Purpose: The purpose of this study was to establish a deep learning model for automated sub-basal corneal nerve fiber (CNF) segmentation and evaluation with in vivo confocal microscopy (IVCM). Methods: A corneal nerve segmentation network (CNS-Net) was established with convolutional neural networks based on a deep learning algorithm for sub-basal corneal nerve segmentation and evaluation. CNS-Net was trained with 552 and tested on 139 labeled IVCM images as supervision information collected from July 2017 to December 2018 in Peking University Third Hospital. These images were labeled by three senior ophthalmologists with ImageJ software and then considered ground truth. The areas under the receiver operating characteristic curves (AUCs), mean average precision (mAP), sensitivity, and specificity were applied to evaluate the efficiency of corneal nerve segmentation. The relative deviation ratio (RDR) was leveraged to evaluate the accuracy of the corneal nerve fiber length (CNFL) evaluation task. Results: The model achieved an AUC of 0.96 (95% confidence interval [CI] = 0.935–0.983) and an mAP of 94% with minimum dice coefficient loss at 0.12. For our dataset, the sensitivity was 96% and specificity was 75% in the CNF segmentation task, and an RDR of 16% was reported in the CNFL evaluation task. Moreover, the model was able to segment and evaluate as many as 32 images per second, much faster than skilled ophthalmologists. Conclusions: We established a deep learning model, CNS-Net, which demonstrated a high accuracy and fast speed in sub-basal corneal nerve segmentation with IVCM. The results highlight the potential of the system in assisting clinical practice for corneal nerves segmentation and evaluation. Translational Relevance: The deep learning model for IVCM images may enable rapid segmentation and evaluation of the corneal nerve and may provide the basis for the diagnosis and treatment of ocular surface diseases associated with corneal nerves.
Poly(acrylic acid-co-2-hydroxyethyl acrylate)/Fe_3O_4 magnetic fluid was prepared by modifying nanometer Fe_3O_4 particles with poly(acrylic acidco-2-hydroxyethyl acrylate) [P(AA-co-HEA)].The copolymer was characterized by gel permeation chromatograph(GPC),nuclear magnetic resonance spectrometry(NMR),and the P(AA-co-HEA)/Fe_3O_4 composite was characterized by transmission electron microscope(TEM),fourier transform infrared spectroscope(FT-IR),ultravioletvisible(UV-Vis)(absorption) spectroscopy and super conducting quantum interference device(SQUID) magnetometer.The(results) show that the modified can be dispersed stably in ethanol for long period and have rapid magnetic(response) against external field.P(AA-co-HEA) adsorbed onto the surface of nano-Fe_3O_4 by coordinate bond.The diameter of the modified magnetic particles is about 10 nm and the material exhibit super paramagnetic properties above the blocking temperature.
Objective Carotid atherosclerosis is a chronic progressive vascular disease that can be complicated by stroke in severe cases. Prompt diagnosis and treatment of high‐risk patients are quite difficult due to the lack of reliable clinical biomarkers. This study aimed to explore potential plaque metabolic markers of stroke‐prone risk and relevant targets for pharmacological intervention. Method Carotid intima and plaque sample tissues were obtained from 20 patients with cerebrovascular symptoms of carotid origin. An untargeted metabolomics approach based on liquid chromatography–tandem mass spectrometry was utilized to characterize the metabolic profiles of the tissues. Multivariate and univariate analysis tools were used. Results A total of 154 metabolites were significantly altered in carotid plaque when compared with thickened intima. Of these, 62 metabolites were upregulated, whereas 92 metabolites were downregulated. Support vector machines identified the 15 most important metabolites, such as N ‐(cyclopropylmethyl)‐ N ′‐phenylurea, 9( S )‐HOTrE, ACar 12:2, quinoxaline‐2,3‐dithiol, and l ‐thyroxine, as biomarkers for high‐risk plaques. Metabolic pathway analysis showed that abnormal purine and nucleotide metabolism, amino acid metabolism, glutathione metabolism, and vitamin metabolism may contribute to the occurrence and progression of carotid atherosclerotic plaque. Conclusions Our study identifies the biomarkers and related metabolic mechanisms of carotid plaque, which is stroke‐prone, and provides insights and ideas for the precise prevention and targeted intervention of the disease.
We propose and demonstrate a cladding-pumped, erbium-ytterbium co-doped fiber amplifier (EYDFA) scheme based on dual wavelength auxiliary signal injection technique to solve the issue of the backward Yb-ASE self-lasing under strong pumping for the high-power fiber amplification of 1.5-μm kHz-linewidth linearly-polarized laser signal. With the dual wavelength auxiliary signal of 1030 and 1040 nm injection, the allowable maximum pump power without triggering the backward Yb-ASE self-lasing can be greatly increased due to the relieving effect on the inhomogeneous gain broadening. For an EYDFA with 3.8-m erbium-ytterbium co-doped double-clad fiber, the net output power is improved to 13.8 W, while the linewidth of the amplified single-transverse-mode linearly-polarized 1560 nm laser signal is still only 3.5 kHz. The SBS effect is observed to be trivial during the fiber amplification.
HFPA6/mMCA and HFPA6/MCA composites were prepared by means of molten blending the high flow-ability polyamide 6 (HFPA6) with the graphene oxide modified melamine cyanuric (mMCA) and the unmodified melamine cyanuric (MCA) respectively. The structure of MCA and mMCA, as well as the flame retardant properties, thermal stability and mechanical properties of the two HFPA 6 composites were characterized. Results show that the flame retardancy, tensile strength, flexural strength and charring of the HFPA6/mMCA composite are superior to that of the HFPA 6/MCA composite, while the impact strength of HFPA6/mMCA composite is slightly lower. When the content of flame retardants mMCA is 14%, the flame retardancy of HFPA6/mMCA composite reaches the UL 94 V-0 rating.
Compressed Sensing is for sparse and compressible signals, the data is compressed while the signal is sampled. This paper proposes the new deterministic measurement matrices that are studied: according to the compressible signal characteristics, we will use the unit matrix added with random orthogonal matrix and complementary sequences as the measurement matrix, and then using orthogonal matching pursuit (OMP) algorithm to reconstruct the signal, we can safely draw that as deterministic measurement matrix, they are feasible to reconstruct the original signal accurately. The simulation results show that the performances of the unit matrix added with random orthogonal matrix and complementary sequences are not only superior the partial Hadamard matrix, but also better than the Gaussian random measurement matrix.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)