Self-adaption and texture generation: A hybrid loss function for low-dose CT denoising.

BACKGROUND: Deep learning has been successfully applied to low-dose CT (LDCT) denoising. But the training of the model is very dependent on an appropriate loss function. Existing denoising models often use per-pixel loss, including mean abs error (MAE) and mean square error (MSE). This ignores the difference in denoising difficulty between different regions of the CT images and leads to the loss of large texture information in the generated image. PURPOSE: In this paper, we propose a new hybrid loss function that adapts to the noise in different regions of CT images to balance the denoising difficulty and preserve texture details, thus acquiring CT images with high-quality diagnostic value using LDCT images, providing strong support for condition diagnosis. METHODS: We propose a hybrid loss function consisting of weighted patch loss (WPLoss) and high-frequency information loss (HFLoss). To enhance the model's denoising ability of the local areas which are difficult to denoise, we improve the MAE to obtain WPLoss. After the generated image and the target image are divided into several patches, the loss weight of each patch is adaptively and dynamically adjusted according to its loss ratio. In addition, considering that texture details are contained in the high-frequency information of the image, we use HFLoss to calculate the difference between CT images in the high-frequency information part. RESULTS: Our hybrid loss function improves the denoising performance of several models in the experiment, and obtains a higher peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). Moreover, through visual inspection of the generated results of the comparison experiment, the proposed hybrid function can effectively suppress noise and retain image details. CONCLUSIONS: We propose a hybrid loss function for LDCT image denoising, which has good interpretation properties and can improve the denoising performance of existing models. And the validation results of multiple models using different datasets show that it has good generalization ability. By using this loss function, high-quality CT images with low radiation are achieved, which can avoid the hazards caused by radiation and ensure the disease diagnosis for patients.

Reinforcement Learning Based Diagnosis and Prediction for COVID-19 by Optimizing a Mixed Cost Function From CT Images.

A novel coronavirus disease (COVID-19) is a pandemic disease has caused 4 million deaths and more than 200 million infections worldwide (as of August 4, 2021). Rapid and accurate diagnosis of COVID-19 infection is critical to controlling the spread of the epidemic. In order to quickly and efficiently detect COVID-19 and reduce the threat of COVID-19 to human survival, we have firstly proposed a detection framework based on reinforcement learning for COVID-19 diagnosis, which constructs a mixed loss function that can integrate the advantages of multiple loss functions. This paper uses the accuracy of the validation set as the reward value, and obtains the initial model for the next epoch by searching the model corresponding to the maximum reward value in each epoch. We also have proposed a prediction framework that integrates multiple detection frameworks using parameter sharing to predict the progression of patients' disease without additional training. This paper also constructed a higher-quality version of the CT image dataset containing 247 cases screened by professional physicians, and obtained more excellent results on this dataset. Meanwhile, we used the other two COVID-19 datasets as external verifications, and still achieved a high accuracy rate without additional training. Finally, the experimental results show that our classification accuracy can reach 98.31%, and the precision, sensitivity, specificity, and AUC (Area Under Curve) are 98.82%, 97.99%, 98.67%, and 0.989, respectively. The accuracy of external verification can reach 93.34% and 91.05%. What's more, the accuracy of our prediction framework is 91.54%. A large number of experiments demonstrate that our proposed method is effective and robust for COVID-19 detection and prediction.

Data collection for mobile crowdsensing in the presence of selfishness.

Mobile crowdsensing is an emerging approach to data collection by exploiting the sensing abilities offered by smart phones and users' mobility. Data collection can be implemented by exploiting the forwarding opportunities given by the contacts between nodes. However, as cell phones are still resource constrained, most people are socially selfish so that they may not always cooperate with each other in data collection. In this paper, we propose a routing protocol, called Accept aNd Tolerate (ANT), which is tailored for data collection in a social environment with selfish individuals. ANT works by accepting and tolerating social selfishness as an unavoidable human characteristic. It makes relay selection based on nodes' contacts and their willingness to cooperate. The cooperative willingness of selfish nodes is measured rationally according to the reciprocity relationship between nodes and their resource constraints. Through assessing the worthiness of carrying and forwarding a packet, ANT proposes a buffer management scheme and makes forwarding decisions. Simulations based on real traces show that ANT achieves better performance under resource-constrained circumstances than other comparable approaches.

On prolonging network lifetime through load-similar node deployment in wireless sensor networks.

This paper is focused on the study of the energy hole problem in the Progressive Multi-hop Rotational Clustered (PMRC)-structure, a highly scalable wireless sensor network (WSN) architecture. Based on an analysis on the traffic load distribution in PMRC-based WSNs, we propose a novel load-similar node distribution strategy combined with the Minimum Overlapping Layers (MOL) scheme to address the energy hole problem in PMRC-based WSNs. In this strategy, sensor nodes are deployed in the network area according to the load distribution. That is, more nodes shall be deployed in the range where the average load is higher, and then the loads among different areas in the sensor network tend to be balanced. Simulation results demonstrate that the load-similar node distribution strategy prolongs network lifetime and reduces the average packet latency in comparison with existing nonuniform node distribution and uniform node distribution strategies. Note that, besides the PMRC structure, the analysis model and the proposed load-similar node distribution strategy are also applicable to other multi-hop WSN structures.

A distance-aware replica adaptive data gathering protocol for Delay Tolerant Mobile Sensor Networks.

In Delay Tolerant Mobile Sensor Networks (DTMSNs) that have the inherent features of intermitted connectivity and frequently changing network topology it is reasonable to utilize multi-replica schemes to improve the data gathering performance. However, most existing multi-replica approaches inject a large amount of message copies into the network to increase the probability of message delivery, which may drain each mobile node's limited battery supply faster and result in too much contention for the restricted resources of the DTMSN, so a proper data gathering scheme needs a trade off between the number of replica messages and network performance. In this paper, we propose a new data gathering protocol called DRADG (for Distance-aware Replica Adaptive Data Gathering protocol), which economizes network resource consumption through making use of a self-adapting algorithm to cut down the number of redundant replicas of messages, and achieves a good network performance by leveraging the delivery probabilities of the mobile sensors as main routing metrics. Simulation results have shown that the proposed DRADG protocol achieves comparable or higher message delivery ratios at the cost of the much lower transmission overhead than several current DTMSN data gathering schemes.

Energy-efficient algorithm for sensor networks with non-uniform maximum transmission range.

In wireless sensor networks (WSNs), the energy hole problem is a key factor affecting the network lifetime. In a circular multi-hop sensor network (modeled as concentric coronas), the optimal transmission ranges of all coronas can effectively improve network lifetime. In this paper, we investigate WSNs with non-uniform maximum transmission ranges, where sensor nodes deployed in different regions may differ in their maximum transmission range. Then, we propose an Energy-efficient algorithm for Non-uniform Maximum Transmission range (ENMT), which can search approximate optimal transmission ranges of all coronas in order to prolong network lifetime. Furthermore, the simulation results indicate that ENMT performs better than other algorithms.

An agenda-based routing protocol in delay tolerant mobile sensor networks.

Routing in delay tolerant mobile sensor networks (DTMSNs) is challenging due to the networks' intermittent connectivity. Most existing routing protocols for DTMSNs use simplistic random mobility models for algorithm design and performance evaluation. In the real world, however, due to the unique characteristics of human mobility, currently existing random mobility models may not work well in environments where mobile sensor units are carried (such as DTMSNs). Taking a person's social activities into consideration, in this paper, we seek to improve DTMSN routing in terms of social structure and propose an agenda based routing protocol (ARP). In ARP, humans are classified based on their agendas and data transmission is made according to sensor nodes' transmission rankings. The effectiveness of ARP is demonstrated through comprehensive simulation studies.

Minimum expected delay-based routing protocol (MEDR) for Delay Tolerant Mobile Sensor Networks.

It is a challenging work to develop efficient routing protocols for Delay Tolerant Mobile Sensor Networks (DTMSNs), which have several unique characteristics such as sensor mobility, intermittent connectivity, energy limit, and delay tolerability. In this paper, we propose a new routing protocol called Minimum Expected Delay-based Routing (MEDR) tailored for DTMSNs. MEDR achieves a good routing performance by finding and using the connected paths formed dynamically by mobile sensors. In MEDR, each sensor maintains two important parameters: Minimum Expected Delay (MED) and its expiration time. According to MED, messages will be delivered to the sensor that has at least a connected path with their hosting nodes, and has the shortest expected delay to communication directly with the sink node. Because of the changing network topology, the path is fragile and volatile, so we use the expiration time of MED to indicate the valid time of the path, and avoid wrong transmissions. Simulation results show that the proposed MEDR achieves a higher message delivery ratio with lower transmission overhead and data delivery delay than other DTMSN routing approaches.

A community-based event delivery protocol in publish/subscribe systems for delay tolerant sensor networks.

The basic operation of a Delay Tolerant Sensor Network (DTSN) is to finish pervasive data gathering in networks with intermittent connectivity, while the publish/subscribe (Pub/Sub for short) paradigm is used to deliver events from a source to interested clients in an asynchronous way. Recently, extension of Pub/Sub systems in DTSNs has become a promising research topic. However, due to the unique frequent partitioning characteristic of DTSNs, extension of a Pub/Sub system in a DTSN is a considerably difficult and challenging problem, and there are no good solutions to this problem in published works. To ad apt Pub/Sub systems to DTSNs, we propose CED, a community-based event delivery protocol. In our design, event delivery is based on several unchanged communities, which are formed by sensor nodes in the network according to their connectivity. CED consists of two components: event delivery and queue management. In event delivery, events in a community are delivered to mobile subscribers once a subscriber comes into the community, for improving the data delivery ratio. The queue management employs both the event successful delivery time and the event survival time to decide whether an event should be delivered or dropped for minimizing the transmission overhead. The effectiveness of CED is demonstrated through comprehensive simulation studies.