The Curious Layperson: Fine-Grained Image Recognition Without Expert Labels.

Most of us are not experts in specific fields, such as ornithology. Nonetheless, we do have general image and language understanding capabilities that we use to match what we see to expert resources. This allows us to expand our knowledge and perform novel tasks without ad-hoc external supervision. On the contrary, machines have a much harder time consulting expert-curated knowledge bases unless trained specifically with that knowledge in mind. Thus, in this paper we consider a new problem: fine-grained image recognition without expert annotations, which we address by leveraging the vast knowledge available in web encyclopedias. First, we learn a model to describe the visual appearance of objects using non-expert image descriptions. We then train a fine-grained textual similarity model that matches image descriptions with documents on a sentence-level basis. We evaluate the method on two datasets (CUB-200 and Oxford-102 Flowers) and compare with several strong baselines and the state of the art in cross-modal retrieval. Code is available at: https://github.com/subhc/clever.

Nondestructive thermographic detection of internal defects using pixel-pattern based laser excitation and photothermal super resolution reconstruction.

In this work, we present a novel approach to photothermal super resolution based thermographic resolution of internal defects using two-dimensional pixel pattern-based active photothermal laser heating in conjunction with subsequent numerical reconstruction to achieve a high-resolution reconstruction of internal defect structures. With the proposed adoption of pixelated patterns generated using laser coupled high-power DLP projector technology the complexity for achieving true two-dimensional super resolution can be dramatically reduced taking a crucial step forward towards widespread practical viability. Furthermore, based on the latest developments in high-power DLP projectors, we present their first application for structured pulsed thermographic inspection of macroscopic metal samples. In addition, a forward solution to the underlying inverse problem is proposed along with an appropriate heuristic to find the regularization parameters necessary for the numerical inversion in a laboratory setting. This allows the generation of synthetic measurement data, opening the door for the application of machine learning based methods for future improvements towards full automation of the method. Finally, the proposed method is experimentally validated and shown to outperform several established conventional thermographic testing techniques while conservatively improving the required measurement times by a factor of 8 compared to currently available photothermal super resolution techniques.

Unsupervised Learning of Probably Symmetric Deformable 3D Objects From Images in the Wild (Invited Paper).

We propose a method to learn 3D deformable object categories from raw single-view images, without external supervision. The method is based on an autoencoder that factors each input image into depth, albedo, viewpoint and illumination. In order to disentangle these components without supervision, we use the fact that many object categories have, at least approximately, a symmetric structure. We show that reasoning about illumination allows us to exploit the underlying object symmetry even if the appearance is not symmetric due to shading. Furthermore, we model objects that are probably, but not certainly, symmetric by predicting a symmetry probability map, learned end-to-end with the other components of the model. Our experiments show that this method can recover very accurately the 3D shape of human faces, cat faces and cars from single-view images, without any supervision or a prior shape model. On benchmarks, we demonstrate superior accuracy compared to another method that uses supervision at the level of 2D image correspondences.

Neurosteroids and translocator protein 18 kDa (TSPO) in depression: implications for synaptic plasticity, cognition, and treatment options.

There is need for novel fast acting treatment options in affective disorders. 3α-reduced neurosteroids such as allopregnanolone are powerful positive allosteric modulators of GABAA receptors and target also extrasynaptic receptors. Their synthesis is mediated by the translocator protein 18 kDa (TSPO). TSPO ligands not only promote endogenous neurosteroidogenesis, but also exert a broad spectrum of functions involving modulation of mitochondrial activity and acting as anti-inflammatory and neuroregenerative agents. Besides affective symptoms, in depression cognitive impairment can be frequently observed, which may be ameliorated through targeting of extrasynaptic GABAA receptors either via TSPO ligands or exogenously administered 3α-reduced neurosteroids. Interestingly, recent findings indicate an enhanced activation of the complement system, e.g., enhanced expression of C1q, both in depression and dementia. It is of note that benzodiazepines have been shown to reduce long-term potentiation and to cause cognitive decline. Intriguingly, TSPO may be crucial in mediating the effects of benzodiazepines on synaptic pruning. Here, we discuss how benzodiazepines and TSPO may interfere with synaptic pruning. Moreover, we highlight recent developments of TSPO ligands and 3α-reduced neurosteroids as therapeutic agents. Etifoxine is the only clinically available TSPO ligand so far and has been studied in anxiety disorders. Regarding 3α-reduced neurosteroids, brexanolone, an intravenous formulation of allopregnanolone, has been approved for the treatment of postpartum depression and zuranolone, an orally available 3α-reduced neurosteroid, is currently being studied in major depressive disorder and postpartum depression. As such, 3α-reduced neurosteroids and TSPO ligands may constitute promising treatment approaches for affective disorders.

Morphological Representation of C1q in the Aging Central Nervous System.

INTRODUCTION: The complement protein C1q is essential for the innate immune system and neurophysiological and neuropathological processes. To gain more insight into these functions in the CNS, a comprehensive understanding of the morphological representation, especially of its cellular and subcellular target structures, is of great importance. METHODS: For a free-floating preparation, the brains of wild-type and ArcAß mice were cut into 100 µm slices. Living slices were incubated in Ringer's solution and then fixed in 4% paraformaldehyde (PFA) and stained with different primary and secondary antibodies or methoxy-X04. RESULTS: C1q was abundant in the entire brain. Interestingly, C1q accumulated around cell nuclei, with a perineuronal localization around neuronal somata and a paraneuronal accumulation around non-neuronal cells, e. g., microglia. Moreover, dendritic-like, linear, branched C1q signals were observed in the area between the dentate gyrus and the CA1 region of the hippocampus. Complementary staining revealed an overlap with ß-amyloid accumulation reflected by the deposition of C1q within plaques and modified basal C1q levels in the brains of transgenic ArcAß animals. DISCUSSION: The applied free-floating approach is suitable for C1q immunofluorescence imaging. The consistent colocalization of the complement protein C1q with ß-amyloid plaques may reflect an activated immune response, whereas the accumulation of C1q around neuronal structures such as somata and dendrites is still a matter of debate. Intriguingly, C1q surrounds those structures in older brains of both wild-type and ArcAß mice. Our results also indicate an involvement of C1q in neurophysiological and neurodegenerative processes.

Neuroinflammation and psychiatric disorders: Relevance of C1q, translocator protein (18 kDa) (TSPO), and neurosteroids.

There is increasing evidence that neuroinflammatory processes may play a role in the pathophysiology of psychiatric disorders. Recently, the complement protein C1q and the translocator protein (18 kDa) (TSPO) have attracted considerable interest in this context. C1q is a small molecule which is involved into synaptic pruning mechanisms, increases during ageing and may contribute to neurodegenerative disorders. TSPO is a transmembrane channel protein and mediates numerous biological functions such as bioenergetics and steroid synthesis. Meanwhile, there is evidence that both C1q and TSPO may be elevated in psychiatric disorders, e.g. major depression. Moreover, preclinical and first clinical studies suggest that TSPO ligands may exert antidepressant and anxiolytic properties by promoting endogenous neurosteroid synthesis. In addition, certain neurosteroids, e.g. allopregnanolone, are potent positive allosteric modulators of GABAA receptors and their composition is altered in depression and anxiety disorders. Recently, neurosteroid compounds such as brexanolone or zuranolone have been reported to reduce depressive and anxiety symptoms in postpartum depression and major depressive disorder. In conclusion, compounds enhancing GABAergic neurotransmission such as neurosteroids and TSPO ligands, which also may exert anti-inflammatory properties in concert with immunomodulators such as C1q may open new avenues for the treatment of psychiatric disorders.

Spread of Multidrug-Resistant Bacteria by Moth Flies from Hospital Waste Water System.

We documented and analyzed moth fly occurrence and spread of multidrug-resistant bacteria in a tertiary care hospital in Germany. The moth flies (Clogmia albipunctata) bred in the sewage system, then moved into the hospital, carrying biofilm and multidrug-resistant bacteria on their feet. Subsequently, the hospital developed a pest control protocol.

Classification of Polar Maps from Cardiac Perfusion Imaging with Graph-Convolutional Neural Networks.

Myocardial perfusion imaging is a non-invasive imaging technique commonly used for the diagnosis of Coronary Artery Disease and is based on the injection of radiopharmaceutical tracers into the blood stream. The patient's heart is imaged while at rest and under stress in order to determine its capacity to react to the imposed challenge. Assessment of imaging data is commonly performed by visual inspection of polar maps showing the tracer uptake in a compact, two-dimensional representation of the left ventricle. This article presents a method for automatic classification of polar maps based on graph convolutional neural networks. Furthermore, it evaluates how well localization techniques developed for standard convolutional neural networks can be used for the localization of pathological segments with respect to clinically relevant areas. The method is evaluated using 946 labeled datasets and compared quantitatively to three other neural-network-based methods. The proposed model achieves an agreement with the human observer on 89.3% of rest test polar maps and on 91.1% of stress test polar maps. Localization performed on a fine 17-segment division of the polar maps achieves an agreement of 83.1% with the human observer, while localization on a coarse 3-segment division based on the vessel beds of the left ventricle has an agreement of 78.8% with the human observer. Our method could thus assist the decision-making process of physicians when analyzing polar map data obtained from myocardial perfusion images.

A novel FbFP-based biosensor toolbox for sensitive in vivo determination of intracellular pH.

The intracellular pH is an important modulator of various bio(techno)logical processes such as enzymatic conversion of metabolites or transport across the cell membrane. Changes of intracellular pH due to altered proton distribution can thus cause dysfunction of cellular processes. Consequently, accurate monitoring of intracellular pH allows elucidating the pH-dependency of (patho)physiological and biotechnological processes. In this context, genetically encoded biosensors represent a powerful tool to determine intracellular pH values non-invasively and with high spatiotemporal resolution. We have constructed a toolbox of novel genetically encoded FRET-based pH biosensors (named Fluorescence Biosensors for pH or FluBpH) that utilizes the FMN-binding fluorescent protein EcFbFP as donor domain. In contrast to many fluorescent proteins of the GFP family, EcFbFP exhibits a remarkable tolerance towards acidic pH (pKa∼3.2). To cover the broad range of physiologically relevant pH values, three EYFP variants exhibiting pKa values of 5.7, 6.1 and 7.5 were used as pH-sensing FRET acceptor domains. The resulting biosensors FluBpH 5.7, FluBpH 6.1 and FluBpH 7.5 were calibrated in vitro and in vivo to accurately evaluate their pH indicator properties. To demonstrate the in vivo applicability of FluBpH, changes of intracellular pH were ratiometrically measured in E. coli cells during acid stress.

Automatic Segmentation of Kidneys using Deep Learning for Total Kidney Volume Quantification in Autosomal Dominant Polycystic Kidney Disease.

Autosomal Dominant Polycystic Kidney Disease (ADPKD) is the most common inherited disorder of the kidneys. It is characterized by enlargement of the kidneys caused by progressive development of renal cysts, and thus assessment of total kidney volume (TKV) is crucial for studying disease progression in ADPKD. However, automatic segmentation of polycystic kidneys is a challenging task due to severe alteration in the morphology caused by non-uniform cyst formation and presence of adjacent liver cysts. In this study, an automated segmentation method based on deep learning has been proposed for TKV computation on computed tomography (CT) dataset of ADPKD patients exhibiting mild to moderate or severe renal insufficiency. The proposed method has been trained (n = 165) and tested (n = 79) on a wide range of TKV (321.2-14,670.7 mL) achieving an overall mean Dice Similarity Coefficient of 0.86 ± 0.07 (mean ± SD) between automated and manual segmentations from clinical experts and a mean correlation coefficient (ρ) of 0.98 (p < 0.001) for segmented kidney volume measurements in the entire test set. Our method facilitates fast and reproducible measurements of kidney volumes in agreement with manual segmentations from clinical experts.

Leukotriene receptor blockade as a life-saving treatment in severe bronchopulmonary dysplasia.

BACKGROUND: Bronchopulmonary dysplasia (BPD) is a major cause of mortality and morbidity in infants with an extremely low birth weight. Because there is no effective therapy, the mortality of this condition in severely affected patients is high. Therapeutic blocking of the leukotriene system seems to be a logical approach due to the known pathophysiology of BPD. OBJECTIVES: The aim of this study was to examine the therapeutic effect of montelukast in preterm children suffering from severe BPD. METHODS: We performed an unblinded, prospective trial including infants born between 23 and 27 weeks of gestation suffering from severe BPD. The study drug was montelukast (1 mg/kg of body weight as a single dose daily in the 1st week of therapy, increasing to 1.5 mg/kg of body weight in the 2nd week and finally to 2 mg/kg of body weight in the 3rd week). Treatment was continued until the radiological signs and the clinical symptoms of BPD disappeared or the patient was discharged from the hospital. Each patient included in this study was matched for gestational age, birth weight, and pulmonary severity score to a control. RESULTS: Until March 2014, a total of 22 infants were enrolled into the study. The rates of the primary outcome differed significantly between the montelukast-treated group and the control group. All but 1 of the children in the treatment group survived (91%), whereas 7 of the 11 children in the control group died (survival rate 36%; p = 0.002 using Fisher's exact test). The mean mechanical ventilation time (41.2 ± 25.3 vs. 103.7 ± 90.6 days) was significantly shorter and the mean preterm complication score (3.0 ± 1.7 vs. 5.6 ± 1.4) was significantly lower in treated patients compared to the control group. (p = 0.05 for both items; Wilcoxon's matched-pairs test). CONCLUSION: Based on the clinical observations, the statistical results, and the relatively low risk of the study drug montelukast, we recommend using this treatment in severe cases of BPD for infants facing a high risk of death.

The photophysics of LOV-based fluorescent proteins--new tools for cell biology.

LOV-based fluorescent proteins (FPs) are an alternative class of fluorescent reporters with unique properties which complement the well-established proteins of the GFP family. One of the most important features of LOV-based FPs is the independence of molecular oxygen for the development of their specific fluorescence. Furthermore, they are characterized by small size and rapid signal development. Over the last few years, a number of different bacterial and plant LOV-based fluorescent proteins such as FbFP, iLOV and miniSOG have been developed and optimized. In this report, we comparatively have characterized the photophysical properties of nine different LOV-based fluorescent proteins including the excitation and emission maxima, the extinction coefficient, the fluorescence quantum yield, the average fluorescence lifetime and the photostability. The unified characterization of the LOV-based FPs provides a useful guide to apply them as in vivo tools for quantitative analyses and biological imaging.