Nowadays, deep learning techniques are extensively applied in the field of automatic target recognition (ATR) for radar images. However, existing data-driven approaches frequently ignore prior knowledge of the target, leading to a lack of interpretability and poor performance of trained models. To address this issue, we first integrate the knowledge of structural attributes into the training process of an ATR model, providing both category and structural information at the dataset level. Specifically, we propose a Structural Attribute Injection (SAI) module that can be flexibly inserted into any framework constructed based on neural networks for radar image recognition. Our proposed method can encode the structural attributes to provide structural information and category correlation of the target and can further apply the proposed SAI module to map the structural attributes to something high-dimensional and align them with samples, effectively assisting in target recognition. It should be noted that our proposed SAI module can be regarded as a prior feature enhancement method, which means that it can be inserted into all downstream target recognition methods on the same dataset with only a single training session. We evaluated the proposed method using two types of radar image datasets under the conditions of few and sufficient samples. The experimental results demonstrate that our application of our proposed SAI module can significantly improve the recognition accuracy of the baseline models, which is equivalent to the existing state-of-the-art (SOTA) ATR approaches and outperforms the SOTA approaches in terms of resource consumption. Specifically, with the SAI module, our approach can achieve substantial accuracy improvements of 3.48%, 18.22%, 1.52%, and 15.03% over traditional networks in four scenarios while requiring 1/5 of the parameter count and just 1/14 of the FLOPs on average.
Degree distribution lies at the heart of design an LT code. It affects the encoding and decoding complexity and error performance of LT code. In this paper, some metrics of well-design LT code such as average degree, release probability, overhead factor is analyzed. It will compare the metrics among robust LT code, sub optimal LT code and SF-LT code. Moreover, it provides some mathematical guideline on how to design a well designed LT code.
To evaluate the expressions of P504s, 34β-E12, and P63 proteins in PCa and the correlation between the parameters of multi-modality MRI and the expression of P504s.We retrospectively analyzed the multi-modality MRI data on 43 PCa and 64 non-PCa patients. We obtained the signal intensity-time (SI-T) curves, maximum SI (SImax), time to SImax (Tmax), rate of maximum enhancement (Rmax), and automatically generated apparent diffusion coefficient (ADC) values by conventional, diffusion-weighted and dynamic contrast-enhanced MRI, and determined the expressions of P504s, 34β-E12 and P63 in the prostatic tissues of the patients by immunohistochemistry.Statistically significant differences were observed between the PCa and non-PCa groups in the positive expressions of P504s (83.7% vs 0%, P < 0.05), 34β-E12 (25.6% vs 91.0%, P < 0.05) and P63 (25.6% vs 86.0%, P < 0.05) in the prostatic tissue, the ADC value (［0.83 ± 0.22］ vs ［1.34 ± 0.28］ ×10－3mm2/s, P < 0.05), Tmax (［21.30 ± 10.78］ vs ［50.22 ± 36.31］ s, P < 0.05), SImax (［1.75 ± 0.39］% vs ［1.24 ± 0.41］%, P < 0.05), and Rmax (［20.20 ± 15.50］% vs ［7.98 ± 6.25］%, P < 0.05). The expression of P504s was correlated negatively with the ADC value and Tmax (r = －0.60 and －0.37, P < 0.01) but positively with SImax and Rmax (r = 0.50 and 0.45, P < 0.01). The parameters of multi-modality MRI are correlated with the expression of P504s in PCa and can be used as imaging biomarkers for predicting the degrees of its malignancy.
Objective
To investigate the effect of 810 nm low-level laser on neuronal axonal regeneration of mice with spinal cord injury and its related mechanism.
Methods
In vivo experiment: 20 Balb/c mice were randomly divided into the spinal cord injury group (SCI group) and the 810 nm low-level laser irradiation group (low-level laser group) after spinal cord injury according to the random number table method, with each group containing ten mice. A mice SCI model was established through clamp injury and the low-level laser group continuously irradiated the damaged area with weak 810 nm low-level laser with selected parameters (continuous wave with wave length 810 nm, power density 2 mW/cm2, spot are 4.5 cm2, irradiation time 50 minutes, energy 6 000 J/cm2). Then immunofluorescence staining was used to observe the M1 macrophage marker-inducible nitric oxide synthase (iNOS), the M2 macrophage marker arginase 1 (Arg-1) and the universal marker F4/80 of macrophages after 14 days. Furthermore, in the in vitro experiment, standardized low-level laser-macrophage irradiation model was established. Another 20 Balb/c mice were used to obtain primary bone marrow-derived macrophages which were induced into M1 macrophages using lipopolysaccharide (LPS) and interferon-gamma (INF-γ). The M1 macrophages were randomly divided into the M1 macrophage group (M1 group) and the low-level laser therapy group (M1+ low-level laser group) equally according to the random number table method. The M1 group was not treated, and the M1+ low-level laser group was treated with low-level laser of selected parameters. RT-qPCR and ELISA were used to detect the expression of interleukin-1 receptor antagonist (IL-1RA) and interleukin-10 (IL-10) in M1 macrophages 24 hours after irradiation. Western blot was used to analyze the expression of iNOS, Arg-1, differentiation antigen cluster 206 (CD206), protein kinase B (AKT), phosphorylated protein kinase B (p-AKT), cyclic adenosine response element binding protein (CREB) and phosphorylated cyclic adenosine response element binding protein (p-CREB) in M1 macrophages 48 hours after irradiation. Dorsal root ganglion neurons (DRG) were cultured in two groups of macrophage conditioned medium, and the length of DRG axon growth was measured 48 h later to evaluate the effect of low-level laser on neuronal axon growth.
Results
In the in vivo experiment, compared with mice with spinal cord injury alone, the fluorescence intensity of F4/80+ iNOS+ in the spinal cord injury area decreased (1.00±0.08 vs. 0.06±0.04)(P<0.05) and the fluorescence intensity of F4/80+ Arg-1+ increased after low-level laser (1.00±0.07 vs. 2.15±0.12)(P<0.01). In the in vitro experiment, compared with the M1 group, the expression of the M1 macrophage marker iNOS in the M1+ low-level laser group decreased (1.00±0.11 vs. 0.08±0.01) (P<0.01); the M2 macrophage marker Arg-1 (1.00±0.14 vs. 2.44±0.16) (P<0.01), and the expression of CD206 (1.00±0.12 vs. 1.83±0.05) (P<0.01) increased. In addition, IL-1RA expression was increased in the M1+ low-level laser group compared with the M1 group (RT-qPCR: 1.00±0.00 vs. 2.27±0.22) (P<0.01) (ELISA: 1 435.58±100.48 vs. 2 006.12±123.91 (P<0.05); IL-10 expression was also increased in the M1+ low-level laser group compared with the M1 group (RT-qPCR: 1.00±0.00 vs. 3.45±0.56) (P<0.05) (ELISA: 137.13±4.20 vs.188.29±8.49) (P<0.01); compared with the M1 group, the macrophage polarization pathway protein in the M1+ low-level laser group increased, AKT (1.07±0.12 vs. 1.74±0.04) (P<0.01), p-AKT (1.00±0.12 vs. 1.64±0.15) (P<0.05), p-CREB (1.00±0.10 vs. 2.12±0.18) (P<0.01). Compared with the M1 group, the conditioned medium of the M1+ low-level laser group significantly promoted DRG axon growth (567.66±63.59 vs. 1 068.95±130.14) (P<0.05).
Conclusions
The 810 nm low-level laser irradiation can promote neuronal axon regeneration of mice with spinal cord injury, which may be related to the regulation of macrophage polarization phenotype by low-level laser through AKT/CREB pathway.
Key words:
Spinal cord injuries; Laser therapy; Macrophages; Nerve regeneration
Epilepsy is one of the most harmful diseases to human society. EEG(Electroencephalogram, EEG) can be used to detect the onset of epilepsy in a timely and effective manner. In this work, a detecting system based on convolutional neural network is designed. The system consists signal acquisition module, pre-processing module and neuromophic calculation module, and is capable of detecting epilepsy activity from EEG signal, the neuromophic module is designed based on memristor array. The system is simulated and tested on a 480-subject database corresponding to the 8 most significant channels from each patient for a total of 64 EEG signal channels, and have obtained the detecting accuracy of 98.46%, which proved its excellent performance on epilepsy detecting.
The observations and researches of five-planet are one of the important part of ancient calendars and also one of the methods to evaluate their accuracies.So astronomers paid much attention to this field.In《Hanshu·Tian wen zhi》and《Xuhanshu·Tian wen zhi》,there are 160 records with detailed dates and positions,which are calculated and studied by the modern astronomical method in this paper.The calculated results show that these positions are mostly correct,taking up 77.5%of the total records.While the rest 36 records are incorrect,taking up 22.5%.In addition,there are three typical or special forms of five-planet movements.The numbers ofshou,he,fanmovements are 14,22 and 46,taking up 9%,14%and 29%,respectively.In this paper,a detailed research on these three typical forms of five-planet movements is carried out.We think that the 36 incorrect records are caused by various reasons,but mainly in the data processes carried out by later generations.
The cancer area segmentation of esophageal histopathology images is a crucial step in determining the stage of esophageal cancer. This task is very important. However, manual segmentation will cost a lot of time. The rise of computational pathology has led to the development of automatic methods for cancer area detection. In the automatic segmentation problem, a well-labeled dataset is the most important part. One of the main contributions of this paper is to establish a dataset contains 1388 patches (958 Normal and 430 Abnormal containing tumor cells), marked with cancer, all of which are manually labeled and supervised by professional pathologists. We test the currently popular networks on our dataset, such as DeeplabV3, FCN+ResNet, Unet and so on. And FCN+ResNet achieves the best performance on our dataset with the highest Mean IoU (85.06%) and Pixel Acc (92.63%).
Head motion during brain PET imaging can introduce significant artifacts, reducing image resolution and affecting tracer distribution estimation. To address this issue, real-time head motion tracking and correction have become essential. In this study, we validate the United Imaging Healthcare markerless motion tracking system (UMT) embedded in the next-generation ultra-high performance brain PET scanner, NeuroEXPLORER (NX), using a phantom study and a human FDG study. We apply event-by-event and frame-based motion correction methods to reduce motion blurring and compare the UMT tracking result with the ground truth. Our results show that UMT achieves precise and real-time motion correction for rigid movements, with an error of 0.05 ± 0.04 mm and 0.02 ± 0.02°. The FWHM result demonstrates that employing UMT for motion tracking can improve spatial resolution. Human study results suggest that UMT-based motion correction has the potential to improve clinical applicability by ameliorating PET motion blurring. This study highlights the importance of real-time motion tracking and correction in achieving ideal spatial resolution and improving the clinical value of brain PET imaging. Future studies will focus on real-time motion correction in clinical studies and further evaluation of the UMT-based motion correction in NX.
'/%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)