Purpose
This prospective study aimed to evaluate the difference between
99m
Tc-PSMA single-photon emission computed tomography (SPECT)/CT and multiparametric magnetic resonance imaging (mpMRI) in the ...detection of primary prostate cancer (PCa).
Materials and methods
Fifty-six men with suspected PCa between October 2019 and November 2022 were prospectively enrolled in this study. The median age of the patients was 70 years (range, 29-87 years). Patients were divided into high-(Gleason score>7, n=31), medium- (Gleason score=7, n=6) and low-risk groups (Gleason score < 7, n=6). All patients underwent
99m
Tc-PSMA SPECT/CT and mpMRI at an average interval of 3 days (range, 1-7 days). The maximum standardized uptake value (SUV
max
), the minimum apparent diffusion coefficient (ADC
min
), and their ratio (SUV
max
/ADC
min
) were used as imaging parameters to distinguish benign from malignant prostatic lesions.
Results
Of the 56 patients, 12 were pathologically diagnosed with a benign disease, and 44 were diagnosed with PCa.
99m
Tc-PSMA SPECT/CT and mpMRI showed no significant difference in the detection of primary PCa (kappa =0.401,
P
=0.002), with sensitivities of 97.7% (43/44) and 90.9% (40/44), specificities of 75.0% (9/12) and 75.0% (9/12), and AUC of 97.4% and 95.1%, respectively. The AUC of SUV
max
/ADC
min
was better than those of SUV
max
or ADC
min
alone. When SUV
max
/ADC
min
in the prostatic lesion was >7.0×10
3
, the lesion was more likely to be malignant. When SUV
max
/ADC
min
in the prostatic lesion is >27.0×10
3
, the PCa patient may have lymph node and bone metastases. SUV
max
was positively correlated with the Gleason score (
r
=0.61, P=0.008), whereas ADC
min
was negatively correlated with the Gleason score (
r
=-0.35,
P
=0.023). SUV
max
/ADC
min
was positively correlated with the Gleason score (
r
=0.59,
P
=0.023). SUV
max
/ADC
min
was the main predictor of the high-risk group, with an optimal cut-off value of 15.0×10
3
.
Conclusions
The combination of
99m
Tc-PSMA SPECT/CT and mpMRI can improve the diagnostic efficacy for PCa compared with either modality alone; SUV
max
/ADC
min
is a valuable differential diagnostic imaging parameter.
Abstract
The most common site of metastasis of prostate cancer (PCa) is bone. Skeletal-related events can increase the risk of death in patients with PCa by 28%. Due to the low detection rate of ...lesions in patients with low prostate-specific antigen (PSA) levels, the value of
99m
Tc methylene diphosphonate (
99m
Tc-MDP) bone scintigraphy is limited. Prostate-specific membrane antigen (PSMA) is a small molecular probe that can efficiently and specifically detect PCa lesions. This prospective study aimed to evaluate the difference between
99m
Tc-PSMA single-photon emission computed tomography (SPECT)/CT and
99m
Tc-MDP SPECT/CT in the detection of bone metastasis in PCa. A total of 74 men with pathologically confirmed PCa from October 2019 to November 2021 were prospectively enrolled in this study. The median age was 70 (range, 55–87) years. All patients underwent both
99m
Tc-PSMA SPECT/CT and
99m
Tc-MDP SPECT/CT at an average interval of 12.1 (range, 1–14) days. The detected imaging-positive bone lesions were scored as “typical metastasis” or “equivocal metastasis” by a standard reporting schema. Subsequent therapy modality details were observed through follow-up. Twenty-five of the 74 patients were diagnosed with bone metastases.
99m
Tc-PSMA SPECT/CT and
99m
Tc-MDP SPECT/CT detected 20 and 18 bone metastases, with sensitivities of 80.0% (20/25) and 72.0% (18/25), specificities of 100.0% (49/49) and 81.3% (40/49), and AUCs of 88.0% and 84.9%, respectively. There was a significant difference in the AUC between the two imaging methods (
P
< 0.001). In an analysis of the number of bone metastasis lesions, the proportion of “typical metastasis” versus “equivocal metastasis” detected by the two imaging methods was 26.3:1 (PSMA) and 2.9:1 (MDP), and the difference was statistically significant (
P
= 0.005). There was a significant difference in the detection of bone metastatic lesions by
99m
Tc-PSMA and
99m
Tc-MDP when the maximum diameter of the lesions was ≤ 0.6 cm (
P
< 0.05). The optimal cut-off value for PSA was 2.635 ng/mL (PSMA) and 15.275 ng/mL (MDP).
99m
Tc-PSMA SPECT/CT led to a change in management to a more individualized therapy modality for 11 of 74 men (14.9%).
99m
Tc-PSMA SPECT/CT was superior to
99m
Tc-MDP SPECT/CT in the detection of bone metastases in PCa, especially for small lesions and in patients with low PSA levels, and demonstrated an additional benefit of providing information on extraskeletal metastases. With regard to therapy,
99m
Tc-PSMA scans might have utility in improving the subsequent therapy modality.
Spectrum sensing is the key technology in enabling spectrum awareness in cognitive radio. The performance of the spectrum sensing depends on the sensing method. In this paper, by means of random ...matrix theory (RMT), a novel spectrum sensing scheme and a threshold decision rule are proposed based on the difference between the maximum and the minimum eigenvalue (DMM) cooperative spectrum sensing algorithm, analyzing the minimum eigenvalue limiting distribution of the covariance matrix of the received signals from multiple cognitive users(CUs). The proposed scheme can overcome the noise uncertainty and need nothing about prior knowledge of the signal transmitted from the primary user (PU). Simulation results show that the proposed scheme outperforms DMM under lower signal-to-noise ratio (SNR) and smaller samples.
Aiming at the application of multi-target passive tracking, a new multi -target passive tracking method based on three array fusion and particle fitler was proposed. Firstly, the "fake points" could ...be almost entirely and exactly deleted with the aids of the three-array at the expense of an additional array. Secondly, considered the fact that the measurements gotten from each array were independent in passive tracking system, a novel sequential particle filter with improved distribution was proposed, and the influence of non-linear and non-Gauss could be eliminated by the presented method. The relative performances, as measurement by track lifetime, position tracking error and RMS tracking error are compared with the classical methods. The results show that there is significant improvement in tracking capability over MHEKF and IMM-PD methods.
The rapid development in deep learning and computer vision has introduced new opportunities and paradigms for building extraction from remote sensing images. In this paper, we propose a novel fully ...convolutional network (FCN), in which a spatial residual inception (SRI) module is proposed to capture and aggregate multi-scale contexts for semantic understanding by successively fusing multi-level features. The proposed SRI-Net is capable of accurately detecting large buildings that might be easily omitted while retaining global morphological characteristics and local details. On the other hand, to improve computational efficiency, depthwise separable convolutions and convolution factorization are introduced to significantly decrease the number of model parameters. The proposed model is evaluated on the Inria Aerial Image Labeling Dataset and the Wuhan University (WHU) Aerial Building Dataset. The experimental results show that the proposed methods exhibit significant improvements compared with several state-of-the-art FCNs, including SegNet, U-Net, RefineNet, and DeepLab v3+. The proposed model shows promising potential for building detection from remote sensing images on a large scale.
Traveling Salesman Problem(TSP) is a main attention issue at present. Neural network can be used to solve combinatorial optimization problems. In recent years, there have existed many neural network ...methods for solving TSP, which has made a big step forward for solving combinatorial optimization problems. This paper reviews the neural network methods for solving TSP in recent years, including Hopfield neural network, graph neural network and neural network with reinforcement learning. Using neural network to solve TSP can effectively improve the accuracy of the approximate solution. Finally, we put forward the prospect of solving TSP in the future.
Epoxy resins are a class of viscoelastic hygroscopic polymers widely used in plastic packaged devices. Stress relaxation and yielding may occur in plastic packaged devices under thermal-hygro ...environments, which seriously affect the reliability of the devices. It is important to obtain accurate viscoelastic parameters of the hygroscopic materials for analyzing the stresses and strains of plastic packaged devices. In this paper, a novel method for measuring the viscoelastic parameters of epoxy resins under water bath environments by using embedded strain gauges is proposed. A kind of epoxy resin-copper bi-layer plate is fabricated, and the biaxial strain gauges are embedded between the epoxy resin layer and the copper layer to monitor the deformations. The interfacial strains of the two test pieces at 85 °C/100% relative humidity are continuously measured up to 144 h. Combine the experimental strain curves and the analytical representation, the viscoelastic parameters of the epoxy resins are obtained by curve fitting with MATLAB and genetic algorithm. It is found that the viscoelasticity of epoxy resins could be described by an exponential function of order five shear modulus and relaxation times, and the interfacial strain curves obtained by genetic algorithm fit well with the experimental ones. The proposed testing and solving methods provide a valuable reference for getting the time-dependent viscoelasticities of epoxy resins under thermal-hygro environments.
•The analytical representations on the hygro-mechanical behavior of the elastic-viscoelastic bi-layer plate are deduced.•The interfacial hygro-mechanical strains of the epoxy resin-copper bi-layer plates are measured exploratory in water.•The viscoelasticities of the epoxy resins are solved by fitting the strain curves with MATLAB based on the genetic algorithm.•The method provides a novel means to analyze the viscoelasticity of hygroscopic materials under water bath environments.
In plants, reactive oxygen species (ROS) produced following the expression of the respiratory burst oxidase homolog (Rboh) gene are important regulators of stress responses. However, little is known ...about how plants acclimate to salt stress through the Rboh-derived ROS signaling pathway. Here, we showed that a 400-bp fragment of the tobacco (Nicotiana tabacum) NtRbohE promoter played a critical role in the salt response. Using yeast one-hybrid (Y1H) screens, NtbHLH123, a bHLH transcription factor, was identified as an upstream partner of the NtRbohE promoter. These interactions were confirmed by Y1H, electrophoretic mobility assay, and chromatin immunoprecipitation assays. Overexpression of NtbHLH123 resulted in greater resistance to salt stress, while NtbHLH123-silenced plants had reduced resistance to salt stress. We also found that NtbHLH123 positively regulates the expression of NtRbohE and ROS production soon after salt stress treatment. Moreover, knockout of NtRbohE in the 35S::NtbHLH123 background resulted in reduced expression of ROS-scavenging and salt stress-related genes and salt tolerance, suggesting that NtbHLH123-regulated salt tolerance is dependent on the NtbHLH123-NtRbohE signaling pathway. Our data show that NtbHLH123 is a positive regulator and acts as a molecular switch to control a Rboh-dependent mechanism in response to salt stress in plants.
•Present a federated learning method called Fed ℓ1 to address client drift.•Introduce ℓ1 regularizer to control clients and avoid unnecessary parameter updates.•Design a stochastic subgradient ...descent algorithm to train the model.•Compare Fed ℓ1 with baselines on several datasets and discuss the improvement.
Federated Learning (FL) is a widely adopted deep learning method that does not require the collection of raw training data and solves specific learning tasks by federating distributed devices. Due to the heterogeneous distribution of data across clients, the clients will drift toward the local optimal solutions during local training and result in different local models. The global model after aggregating these different local models may keep away from the global optimal solution. This phenomenon is known as client drift, which often hinders the performance of FL. Parameter regularization methods address this challenge of client drift by controlling the update direction of each client. They consider the global model as both the starting point and a reference for the induction bias in the penalty. However, the existing regularization approaches produce dense solutions so that all parameters need to be updated during local model training. At the same time, we note that some studies on deep learning have found that it is unnecessary to update all parameters at each round. Therefore, in this work, we design a novel FL training approach called Fedℓ1 which can alleviate the performance degradation of FL by updating only part of the parameters at each round. ℓ1 regularization is utilized to control the update direction of each client and avoid unnecessary parameter updates at the same time. To our knowledge, our study is the first to introduce sparse regularization term to correct the local training of individual clients in FL. We design a stochastic subgradient descent algorithm to train the ℓ1-regularized nonsmooth model. The comparison experiments with state-of-the-art baselines verify the superiority of the proposed approach.