No abstract

Platelet detection and counting play a greatly significant role in medical field, especially in routine blood tests which can be used to judge blood status and diagnose related diseases. Therefore, platelet detection is valuable for diagnosing related blood diseases such as liver-related diseases. Blood analyzers and visual microscope counting were widely used for platelet detection, but the experimental procedure took nearly 20 minutes and can only be performed by a professional doctor. In recent years, technological breakthroughs in artificial intelligence have made it possible to detect red blood cells through deep learning methods. However, due to the inaccessibility of platelet datasets and the small size of platelets, deep learning-based platelet detection studies are almost nonexistent. In this paper, we carried out experiments for platelet detection based on commonly used object detection models, such as Single Shot Multibox Detector (SSD), RetinaNet, Faster_rcnn, and You Only Look Once_v3 (YOLO_v3). Compared with the other three models, YOLO_v3 can detect platelets more effectively. And we proposed three ideas for improvement based on YOLO_v3. Our study demonstrated that YOLO_v3 can be adopted for platelet detection accurately and in real time. We also implemented YOLO_v3 with multiscale fusion, YOLO_v3 with anchor box clustering, and YOLO_v3 with match parameter on our self-created dataset and, respectively, achieved 1.8% higher average precision (AP), 2.38% higher AP, and 2.05% higher AP than YOLO_v3. The comprehensive experiments revealed that YOLO_v3 with the improved ideas performs better in platelet detection than YOLO_v3.

In image deconvolution, the boundary value problem, if not appropriately handled, often causes serious ringing artifacts in the restored results. This paper proposes a simple method to tackle this problem without any assumption on the noise level and the symmetry of the Point Spread Function (PSF). We establish new boundary conditions by smoothly expanding the input image to a large tile. It helps reducing the boundary discontinuities and accordingly makes all restoration filters based on Fast Fourier Transform (FFT) not produce obvious image border artifacts.

This paper presents a novel dynamic load-aware based load-halanced routing (DLBL) algorithm for ad hoc networks. DLBL considers intermediate node routing load as the primary route selection metric while discovering a route, which can lead to less network congestion and bottlenecks. During the route maintaining periods, DLBL deals by a distributed mechanism with the congestion of active routes when nodes of the routes have their queue overload, at the same time with little overhead increment. When link breaks because of the node mobility or power off, etc, DLBL provides efticient path maintenance to patch up broken links to help get the robust route from the source to the destination. By presenting and analyzing simulation results, the DLBL is shown to result in good performance of packet delivery ratio and average end-to-end delay, while exhibiting many attractive features of distributed control to adapt to the dynamic ad hoc networks.

The ad-hoc network [1] integrates the computer network and mobile communication without control centers, and is self-built, self-organized, and self-managed. It is very useful in military and commercial applications such as national defense readiness, emergency rescue, exploration missions, and sensor networks, all of which have a need for ubiquitous communication services without the presence of a fixed infrastructure. Route protocol design is a more difficult task in ad-hoc networks because of the frequent node mobility, limited battery energy and bandwidth, and high bit error rate. In order to manage frequent topology changes caused by node mobility and optimize multiple quality-of-service (QoS) parameters in ad-hoc networks, the control overheads must be reduced as much as possible. The alternate path route (APR) protocols improve the robustness of existing on-demand routing protocols against high node mobility through alternate path construction, thus reducing the flooding frequency and overheads of route rediscovery, whereas most existing alternate path routing protocols focus on the alternate path construction and handover and do not consider the aging mechanism of alternate paths. Therefore, it is difficult to assess the status of the APR set. Although most current APR protocols achieve as many alternate paths as possible for a single source-destination pair during route discovery at a low cost, the alternate path route set is unidirectionally constructed just from source to the destination by flooding, which is inefficient for bilateral communication. Besides, few existing APR protocols support QoS. In addition, centralized routing mechanism in ad-hoc networks with frequent topology changes may result in the increase of overheads, low reliability, and scalability, resulting in a bottleneck of traffic flow at the control center. Owing to its good distributed control mechanism, ant-colony optimization based routing protocols provide many redundant routes and reliable connectivity, which makes the system very robust against Abstract In order to periodically reassess the status of the alternate path route (APR) set and to improve the efficiency of alternate path construction existing in most current alternate path routing protocols, we present a cross-layer design and ant-colony optimization based load-balancing routing protocol for ad-hoc networks (CALRA) in this paper. In CALRA, the APR set maintained in nodes is aged and reassessed by the inherent mechanism of pheromone evaporation of ant-colony optimization algorithm, and load balance of network is achieved by ant-colony optimization combining with cross-layer synthetic optimization. The efficiency of APR set construction is improved by bidirectional and hop-by-hop routing update during routing discovery and routing maintenance process. Moreover, ants in CALRA deposit simulated pheromones as a function of multiple parameters corresponding to the information collected by each layer of each node visited, such as the distance from their source node, the con...