Controllable Image Synthesis With Attribute-Decomposed GAN.

This paper proposes Attribute-Decomposed GAN (ADGAN) and its enhanced version (ADGAN++) for controllable image synthesis, which can produce realistic images with desired attributes provided in various source inputs. The core ideas of the proposed ADGAN and ADGAN++ are both to embed component attributes into the latent space as independent codes and thus achieve flexible and continuous control of attributes via mixing and interpolation operations in explicit style representations. The major difference between them is that ADGAN processes all component attributes simultaneously while ADGAN++ utilizes a serial encoding strategy. More specifically, ADGAN consists of two encoding pathways with style block connections and is capable of decomposing the original hard mapping into multiple more accessible subtasks. In the source pathway, component layouts are extracted via a semantic parser and the segmented components are fed into a shared global texture encoder to obtain decomposed latent codes. This strategy allows for the synthesis of more realistic output images and the automatic separation of un-annotated component attributes. Although the original ADGAN works in a delicate and efficient manner, intrinsically it fails to handle the semantic image synthesizing task when the number of attribute categories is huge. To address this problem, ADGAN++ employs the serial encoding of different component attributes to synthesize each part of the target real-world image, and adopts several residual blocks with segmentation guided instance normalization to assemble the synthesized component images and refine the original synthesis result. The two-stage ADGAN++ is designed to alleviate the massive computational costs required when synthesizing real-world images with numerous attributes while maintaining the disentanglement of different attributes to enable flexible control of arbitrary component attributes of the synthesized images. Experimental results demonstrate the proposed methods' superiority over the state of the art in pose transfer, face style transfer, and semantic image synthesis, as well as their effectiveness in the task of component attribute transfer. Our code and data are publicly available at https://github.com/menyifang/ADGAN.

High mobility group box-1 serves a pathogenic role in spinal cord injury via the promotion of pro-inflammatory cytokines.

Traumatic spinal cord injury (SCI) is a devastating condition marked by permanent motor, sensory, and autonomic dysfunction, in which the inflammatory response serves an important and preventable role. High mobility group box-1 (HMGB1) is a potent regulator of inflammation in numerous acute and chronic inflammatory conditions.; however, the role of HMGB1 in SCI remains unclear. The present study aimed to characterize the temporal dynamics of HMGB1 release after SCI, to investigate the role of spinal microglia activation in mediating the effects of HMGB1 on SCI, and to explore the therapeutic potential of intrathecal anti-HMGB1 polyclonal antibody on alleviating SCI. The present study demonstrated that HMGB1 expression was increased immediately after traumatic injury of a primary spinal neuron culture. It was found that neutralizing HMGB1 significantly ameliorated SCI pathogenesis and hind limb paralysis. Moreover, the levels of a number of pro-inflammatory cytokines in the SCI lesion were reduced when local HMGB1 was blocked by anti-HMGB1 antibody. In addition, the injured neuron-derived conditioned medium increased TNF-α secretion and the NF-κB pathway in the BV2 microglia cell line via HMGB1. Collectively, these results indicated that HMGB1 served an important role in SCI inflammation and suggested the therapeutic potential of an anti-HMGB1 antibody for SCI.

Albendazole and Corticosteroids for the Treatment of Solitary Cysticercus Granuloma: A Network Meta-analysis.

BACKGROUND: Solitary cysticercus granuloma (SCG) is the commonest form of neurocysticercosis in the Indian subcontinent and in travelers. Several different treatment options exist for SCG. We conducted a Bayesian network meta-analysis of randomized clinical trials (RCTs) to identify the best treatment option to prevent seizure recurrence and promote lesion resolution for patients with SCG. METHODS AND PRINCIPAL FINDINGS: PubMed, EMBASE and the Cochrane Library databases (up to June 1, 2015) were searched for RCTs that compared any anthelmintics or corticosteroids, alone or in combination, with placebo or head to head and reported on seizure recurrence and lesion resolution in patients with SCG. A total of 14 RCTs (1277 patients) were included in the quantitative analysis focusing on four different treatment options. A Bayesian network model computing odds ratios (OR) with 95% credible intervals (CrI) and probability of being best (Pbest) was used to compare all interventions simultaneously. Albendazole and corticosteroids combination therapy was the only regimen that significantly decreased the risk of seizure recurrence compared with conservative treatment (OR 0.32, 95% CrI 0.10-0.93, Pbest 73.3%). Albendazole and corticosteroids alone or in combination were all efficacious in hastening granuloma resolution, but the combined therapy remained the best option based on probability analysis (OR 3.05, 95% CrI 1.24-7.95, Pbest 53.9%). The superiority of the combination therapy changed little in RCTs with different follow-up durations and in sensitivity analyses. The limitations of this study include high risk of bias and short follow-up duration in most studies. CONCLUSIONS: Dual therapy of albendazole and corticosteroids was the most efficacious regimen that could prevent seizure recurrence and promote lesion resolution in a follow-up period of around one year. It should be recommended for the management of SCG until more high-quality evidence is available.

Pneumothorax as the initial manifestation of idiopathic hypereosinophilic syndrome.

We report a case of hypereosinophilic syndrome in a 47-year-old man who had acute pneumothorax as the initial presentation. Peripheral blood eosinophil count increased continuously over a period of 1 month and was associated with pulmonary changes and appearance of skin lesions on the right chest wall. Idiopathic hypereosinophilic syndrome was confirmed by bone marrow aspiration biopsy and skin lesion biopsy after exclusion of all possible secondary etiologies. The clinical status and chest radiographs showed marked improvement after treatment with corticosteroids.

Flickr distance: a relationship measure for visual concepts.

This paper proposes the Flickr Distance (FD) to measure the visual correlation between concepts. For each concept, a collection of related images are obtained from the Flickr website. We assume that each concept consists of several states, e.g., different views, different semantics, etc., which are considered as latent topics. Then a latent topic visual language model (LTVLM) is built to capture these states. The Flickr distance between two concepts is defined as the Jensen-Shannon (J-S) divergence between their LTVLM. Differently from traditional conceptual distance measurements, which are based on Web textual documents, FD is based on the visual information. Comparing with the WordNet distance, FD can easily scale up with the increasing size of the conceptual corpus. Comparing with the Google Distance (NGD) and Tag Concurrence Distance (TCD), FD uses the visual information and can properly measure the conceptual relations. We apply FD to multimedia-related tasks and find methods based on FD significantly outperform those based on NGD and TCD. With the FD measurement, we also construct a large-scale visual conceptual network (VCNet) to store the knowledge of conceptual relationship. Experiments show that FD is more coherent to human cognition and it also outperforms text-based distances in real-world applications.

High-mobility group box-1 and its receptors contribute to proinflammatory response in the acute phase of spinal cord injury in rats.

STUDY DESIGN: To examine the localization and expression of high-mobility group box-1 (HMGB-1) protein and its receptors after rat spinal cord injury. OBJECTIVE: To elucidate the contribution of HMGB-1 and its receptors as potential candidates in a specific upstream pathway to the proinflammatory response leading to a cascade of secondary tissue damage after spinal cord injury. SUMMARY OF BACKGROUND DATA: HMGB-1 was recently characterized as a key cytokine with a potential role in nucleosome formation and regulation of gene transcription. No studies have investigated the role of HMGB-1 in spinal cord injury. METHODS: Injured thoracic spinal cord from 62 rats aged 8 to 12 weeks and spinal cord from 20 control rats were examined. HMGB-1 was localized by immunofluorescence staining, costaining with cell markers, and by immunoelectron microscopy. The expression of HMGB-1 and its receptors, receptor for advanced glycation end products (RAGE), toll-like receptor (TLR)2, and TLR4 were also examined by immunohistochemistry. RESULTS: HMGB-1 expression appeared earlier than that of tumor necrosis factor-α, interleukin (IL)-1ß, and IL-6 in the spinal cord injury rats, with the HMGB-1 produced by both macrophages and neurons. HMGB-1 translocated from nucleus to cytoplasm in some neurons at an early stage after neural injury. Increased expression of HMGB-1, RAGE, and TLRs was observed after injury, and interaction of HMGB-1 with RAGE or TLRs, particularly in macrophage, was confirmed at 3 days after injury. CONCLUSION: Our results demonstrated an earlier onset in the expression of HMGB-1 than in tumor necrosis factor-α, IL-1ß, and IL-6 after spinal cord injury. The release of HMGB-1 from neurons and macrophages is mediated through the HMGB-1/RAGE or TLR pathways. HMGB-1 seems to play at least some roles in the proinflammatory cascade originating the secondary damage after the initial spinal cord injury.

Tumor necrosis factor-α antagonist reduces apoptosis of neurons and oligodendroglia in rat spinal cord injury.

STUDY DESIGN: To examine the effects of a tumor necrosis factor (TNF)-α antagonist (etanercept) on rat spinal cord injury and identify a possible mechanism for its action. OBJECTIVE: To elucidate the contribution of etanercept to the pathologic cascade in spinal cord injury and its possible suppression of neuronal and oligodendroglial apoptosis. SUMMARY OF BACKGROUND DATA: Etanercept has been recently used successfully for treatment of inflammatory disorders. However, only a few studies have examined its role in suppressing neuronal and oligodendroglial apoptosis in spinal cord injury. METHODS: Etanercept or saline (control) was administered by intraperitoneal injection 1 hour after thoracic spinal cord injury in rats. The expressions and localizations of TNF-α, TNF receptor 1 (TNFR1), and TNF receptor 2 (TNFR2) were examined by immunoblot and immunohistochemical analyses. Spinal cord tissue damage between saline- and etanercept-treated groups was also compared after hematoxylin-eosin and luxol fast blue (LFB) staining. The Basso-Beattie-Bresnahan (BBB) scale was used to evaluate rat locomotor function after etanercept administration. Terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick end labeling (TUNEL)-positive cells were counted and the immunoreactivity to active caspase-3 and caspase-8 was examined after etanercept administration. RESULTS: Immunoblot and double immunofluorescence staining revealed suppression of TNF-α, TNFR1, and TNFR2 expression after administration of etanercept in the acute phase of spinal cord injury. LFB staining demonstrated potential myelination in the etanercept-treated group from 2 week after spinal cord injury, together with an increased BBB locomotor score. Double immunofluorescence staining showed a significant decrease in TUNEL-positive neurons and oligodendroglia from 12 hour to 1 week in the gray and white matters after etanercept administration. Immunoblot analysis demonstrated overexpression of activated caspase-3 and caspase-8 after spinal cord injury, which was markedly inhibited by etanercept. CONCLUSION: Our results indicated that etanercept reduces the associated tissue damage of spinal cord injury, improves hindlimb locomotor function, and facilitates myelin regeneration. This positive effect of etanercept on spinal cord injury is probably attributable to the suppression of TNF-α, TNFR1, TNFR2, and activated caspase-3 and caspase-8 overexpressions, and the inhibition of neuronal and oligodendroglial apoptosis.

A synthetic cell-penetrating peptide antagonizing TrkA function suppresses neuropathic pain in mice.

Nerve growth factor (NGF) and its high-affinity receptor, TrkA, are one of the targets in the production of new drugs for the treatment of neuropathic pain. NGF contributes to both the initiation and maintenance of sensory abnormalities after peripheral nerve injury. This study examined the effects of IPTRK3, a new synthetic cell-penetrating peptide that antagonizes TrkA function, on neuropathic pain in mice. Partial sciatic nerve ligation (PSNL) was used to generate neuropathic pain, and we injected IPTRK3 (2 or 10 mg/kg) intraperitoneally on day 7 after PSNL. Effects of the peptide on hyperalgesia, allodynia, and expression of Fos in the spinal cord were examined. Single administration of the peptide on day 7 significantly suppressed both thermal hyperalgesia and mechanical allodynia. Gentle touch stimuli-evoked Fos expression in the lumbar spinal cord was also significantly reduced. Intraperitoneal injection of a cell-penetrating peptide antagonizing TrkA function appears effective for treatment of neuropathic pain in a mouse pain model.

Annotating images by mining image search results.

Although it has been studied for years by the computer vision and machine learning communities, image annotation is still far from practical. In this paper, we propose a novel attempt at model-free image annotation, which is a data-driven approach that annotates images by mining their search results. Some 2.4 million images with their surrounding text are collected from a few photo forums to support this approach. The entire process is formulated in a divide-and-conquer framework where a query keyword is provided along with the uncaptioned image to improve both the effectiveness and efficiency. This is helpful when the collected data set is not dense everywhere. In this sense, our approach contains three steps: 1) the search process to discover visually and semantically similar search results, 2) the mining process to identify salient terms from textual descriptions of the search results, and 3) the annotation rejection process to filter out noisy terms yielded by Step 2. To ensure real-time annotation, two key techniques are leveraged-one is to map the high-dimensional image visual features into hash codes, the other is to implement it as a distributed system, of which the search and mining processes are provided as Web services. As a typical result, the entire process finishes in less than 1 second. Since no training data set is required, our approach enables annotating with unlimited vocabulary and is highly scalable and robust to outliers. Experimental results on both real Web images and a benchmark image data set show the effectiveness and efficiency of the proposed algorithm. It is also worth noting that, although the entire approach is illustrated within the divide-and conquer framework, a query keyword is not crucial to our current implementation. We provide experimental results to prove this.

Learning similarity measure for natural image retrieval with relevance feedback.

A new scheme of learning similarity measure is proposed for content-based image retrieval (CBIR). It learns a boundary that separates the images in the database into two clusters. Images inside the boundary are ranked by their Euclidean distances to the query. The scheme is called constrained similarity measure (CSM), which not only takes into consideration the perceptual similarity between images, but also significantly improves the retrieval performance of the Euclidean distance measure. Two techniques, support vector machine (SVM) and AdaBoost from machine learning, are utilized to learn the boundary. They are compared to see their differences in boundary learning. The positive and negative examples used to learn the boundary are provided by the user with relevance feedback. The CSM metric is evaluated in a large database of 10009 natural images with an accurate ground truth. Experimental results demonstrate the usefulness and effectiveness of the proposed similarity measure for image retrieval.