This article proposes a novel multiagent deep reinforcement learning-based algorithm which can realize federated learning (FL) computation with Internet-of-Underwater-Things (IoUT) devices in the ...ocean environment. According to the fact that underwater networks are relatively not easy to set up reliable links by huge fading compared to wireless free-space air medium, gathering all training data for conducting centralized deep learning training is not easy. Therefore, FL-based distributed deep learning can be a suitable solution for this application. In this IoUT network (IoUT-Net) scenario, the FL system needs to construct a global learning model by aggregating the local model parameters that are obtained from individual IoUT devices. In order to reliably deliver the parameters from IoUT devices to a centralized FL machine, base station like devices are needed. Therefore, a joint cell association and resource allocation (JCARA) method is required and it is designed inspired by multiagent deep deterministic policy gradient (MADDPG) to deal with distributed situations and unexpected time-varying states. The performance evaluation results show that our proposed MADDPG-based algorithm achieves 80% and 41% performance improvements than the standard actor-critic and DDPG, respectively, in terms of the downlink throughput.
Lipid‐based nanoparticles have emerged as a clinically viable platform technology to deliver nucleic acids for a wide range of healthcare applications. Within this scope, one of the most exciting ...areas of recent progress and future innovation potential lies in the material science of lipid‐based nanoparticles, both to refine existing nanoparticle strategies and to develop new ones. Herein, the latest efforts to develop next‐generation lipid‐based nanoparticles are covered by taking a nanoarchitectonics perspective and the design, nucleic acid encapsulation methods, scalable production, and application prospects are critically analyzed for three classes of lipid‐based nanoparticles: 1) traditional lipid nanoparticles (LNPs); 2) lipoplexes; and 3) bicelles. Particular focus is placed on rationalizing how molecular self‐assembly principles enable advanced functionalities along with comparing and contrasting the different nanoarchitectures. The current development status of each class of lipid‐based nanoparticle is also evaluated and possible future directions in terms of overcoming clinical translation challenges and realizing new application opportunities are suggested.
Lipid‐based nanoparticles are transforming nucleic acid delivery options for healthcare and biotechnology applications. One of the most exciting areas of recent progress and future innovation potential lies in the material science of lipid‐based nanoparticles, especially when taking a nanoarchitectonics perspective to rationalize how molecular‐level control over nanoparticle structure can enable an increasingly wide range of application possibilities.
This paper presents a novel approach for enhancing autonomous drone mobility control using deep reinforcement learning (DRL), primarily aimed at improving autonomous navigation in challenging ...environments. Our research tackles the significant issue of real-time obstacle avoidance, a critical aspect in drone control. This is achieved through the integration of sensing-aware nonlinear control mechanisms, facilitating advanced trajectory optimization. A notable contribution of our work is the incorporation of real-time human-in-the-loop feedback through human-computer interaction (HCI), which is crucial when pre-trained DRL models encounter environments they are not fully adapted to. Combining autonomous DRL control and HCI feedback equips our system with the flexibility to handle unforeseen scenarios effectively. Furthermore, the paper showcases a comprehensive software demonstration employing Unity 3D for visualization. This demonstration highlights the practical application of our sensing-aware nonlinear control and the HCI-based feedback system, using keyboard interfaces for real-time interaction. The accompanying demonstration video distinctly exhibits the ability of our proposed algorithm.
Volumetric imaging by fluorescence microscopy is often limited by anisotropic spatial resolution, in which the axial resolution is inferior to the lateral resolution. To address this problem, we ...present a deep-learning-enabled unsupervised super-resolution technique that enhances anisotropic images in volumetric fluorescence microscopy. In contrast to the existing deep learning approaches that require matched high-resolution target images, our method greatly reduces the effort to be put into practice as the training of a network requires only a single 3D image stack, without a priori knowledge of the image formation process, registration of training data, or separate acquisition of target data. This is achieved based on the optimal transport-driven cycle-consistent generative adversarial network that learns from an unpaired matching between high-resolution 2D images in the lateral image plane and low-resolution 2D images in other planes. Using fluorescence confocal microscopy and light-sheet microscopy, we demonstrate that the trained network not only enhances axial resolution but also restores suppressed visual details between the imaging planes and removes imaging artifacts.
Open-top light-sheet microscopy (OT-LSM) is a specialized microscopic technique for the high-throughput cellular imaging of optically cleared, large-sized specimens, such as the brain. Despite the ...development of various OT-LSM techniques, achieving submicron resolution in all dimensions remains. We developed a high-resolution open-top axially swept LSM (HR-OTAS-LSM) for high-throughput and high-resolution imaging in all dimensions. High axial and lateral resolutions were achieved by using an aberration-corrected axially swept excitation light sheet in the illumination arm and a high numerical aperture (NA) immersion objective lens in the imaging arm, respectively. The high-resolution, high-throughput visualization of neuronal networks in mouse brain and retina specimens validated the performance of HR-OTAS-LSM. The proposed HR-OTAS-LSM method represents a significant advancement in the high-resolution mapping of cellular networks in biological systems such as the brain and retina.
This paper describes the preparation of giant unilamellar vesicles with reconstituted hamster P-glycoprotein (Pgp, ABCB1) for studying the transport activity of this efflux pump in individual ...liposomes using optical microscopy. Pgp, a member of ABC (ATP-binding cassette) transporter family, is known to contribute to the cellular multidrug resistance (MDR) against variety of drugs. The efficacy of many therapeutics is, thus, hampered by this efflux pump, leading to a high demand for simple and effective strategies to monitor the interactions of candidate drugs with this protein. Here, we applied small Pgp proteoliposomes to prepare giant Pgp-bearing liposomes via modified electroformation techniques. The presence of Pgp in the membrane of giant proteoliposomes was confirmed using immunohistochemistry. Assessment of Pgp ATPase activity suggested that this transporter retained its activity upon reconstitution into giant liposomes, with an ATPase specific activity of 439 ± 103 nmol/mg protein/min. For further confirmation, we assessed the transport activity of Pgp in these proteoliposomes by monitoring the translocation of rhodamine 123 (Rho123) across the membrane using confocal microscopy at various ATP concentrations (0-2 mM) and in the presence of Pgp inhibitors. Rate of change in Rho123 concentration inside the liposomal lumen was used to estimate the Rho123 transport rates (1/s) for various ATP concentrations, which were then applied to retrieve the Michaelis-Menten constant (Km) of ATP in Rho123 transport (0.42 ± 0.75 mM). Similarly, inhibitory effects of verapamil, colchicine, and cyclosporin A on Pgp were studied in this system and the IC50 values for these Pgp inhibitors were found 26.6 ± 6.1 μM, 94.6 ± 47.6 μM, and 0.21 ± 0.07 μM, respectively. We further analyzed the transport data using a kinetic model that enabled dissecting the passive diffusion of Rho123 from its Pgp-mediated transport across the membrane. Based on this model, the permeability coefficient of Rho123 across the liposomal membrane was approximately 1.25×10-7 cm/s. Comparing the membrane permeability in liposomes with and without Pgp revealed that the presence of this protein did not have a significant impact on membrane integrity and permeability. Furthermore, we used this model to obtain transport rate constants for the Pgp-mediated transport of Rho123 (m3/mol/s) at various ATP and inhibitor concentrations, which were then applied to estimate values of 0.53 ± 0.66 mM for Km of ATP and 25.2 ± 5.0 μM for verapamil IC50, 61.8 ± 34.8 μM for colchicine IC50, and 0.23 ± 0.09 μM for cyclosporin A IC50. The kinetic parameters obtained from the two analyses were comparable, suggesting a minimal contribution from the passive Rho123 diffusion across the membrane. This approach may, therefore, be applied for screening the transport activity of Pgp against potential drug candidates.
In many deep neural network (DNN) applications, the difficulty of gathering high‐quality data in industry fields hinders the practical use of DNN. Thus, the concept of transfer learning (TL) has ...emerged, which leverages the pretrained knowledge of the DNN which was built based on large‐scale datasets. For this TL objective, this paper suggests two‐stage architectural fine‐tuning for reducing the costs and time while exploring the most efficient DNN model, inspired by neural architecture search (NAS). The first stage is mutation, which reduces the search costs using a priori architectural information. Moreover, the next stage is early‐stopping, which reduces NAS costs by terminating the search process in the middle of computation. The data‐intensive experimental results verify that the proposed method outperforms benchmarks.
This paper suggests two‐stage architectural fine‐tuning for reducing the costs and time while exploring the most efficient neural network model, inspired by neural architecture search (NAS). The first stage is mutation, which reduces the search costs using a priori architectural information. Moreover, the next stage is early‐stopping, which reduces NAS costs by terminating the search process in the middle of computation.
RNF43 is an E3 ligase that inhibits Wnt signaling by ubiquitinating Wnt receptors for degradation. It is mutated in various cancer types with the most recurrent mutation being the frameshift ...G659Vfs*41 with frequencies of ~5-8% in colon, stomach and endometrial cancers. This mutation, a deletion of G in a 7-G repeat, has been assumed to encode an inactive enzyme that would lead to increased Wnt signaling and drive tumorigenesis, yet no functional characterization has been reported. We analyzed the distribution of G659Vfs*41 and its association with other cancer gene mutations, and found that the mutation occurred nearly exclusively in tumors with low expression of the DNA mismatch repair gene MLH1. Mutant RNF43-G659Vfs*41 was no different from wild type RNF43 in expression, stability, localization, R-spondin binding, and inhibition of Wnt signaling. No dominant negative activity of the mutant was observed. Colon tumors with RNF43-G659Vfs*41 had low Wnt/β-catenin signaling and were frequently mutated in BRAF. A colon cancer cell line with RNF43-G659Vfs*41 and BRAF-V600E mutations was sensitive to activation of Wnt/β-catenin signaling. These findings suggest that the frequent occurrence of RNF43-G659Vfs*41 may result from error-prone replication of the 7-G repeat in MLH1-deficient tumors and that the mutation itself does not inactivate enzyme.
Green tide, which is a serious water pollution problem, is caused by the complex relationships of various factors, such as flow rate, several water quality indicators, and weather. Because the ...existing methods are not suitable for identifying these relationships and making accurate predictions, a new system and algorithm is required to predict the green tide phenomenon and also minimize the related damage before the green tide occurs. For this purpose, we consider a new network model using smart sensor-based federated learning which is able to use distributed observation data with geologically separated local models. Moreover, we design an optimal scheduler which is beneficial to use real-time big data arrivals to make the overall network system efficient. The proposed scheduling algorithm is effective in terms of (1) data usage and (2) the performance of green tide occurrence prediction models. The advantages of the proposed algorithm is verified via data-intensive experiments with real water quality big-data.
Skin cancer is one of the most prevalent cancers in the Caucasian population. In the United States, it is estimated that at least one in five people will develop skin cancer in their lifetime, ...leading to significant morbidity and a healthcare burden. Skin cancer mainly arises from cells in the epidermal layer of the skin, where oxygen is scarce. There are three main types of skin cancer: malignant melanoma, basal cell carcinoma, and squamous cell carcinoma. Accumulating evidence has revealed a critical role for hypoxia in the development and progression of these dermatologic malignancies. In this review, we discuss the role of hypoxia in treating and reconstructing skin cancers. We will summarize the molecular basis of hypoxia signaling pathways in relation to the major genetic variations of skin cancer.