetic Resonance Imaging.

BACKGROUND This study assessed an innovative tracer-based magnetic resonance imaging (MRI) system to visualize the dynamic transportation of tracers in regions of deep brain extracellular space (ECS) and to measure transportation ability and ECS structure. MATERIAL AND METHODS Gadolinium-diethylene triamine pentaacetic acid (Gd-DTPA) was the chosen tracer and was injected into the caudate nucleus and thalamus. Real-time dynamic transportation of Gd-DTPA in ECS was observed and the results were verified by laser scanning confocal microscopy. Using Transwell assay across the blood-brain barrier, a modified diffusion equation was further simplified. Effective diffusion coefficient D* and tortuosity λ were calculated. Immunohistochemical staining and Western blot analysis were used to investigate the extracellular matrix contributing to ECS structure. RESULTS Tracers injected into the caudate nucleus were transported to the ipsilateral frontal and temporal cortices away from the injection points, while both of them injected into the thalamus were only distributed on site. Although the caudate nucleus was closely adjacent to the thalamus, tracer transportation between partitions was not observed. In addition, D* and the λ showed statistically significant differences between partitions. ECS was shown to be a physiologically partitioned system, and its division is characterized by the unique distribution territory and transportation ability of substances located in it. Versican and Tenascin R are possible contributors to the tortuosity of ECS. CONCLUSIONS Tracer-based MRI will improve our understanding of the brain microenvironment, improve the techniques for local delivery of drugs, and highlight brain tissue engineering fields in the future.

Investigation of an optimized scanning protocol for the dentomaxillofacial region using 320-slice multidetector computed tomography.

OBJECTIVES: To propose an imaging protocol that provides satisfactory image quality for oral examination while minimizing radiation dosage using 320-slice multidetector CT (MDCT). METHODS: An anthropomorphic head phantom was scanned using 320 MDCT with protocols combining different scanning modes: volume scanning (whole or local) vs helical scanning (80- or 64-slice detectors); tube voltage settings (80 kVp, 120 kVp and 135 kVp); and tube current settings (60 mA, 80 mA, 100 mA and 120 mA). A total of six anatomical bone structures and three anatomical soft-tissue structures were assessed using quantitative and qualitative analysis in the three orthographic planes (axial, sagittal and coronal). A figure of merit (FOM) was used to determine the optimal imaging protocol in terms of tube voltage, tube current and scanning mode. RESULTS: The 80-kVp setting had the worst quantitative and qualitative results (both p < 0.001) compared with the 135-kVp and 120-kVp settings, especially for soft-tissue structures. A significant difference was noted for the scores obtained using a tube current between 120 mA and 60 mA by quantitative analysis, but not by qualitative analysis. Volume scans using either whole or local modes had a significantly higher FOM than helical scanning of 80 or 64 slices. CONCLUSIONS: In 320 MDCT, a protocol using 135 kVp, 80 mA and the volume-scanning mode (whole or local) offers adequate visualization of both soft-tissue and bone structures while keeping the radiation dose as low as possible. This may therefore be considered a first choice among a wide selection of scanning protocols for dentomaxillofacial CT.

A multicentre retrospective study of transcatheter angiographic embolization in the treatment of delayed haemorrhage after percutaneous nephrolithotomy.

OBJECTIVE: The objective is to determine the timing and indications of transcatheter angiographic embolization (TAE) for delayed haemorrhage after percutaneous nephrolithotomy (PCNL). METHODS: The medical records of 144 patients who underwent arteriography and TAE for delayed post-PCNL haemorrhage at five university hospitals between January 2005 and December 2012 were reviewed retrospectively. RESULTS: The mean time to the onset of post-PCNL haemorrhage was 10.5 days (2 - 30 days). Clinical presentation included sudden onset bleeding in 51 patients (35.4 %), intermittent bleeding in 67 patients (46. 5 %), and continuous slow bleeding in 26 patients (18.1 %). Hemodynamic instability occurred in 32 patients (22.2 %). The mean haemoglobin decrease from the first post-PCNL day to the day of TAE was 49.5 g/L (31.0 - 79.0 g/L). Renal arteriography showed pseudoaneurysms in 69 (47.9 %) patients, arteriovenous fistulas in 28 (19.4 %) patients, mixed arterial and arteriovenous lesions in 17 (11.8 %) patients, arterial lacerations in 23 (16.0 %) patients, and negative angiographic finding in seven (4.9 %) patients. TAE was successful in stopping bleeding in all 137 patients with vascular lesions. There were no major complications associated with TAE. CONCLUSIONS: TAE should be the recommended treatment for delayed post-PCNL haemorrhage in patients with hemodynamic instability and/or corrected haemoglobin decrease >30 g/L following conservative management. KEY POINTS: • Delayed haemorrhage after percutaneous nephrolithotomy occurs more than 24 hours postoperatively. • Angio-embolization is a safe and effective treatment for delayed post-PCNL haemorrhage. • Angio-embolization can treat hemodynamic instability and/or corrected haemoglobin decrease >30 g/L.

[Effects of miR-106b expression on the proliferation of human hepatocellular carcinoma cells].

OBJECTIVE: To investigate the effects of miR-106b expression on the proliferation of human hepatocellular carcinoma cells. METHODS: Real-time PCR was used to detect the expression of miR-106b in human hepatocellular carcinoma (HCC) cell lines and normal liver epithelial THLE3 cells. Over-expression of miR-106b was transfected by miR-106b mimics, and inhibition of miR-106b expression was transfected by miR-106b inhibitors. The effects of miR-106b expression on the proliferation of HCC cells were detected by MTT, clone formation assay and anchorage-independent growth ability assay. Bromodeoxyuridine labeling and flow cytometry analysis were used to examine the effects of miR-106b expression on the cell cycle distribution of the HCC cells. RESULTS: Compared with that in the normal liver epithelial THLE3 cells, the relative expression of miR-106b in HepG2, QGY-7703, BEL-7402, MHCC97H, HCCC-9810, Hep3B, MHCC97L and Huh7 cell lines were 5.37 ± 0.35, 8.45 ± 0.75, 19.22 ± 1.74, 11.93 ± 1.26, 17.03 ± 0.97, 4.19 ± 0.67, 7.94 ± 1.35 and 3.82 ± 0.87, respectively (P < 0.05 for all). Three days after transfection, the miR-106b over-expression was accelerated, while miR-106b inhibitor suppressed the proliferation of HCC cells. The numbers of clones formed were (4.13 ± 0.75) and (3.78 ± 0.47) times higher than that of control cells, and (147.73 ± 15.56) and (138.87 ± 15.58) clones in diameter >1.0 mm were formed by miR-106b-overexpressing cells. When the miR-106b expression was inhibited in the HepG2 and QGY-7703 cells, the clone numbers were (0.18 ± 0.05) and (0.24 ± 0.07) times of that in the controls, and the numbers of clones formed were (23.29 ± 7.14) and (20.60 ± 8.07) (P < 0.05 for all). The positive rates of BrdU labeled miR-106b-overexpressing HepG2 and QGY-7703 cells were (51.89 ± 4.91) % and (54.74 ± 6.10) %, those of the miR-106b-inhibited cells were (6.48 ± 3.15) % and (7.52 ± 2.03) %, significantly different from that in the control cells (P < 0.05 for all). The S phases were dramatically increased from 29.93% and 31.04% to 56.76% and 57.22% in the miR-106b-overexpressing HepG2 and QGY-7703 cells, while they were 19.43% and 19.92% in the miR-106b-inhibited HepG2 and QGY-7703 cells. CONCLUSIONS: MiR-106b overexpression may promote the proliferation and metastasis of HCC cells. The mechanism of this effect may be related to promoting cells into S phase and inhibiting cell apoptosis.

miR-106b downregulates adenomatous polyposis coli and promotes cell proliferation in human hepatocellular carcinoma.

Aberrant activation of the Wnt/ß-catenin signal pathway is frequently observed in hepatocellular carcinoma (HCC). ß-Catenin is the major cellular effector of Wnt signaling and inactivation of adenomatous polyposis coli (APC) results in nuclear accumulation of ß-catenin. Therefore, it was speculated that APC inhibition could play important roles in activating the Wnt/ß-catenin pathway and in HCC progression. In this study, we report that miR-106b expression is markedly upregulated in hepatoma cells and hepatoma tissues compared with immortalized normal liver epithelial cells and normal hepatic tissues. Ectopic expression of miR-106b induces the proliferation and anchorage-independent growth of hepatoma cells, whereas inhibition of miR-106b reduced this effect. Furthermore, miR-106b upregulation in hepatoma cells modulated entry into the G(1)/S transitional phase by upregulating cyclin D1 and downregulating APC. Moreover, we demonstrated that miR-106b downregulates APC expression by directly targeting the 3'-untranslated region of APC messenger RNA. Taken together, our results suggest that miR-106b plays an important role in promoting the proliferation of human hepatoma cells and presents a novel mechanism of micro RNA-mediated direct suppression of APC expression in cancer cells.

[Relationship between hMSH2 and FHIT gene expression in non-small cell lung cancer].

BACKGROUND & OBJECTIVE: Abnormality of FHIT gene has been proved to be frequent in certain malignant tumors closely related to environmental oncogenic factors, such as lung cancer. Foreign scholars have begun to explore the relationship between FHIT gene and other tumor suppressor genes, which are implicated in the pathogenesis of lung cancer. This study was designed to investigate the relationship between hMSH(2) and FHIT protein expression and to explore the correlation of hMSH(2) and FHIT protein expression with clinicopathologic features of lung cancer. METHODS: Immunohistochemical analysis of hMSH(2) and FHIT protein expression in 40 lung cancer cases and 15 adjacent non-cancer lung tissues was performed; the positive rates of FHIT and hMSH(2) proteins were measured by image analysis system. RESULTS: (1)The positive rates of FHIT and hMSH(2) proteins were 58.2% and 45.8% respectively in lung cancer tissues compared with 89.1% and 65.3% in non-cancer lung tissues. The expression levels of FHIT and hMSH(2) proteins were significantly lower in lung cancer tissues than that in non-cancer lung tissues (P< 0.01). (2)Reduced expression levels of both proteins were significantly related to tumor histology. The positive rate of the FHIT protein was 52.2% in squamous cell carcinoma compared with 63.4% in adenocarcinomas(P< 0.01), whereas the positive rate of the hMSH(2) protein was 35.6% in adenocarcinomas compared with 53.2% in squamous cell carcinoma(P< 0.01). (3)A correlation between FHIT reduced expression and lymph node metastasis was observed(P< 0.01). The positive rate of the FHIT protein was 54.1% in lung cancer tissues with metastasis compared with 60.5% in lung cancer tissues without metastasis. No association was found between hMSH(2) reduced expression and nodal metastasis(P >0.05). (4)Loss of FHIT protein correlated significantly with lasting and heavy smoking(P< 0.01). The positive rate of the FHIT protein was 53.1% in smoking group compared with 66.1% in non-smoking group. The reduction of hMSH(2) expression was not associated with smoking(P >0.05). (5)An inverse correlation was found between hMSH(2) reduced expression and FHIT protein loss (P< 0.01, RR=-0.54). CONCLUSION: FHIT gene may be a negative regulatory gene of hMSH(2) gene, and play an important role in the inactivation mechanism of hMSH(2) gene.

[Helical dual-phase CT scan in evaluating blood supply of primary heptocellular carcinoma after transcatheter hepatic artery chemoembolization with lipiodol].

OBJECTIVE: To evaluate the blood supply of low density viable area of primary heptocellular carcinoma after transcatheter hepatic artery chemoembolization using lipiodol (LP-TACE), by helical dual-phase CT scanning and three dimensional CT (3DCT). METHODS: Thirty-four patients with primary heptocellular carcinoma after LP-TACE were examined by hepatic helical dual-phase CT. 3DCT model of the maximum intensity projection (MIP), surface shaded display (SSD) reconstruction of the hepatic artery and portal vein were simultaneously done in 5 cases. RESULTS: Viable tumor areas of 34 cases of primary heptocellular carcinoma after LP-TACE were divided into four types: peripheral, lateral, central and diffused types. Enhanced tumor vessel or tissue in viable tumor area was found during hepatic dual-phase in 17 cases, during hepatic artery-phase only in 8 and hepatic portal vein-phase only in 3. The viable tumor areas were found to have blood supply from the hepatic vein in 2 cases. The viable tumor area unenhanced during hepatic dual-phase was found in 6 cases. In 5 cases, the relation between the viable tumor area and branches of hepatic artery and portal vein was showed by MIP and SSD of hepatic artery and portal vein. CONCLUSION: Hepatic helical dual-phase CT scan with 3DCT is effective in evaluating the blood supply of viable tumor areas and the therapeutic effect of primary heptocellular carcinoma after LP-TACE.

[Expression of HIF-1 alpha and its relationship to apoptosis and proliferation in lung cancer].

BACKGROUND & OBJECTIVES: Hipoxia-inducible factor-1 is a transcriptive factor that regulates genes involved in metabolism, angiogenesis, proliferation, and apoptosis. This study was designed to investigate the expression of hypoxia inducible facter-1 alpha(HIF-1 alpha) and its relationship to bcl-2, Bax, PCNA in lung cancer. METHOD: Immunohistochemical streptavidin/peroxidase(SP) was used to examine the expression of HIF-1a, bcl-2, Bax, and PCNA in 60 cases of lung cancer. RESULTS: In 60 cases of lung cancer, positive rate for HIF-1a was 28.3% (17/60), specially the positive rate of small cell lung cancer(66.7%) was significantly higher than non-small cell lung cancer (21.6%). HIF-1a expression increased as clinical stage and metastasis increased(P < 0.01). The positive rate of bcl-2, Bax, and PCNA were 31.7% (19/60), 40.0% (24/60), 76.7% (46/60), respectively. Inverse relationship was found between the expression of HIF-1 alpha and bcl-2; while the correlation of HIF-1 alpha and Bax was positive(P < 0.01). The relationship between HIF-1 alpha and Bax was positive(P < 0.01). The relationship between HIF-1 alpha and PCNA was not observed(P > 0.05). CONCLUSION: HIF-1 alpha is correlated with apopotosis, but has no relationship with proliferation.

[Dynamic study on expression of gelatinase A and its natural inhibitor during invasion and metastasis of induced lung cancer in Wistar rats].

OBJECTIVE: To investigate the dynamic expression and its relation of gelatinase A (MMP-2), its natural inhibitor (TIMP-2) and DNA index (DI) changes during carcinogenesis, invasion and metastasis in Wistar rats. METHODS: Squamous cell carcinoma of lung was induced with 3-methylcholanthrene (MCA) and diethyinitrosamine (DEN) in iodized oil by left intra-bronchial instillation in 80 Wistar rats. Immrno histochemistay (IHC) and in situ hybridigation were used in the monitor of MMP-2, TIMP-2 proteins and mRNA expression during invasion and metastasis of lung cancer in these rats, DNA index (DI) value was measured by guantitatove image analysis on feulgen stained sections. RESULTS: Along with the carcinogenis, the average poritive MNP-2 and TIMP-2 expressions increased, with positive rates of 8.5% - 85.7% and 6.4% - 35.7%. DI value also underwent the same changes (1.47 +/- 0.54) - (2.87 +/- 0.55). The difference of MMP-2 expression in carcinoma in situ versus early carcinoma and early carcinoma versus metastatic carcinoma are statistically significant (P < 0.05). Companing lung carcinome, the contrel group and non-cancerous lesions, the elevation of MNP-2 and TIMP-2 expressions were also sigmificant (P < 0.01). The DI elevation in carcinoma in situ and dysplasia were obviously significant (P < 0.05). Meanwhile a negative relation was noted in TINP-2 and MMP-2 expressions during carcinogenesis. There was a positive relation between MMP-2 expression and DNA poikiloidy (P < 0.01), which was related to the close relationship between MMP-2 and metastasis in advanced rat lung carcinoma (P < 0.05). CONCLUSION: The excess degradation and disruption of basement membranes by activated MMP-2 may be a key step in inducing lung cancer invasion and metastasis. The imbalance between MMP and TIMP may be a critical factor which affects biologic behavior of lung carcinogenesis, invasion and metastasis.