Search

Artificial intelligence models implemented in power-efficient Internet-of-Things (IoT) devices have accuracy degradation due to limited power consumption. To mitigate the accuracy loss on IoT devices, an edge-server joint inference system is introduced. On the edge-server inference system, allocate more workloads to the server end can mitigate accuracy loss, but data transmission contributes to the power consumption of the edge device. Thus, in this article, we present a novel two-stage method to allocate workloads to the server or the edge to maximize inference accuracy under a power constraint. In the first stage, we present a clusterwise threshold-based method for estimating the trustworthiness of a prediction made at the edge. In the second stage, we further determine the workload allocation of a trustworthy image based on the probability of the top 1 prediction and the power constraint. In addition, we propose a fine-tuning process to the pretrained model at the edge for achieving better accuracy. In the experiments, we apply the proposed method to several well-known deep neural network models. The results show that the proposed method can improve inference accuracy up to 3.93% under a specific power constraint compared to previous methods.

Background: The use of a nonthermal plasma (NTP) jet in the treatment of living tissue has been the subject of considerable interest in the field of medical technology, and has the potential to reduce the recovery time of open wounds. We aimed to investigate the wound-healing process by clinical observation, blood tests, and expression of cell adhesion markers and reactive oxygen species in NTP jet-treated rats. Methods: This study utilized Sprague-Dawley (SD) rats as experimental subjects, and wounds measuring 2 cm × 2 cm were produced on the animals' backs. The experimental group was treated with NTP for 5 min/d for 4 weeks. The NTP was injected in a diffused manner into the cage housing the rats. The SD rats that had not received plasma treatment were designated as the control group. Blood was drawn on Postoperative Day 2, Day 4, and at 3 months. An immunohistochemical stain of E-cadherin and 4-hydroxy-2-nonenal (4-HNE), a reactive oxygen species marker, were evaluated and quantified for analysis using a CMYK color model. Results: A total of 35 SD rats were included in the study (25 in the NTP group and 10 in the control group). Low dose plasma treatment shortened the wound-healing time without damaging organs. In the NTP group, the white blood cell counts at Day 2 post-NTP treatment was not increased significantly more than that in the control group. After quantification of immunohistochemical staining, 4-HNE was increased at Day 14 compared with Day 7 (16.16 ± 12.81% vs. 55.11 ± 8.11%, p

Rationale: Diabetic retinopathy is characterized by vasopermeability, vascular leakage, inflammation, blood–retinal barrier breakdown, capillary degeneration, and neovascularization. However, the mechanisms underlying the association between diabetes mellitus and progression retinopathy remain unclear. Objective: TPL2 (tumor progression locus 2), a serine-threonine protein kinase, exerts a pathological effect on vascular angiogenesis. This study investigated the role of N ε -(carboxymethyl)lysine, a major advanced glycation end products, and the involved TPL2-related molecular signals in diabetic retinopathy using models of in vitro and in vivo and human samples. Methods and Results: Serum N ε -(carboxymethyl)lysine levels and TPL2 kinase activity were significantly increased in clinical patients and experimental animals with diabetic retinopathy. Intravitreal administration of pharmacological blocker or neutralizing antibody inhibited TPL2 and effectively suppressed the pathological characteristics of retinopathy in streptozotocin-induced diabetic animal models. Intravitreal VEGF (vascular endothelial growth factor) neutralization also suppressed the diabetic retinopathy in diabetic animal models. Mechanistic studies in primary human umbilical vein endothelial cells and primary retinal microvascular endothelial cells from streptozotocin-diabetic rats, db/db mice, and samples from patients with diabetic retinopathy revealed a positive parallel correlation between N ε -(carboxymethyl)lysine and the TPL2/chemokine SDF1α (stromal cell–derived factor-α) axis that is dependent on endoplasmic reticulum stress-related molecules, especially ATF4 (activating transcription factor-4). Conclusions: This study demonstrates that inhibiting the N ε -(carboxymethyl)lysine-induced TPL2/ATF4/SDF1α axis can effectively prevent diabetes mellitus–mediated retinal microvascular dysfunction. This signaling axis may include the therapeutic potential for other diseases involving pathological neovascularization or macular edema.

Based on a porcine model with surgically created myocardial infarction (MI) as a pre-clinical scheme, this study investigates the clinical translation of cell sheet fragments of autologous mesenchymal stem cells (MSCs) for cellular cardiomyoplasty. MSC sheet fragments retaining endogenous extracellular matrices are fabricated using a thermo-responsive methylcellulose hydrogel system. Echocardiographic observations indicate that transplantation of MSC sheet fragments in infarcted hearts can markedly attenuate the adverse ventricular dilation and preserve the cardiac function post MI, which is in contrast to the controlled groups receiving saline or dissociated MSCs. Additionally, histological analyses suggest that administering MSC sheet fragments significantly prevents the scar expansion and left ventricle remodeling after MI. Immunohistochemistry results demonstrate that the engrafted MSCs can differentiate into endothelial cells and smooth muscle cells, implying that angiogenesis and the subsequent regional perfusion improvement is a promising mechanism for ameliorating post-infarcted cardiac function. However, according to the data recorded by an implantable loop recorder, the transplanted MSCs may provoke arrhythmia. Nevertheless, the proposed approach may potentially lead to the eventual translation of MSC-based therapy into practical and effective clinical treatments.

Trypanosoma (subgenus Megatrypanum) theileri was first identified over one hundred years ago, and is a widespread parasite in cattle. Its life cycle within the mammalian host has rarely been reported. Whether there is an intracellular stage in tissues is unknown and such a stage has not been demonstrated experimentally. Intriguingly, using Giemsa staining with light microscopy and transmission electron microscopy examination, we found that the parasite was able not only to attach to cells but also to invade several phagocytic and non-phagocytic mammalian cells. Based on these findings, we conducted further investigations using a special antibody in immunofluorescence confocal images. Moreover, we examined a series of possible events of cell invasion in T. theileri. The results revealed that GM1, a marker of membrane rafts, was implicated in the mechanism of entry by this parasite. After incubation with tissue culture trypomastigotes, the gelatinolytic activity was significantly increased and accumulated at the attachment sites. Using ultrastructural localization detection by CytoTracker live imaging and confocal immunofluorescence microscopy, we found that lysosome fusion and the autophagy pathway were engaged in invaginating processes. T. theileri amastigotes also invaded cells and were enclosed by the lysosomes. Furthermore, tissue-cultured trypomastigotes were found to be capable of triggering intracellular free Ca2+ transients and TGF-β-signaling. Our findings that intracellular amastigote stages exist in mammalian cells infected with T. theileri and that the invasion processes involved various host cell components and cell signalings were extremely surprising and warrant further investigation.

Background This study aimed to investigate the relationship between oxidized regenerated cellulose (ORC) and mucosa healing in an experimental animal model. Methods Fifteen adult Sprague-Dawley rats were randomly divided into three groups that underwent different wound treatments. In Group 1, no pharyngeal wound was created. In Group 2, the pharyngeal wound was sutured with Prolene only. In Group 3, the pharyngeal wound was sutured with Prolene, and covered with one layer of ORC before closure of the skin wound. The animals were euthanized either 5 or 10 days after operation, and wound conditions were inspected and recorded. Specimens including sections of larynx and pharynx/upper esophagus were taken for microscopic and molecular biological examination. Results The pharyngotomy/esophagotomy wounds achieved good healing outcomes 10 days after operation. Wounds treated with ORC had significantly diminished inflammatory cell infiltration in microscopic examination when compared with that of those without ORC 5 days after operation. The matrix metalloproteinases (MMP) expression level was higher in wounds of Group 2 and Group 3, when compared with that of group 1. In addition, the MMP expression level was lower in the ORC-treated wounds when compared with that of those without ORC. There was no significant difference in fibroblast proliferation, collagen deposition, endothelin-1, alpha-smooth muscle actin, and transforming growth factor beta 1 expression level between wounds treated with ORC and those without ORC. Conclusion Reduced inflammatory response and decreased MMP expression level was observed in ORC-treated wounds. Whether ORC facilitates mucosa healing requires further investigation.

Rapid induction and creation of functional vascular networks is essential for the success of treating ischemic tissues. The formation of mature and functional vascular networks requires the cooperation of endothelial cells (ECs) and perivascular cells. In the study, we used a thermo-responsive hydrogel system to fabricate core-shell cell bodies composed of cord-blood mesenchymal stem cells (cbMSCs) and human umbilical vascular ECs (HUVECs) for functional vasculogenesis. When seeded on Matrigel, the shelled HUVECs attempted to interact and communicate vigorously with the cored cbMSCs initially. Subsequently, HUVECs migrated out and formed tubular structures; cbMSCs were observed to coalesce around the HUVEC-derived tubes. With time progressing, the tubular networks continued to expand without regression, indicating that cbMSCs might function as perivascular cells to stabilize the nascent networks. In the in vivo study, cbMSC/HUVEC bodies were embedded in Matrigel and implanted subcutaneously in nude mice. At day 7, visible blood-filled vessels were clearly identified within the implant containing cbMSC/HUVEC bodies, indicating that the formed vessels anastomosed with the host vasculature. The cored cbMSCs were stained positive for smooth muscle actin, suggesting that they underwent smooth muscle differentiation and formed microvessels with the shelled HUVECs, as the role of perivascular cells. These data confirm that the formation of mature vessels requires heterotypic cooperation of HUVECs and MSCs. This study provides a new strategy for therapeutic vasculogenesis, by showing the feasibility of using cbMSC/HUVEC bodies to create functional vascular networks.

Individual activation of nicotinic acetylcholine receptor (nAChR) or nitric oxide (NO) synthase in the dorsal facial area (DFA) increases blood flow of common carotid artery (CCA) supplying intra- and extra-cranial tissues. We investigated whether the activation of nAChR initiated the activation of NO synthase and guanylyl cyclase to increase CCA blood flow in anesthetized cats. Microinjections of nicotine (a non-selective nAChR agonist), or choline (a selective α7-nAChR agonist) in the DFA produced increases in CCA blood flow ipsilaterally. These increases were significantly reduced by pretreatment with NG-nitro-arginine methyl ester (l-NAME, a non-specific NO synthase inhibitor), 7-nitroindazole (7-NI, a relatively selective neuronal NO synthase inhibitor) or methylene blue (MB, a guanylyl cyclase inhibitor) but not by that with N5-(1-iminoethyl)-l-ornithine (l-NIO, a potent endothelial NO synthase inhibitor). Control microinjection with d-NAME (an isomer of l-NAME), artificial cerebrospinal fluid or DMSO (a solvent for 7-NI) did not affect resting CCA blood flow, nor did they affect nicotine- or choline-induced response. In conclusion, activation of nAChR, at least α7-nAChR, led to the activation of neuronal NO synthase and guanylyl cyclase in the DFA, which induced an increase in CCA blood flow.