Vocal Cord Leukoplakia Classification Using Siamese Network Under Small Samples of White Light Endoscopy Images.

OBJECTIVE: Accurate vocal cord leukoplakia classification is instructive for clinical diagnosis and surgical treatment. This article introduces a reliable very deep Siamese network for accurate vocal cord leukoplakia classification. STUDY DESIGN: A study of a classification network based on a retrospective database. SETTING: Academic university and hospital. METHODS: The white light image datasets of vocal cord leukoplakia used in this article were classified into 6 classes: normal tissues, inflammatory keratosis, mild dysplasia, moderate dysplasia, severe dysplasia, and squamous cell carcinoma. The classification performance was assessed by comparing it with 6 classical deep learning models, including AlexNet, VGG Net, Google Inception, ResNet, DenseNet, and Vision Transformer. RESULTS: Experiments show the superior classification performance of our proposed network compared to state-of-the-art methods. The overall accuracy is 0.9756. The values of sensitivity and specificity are very high as well. The confusion matrix provides information for the 6-class classification task and demonstrates the superiority of our proposed network. CONCLUSION: Our very deep Siamese network can provide accurate classification results of vocal cord leukoplakia, which facilitates early detection, clinical diagnosis, and surgical treatment. The excellent performance obtained in white light images can reduce the cost for patients, especially those living in developing countries.

Vocal cord leukoplakia classification using deep learning models in white light and narrow band imaging endoscopy images.

BACKGROUND: Accurate vocal cord leukoplakia classification is critical for the individualized treatment and early detection of laryngeal cancer. Numerous deep learning techniques have been proposed, but it is unclear how to select one to apply in the laryngeal tasks. This article introduces and reliably evaluates existing deep learning models for vocal cord leukoplakia classification. METHODS: We created white light and narrow band imaging (NBI) image datasets of vocal cord leukoplakia which were classified into six classes: normal tissues (NT), inflammatory keratosis (IK), mild dysplasia (MiD), moderate dysplasia (MoD), severe dysplasia (SD), and squamous cell carcinoma (SCC). Vocal cord leukoplakia classification was performed using six classical deep learning models, AlexNet, VGG, Google Inception, ResNet, DenseNet, and Vision Transformer. RESULTS: GoogLeNet (i.e., Google Inception V1), DenseNet-121, and ResNet-152 perform excellent classification. The highest overall accuracy of white light image classification is 0.9583, while the highest overall accuracy of NBI image classification is 0.9478. These three neural networks all provide very high sensitivity, specificity, and precision values. CONCLUSION: GoogLeNet, ResNet, and DenseNet can provide accurate pathological classification of vocal cord leukoplakia. It facilitates early diagnosis, providing judgment on conservative treatment or surgical treatment of different degrees, and reducing the burden on endoscopists.

Salvianolic acid B exerts an anti-hepatocellular carcinoma effect by regulating the Hippo/YAP pathway and promoting pSmad3L to pSmad3C simultaneously.

Salvianolic acid B (Sal B) is a component obtained from Salvia miltiorrhiza and is empirically used for liver diseases. The TGF-ß/Smad and Hippo/YAP pathways may interact with each other in hepatocellular carcinoma (HCC). Previously, we found that Sal B mediates the TGF-ß/Smad pathway in mice and delays liver fibrosis-carcinoma progression by promoting the conversion of pSmad3L to pSmad3C, but the effect of Sal B on the Hippo/YAP pathway has not been determined. Therefore, we used a DEN/CCl4/C2H5OH-induced liver cancer model in mice to analyze liver index and tumor incidence, detect AST and ALT serological markers, observe liver pathology and the number of Ki67-positive cells to evaluate the anti-HCC effect of Sal B in vivo. We used a TGF-ß1-induced HepG2 cell model, and applied an MST1/2 inhibitor, XMU-MP-1, to detect the changes in pSmad3C/pSmad3L signaling induced by MST1/2 inhibition. Sal B significantly inhibited tumorigenesis in DEN/CCl4/C2H5OH-induced mice in vivo, and suppressed the growth of HepG2 cells by inhibiting cell proliferation and migration in vitro. Here, our study also validated the role of Sal B in reversing XMU-MP-1-induced proliferation and migration of HepG2 cells in vitro. Most importantly, we elucidated for the first time the potential mechanism of Sal B against HCC via the Hippo/YAP pathway, which may be specifically related to upregulation of MST1 and inhibition of its downstream effector protein YAP. In conclusion, these findings indicate that Sal B possesses anti- HCC effects both in vivo and in vitro by regulating the Hippo/YAP pathway and promoting pSmad3L to pSmad3C synchronously.

Astragaloside IV inhibits hepatocellular carcinoma by continually suppressing the development of fibrosis and regulating pSmad3C/3L and Nrf2/HO-1 pathways.

ETHNOPHARMACOLOGICAL RELEVANCE: Astragalus is a medicinal herb used in China for the prevention and treatment of diseases such as diabetes and cancer. As one of the main active ingredients of astragalus, Astragaloside IV (AS-IV) has a wide range of pharmacological effects, including anti-inflammation and anti-cancer effects. AIM OF THE STUDY: Different phosphorylated forms of Smad3 differentially regulate the progression of hepatic carcinoma. The phosphorylation of the COOH-terminal of Smad3 (pSmad3C) and activation of the Nrf2/HO-1 pathway inhibits hepatic carcinoma, while phosphorylation of the linker region of Smad3 (pSmad3L) promotes progression. Thus, pSmad3C/3L and Nrf2/HO-1 pathways are potential targets for drug of anti-cancer development. AS-IV is anti-apoptotic and can inhibit hepatocellular carcinoma cell (HCC) proliferation, invasion, and tumor growth in nude mice. However, it is not clear whether AS-IV has a therapeutic effect on inhibiting the progression of primary liver cancer by regulating the pSmad3C/3L and Nrf2/HO-1 pathway. The purpose of this study is to investigate whether AS-IV inhibits hepatocellular carcinoma by regulating pSmad3C/3L and Nrf2/HO-1 pathway. MATERIALS AND METHODS: primary liver cancer in mice induced by DEN/CCl4/C2H5OH (DCC) and HSC-T6/HepG2 cell models activated by TGF-ß1 was investigated for the mechanisms of AS-IV. In vivo assays included liver biopsy, histopathology and post-mortem analysis included immunohistochemistry, immunofluorescent, and Western blotting analysis, and in vitro assays included immunofluorescent, and Western blotting analysis. RESULTS: AS-IV significantly inhibited the development of primary liver cancer, reflecting improved liver biopsy, histopathology. The incidence and multiplicity of primary liver cancer were markedly decreased by AS-IV treatment at the 20th week. AS-IV had observable effects on the TGF-ß1/Smad and Nrf2/HO-1 expression in vivo, especially up-regulated pSmad3C, pNrf2, HO-1, and NQO1, while it down-regulated pSmad2C, pSmad2L, pSmad3L, PAI-1, and α-SMA at the 12th week and the 20th week. Furthermore, in vitro analysis further confirmed that AS-IV regulated the expression of pSmad3C/3L and Nrf2/HO-1 pathway in HSC-T6 and HepG2 cells activated by TGF-ß1. CONCLUSION: AS-IV administration delays the occurrence of primary liver cancer by continually suppressing the development of fibrosis, the mechanism of the therapeutic effect involving the regulation of the pSmad3C/3L and Nrf2/HO-1 pathways, especially in regulation reversibility and antagonism of pSmad3C and pSmad3L and promoting the phosphorylation of Nrf2.

Smad3 gene C-terminal phosphorylation site mutation exacerbates CCl4-induced hepatic fibrogenesis by promoting pSmad2L/C-mediated signaling transduction.

Current researches have confirmed that Smads, mediators of TGF-ß signaling, are strictly controlled by domain-specific site phosphorylation in the process of hepatic disease. Usually, Smad3 phospho-isoform pSmad3L and pSmad3C are reversible and antagonistic; pSmad2L/C could act together with pSmad3L by stimulating PAI-1 expression and ECM synthesis to transmit fibrogenic signals. Our recent study found that pSmad3C mutation is supposed to perform a vigorous role on the early phase of liver injury and abates salvianolic acid B's anti-hepatic fibrotic-carcinogenesis. However, whether pSmad3C mutation expedites pSmad2L/C-mediated signaling transduction during hepatic fibrogenesis remains vague. Presently, Smad3 gene C-terminal phosphorylation site mutation heterozygote (pSmad3C+/-) mice were constructed to probe if and how pSmad3C retards CCl4-induced hepatic fibrogenesis by inhibiting pSmad2L/C-mediated signaling transduction. Twelve 6-week-old pSmad3C+/- C57BL/6J mice were intraperitoneally injection with CCl4 for 6 weeks to induce liver fibrogenesis. Results showed that pSmad3C mutation aggravates the relative liver weight, biochemical parameters, collagenous fibers and fibrotic septa formation, contributes to fibrogenesis in HT-CCl4 mice. Furthermore, fibrotic-related proteins TGF-ß1, pSmad2C, pSmad2L, and PAI-1 were also increased in CCl4-induced pSmad3C+/- mice. These results suggest that pSmad3C mutation exacerbates hepatic fibrogenesis which relates to intensifying pSmad2L/C-mediated signaling transduction.

A joint ROI extraction filter for computer aided lung nodule detection.

Extraction of regions of interest plays an important rule in computer aided lung nodules detection. However, because of the complex background and structure, accurate and robust extraction of ROIs in medical image still remains a problem. Aim at this problem, a two-stage operations joint filter: Hessian-LoB, is proposed. The first stage is blobs (which being taken as candidate ROIs) detection and the second stage is ROIs extraction. In the first stage, the derivatives of a Hessian matrix at multiple scales are convolved with input images to localize blobs. Then in the second stage, Laplacian of bilateral filter (LoB) is convolved with the detected blobs to extract the final ROIs. Experiments show that the proposed filter can deal with images with noise and low brightness contrast, and is effectively in ROI extraction for lung nodule detection.

A new method of detecting pulmonary nodules with PET/CT based on an improved watershed algorithm.

BACKGROUND: Integrated 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) is widely performed for staging solitary pulmonary nodules (SPNs). However, the diagnostic efficacy of SPNs based on PET/CT is not optimal. Here, we propose a method of detection based on PET/CT that can differentiate malignant and benign SPNs with few false-positives. METHOD: Our proposed method combines the features of positron-emission tomography (PET) and computed tomography (CT). A dynamic threshold segmentation method was used to identify lung parenchyma in CT images and suspicious areas in PET images. Then, an improved watershed method was used to mark suspicious areas on the CT image. Next, the support vector machine (SVM) method was used to classify SPNs based on textural features of CT images and metabolic features of PET images to validate the proposed method. RESULTS: Our proposed method was more efficient than traditional methods and methods based on the CT or PET features alone (sensitivity 95.6%; average of 2.9 false positives per scan).

A new method based on MTANNs for cutting down false-positives: an evaluation on different versions of commercial pulmonary nodule detection CAD software.

One of the major problems for computer-aided pulmonary nodule detection in chest radiographs is that a high false-positive (FP) rate exists. In an effort to overcome this problem, a new method based on the MTANN (Massive Training Artificial Neural Network) is proposed in this paper. An MTANN comprises a multi-layer neural network where a linear function rather than a sigmoid function is used as its activity function in the output layer. In this work, a mixture of multiple MTANNs were employed rather than only a single MTANN. 50 MTANNs for 50 different types of FPs were prepared firstly. Then, several effective MTANNs that had higher performances were selected to construct the MTANNs mixture. Finally, the outputs of the multiple MTANNs were combined with a mixing neural network to reduce various different types of FPs. The performance of this MTANNs mixture in FPs reduction is validated on three different versions of commercial CAD software with a validation database consisting of 52 chest radiographs. Experimental results demonstrate that the proposed MTANN approach is useful in cutting down FPs in different CAD software for detecting pulmonary nodules in chest radiographs.