Monocular Depth Estimation from a Fisheye Camera Based on Knowledge Distillation.

Monocular depth estimation is a task aimed at predicting pixel-level distances from a single RGB image. This task holds significance in various applications including autonomous driving and robotics. In particular, the recognition of surrounding environments is important to avoid collisions during autonomous parking. Fisheye cameras are adequate to acquire visual information from a wide field of view, reducing blind spots and preventing potential collisions. While there have been increasing demands for fisheye cameras in visual-recognition systems, existing research on depth estimation has primarily focused on pinhole camera images. Moreover, depth estimation from fisheye images poses additional challenges due to strong distortion and the lack of public datasets. In this work, we propose a novel underground parking lot dataset called JBNU-Depth360, which consists of fisheye camera images and their corresponding LiDAR projections. Our proposed dataset was composed of 4221 pairs of fisheye images and their corresponding LiDAR point clouds, which were obtained from six driving sequences. Furthermore, we employed a knowledge-distillation technique to improve the performance of the state-of-the-art depth-estimation models. The teacher-student learning framework allows the neural network to leverage the information in dense depth predictions and sparse LiDAR projections. Experiments were conducted on the KITTI-360 and JBNU-Depth360 datasets for analyzing the performance of existing depth-estimation models on fisheye camera images. By utilizing the self-distillation technique, the AbsRel and SILog error metrics were reduced by 1.81% and 1.55% on the JBNU-Depth360 dataset. The experimental results demonstrated that the self-distillation technique is beneficial to improve the performance of depth-estimation models.

Metal-Organic Framework Thin Films as Ideal Matrices for Azide Photolysis in Vacuum.

Studies on reactions in solutions are often hampered by solvent effects. In addition, detailed investigation on kinetics is limited to the small temperature regime where the solvent is liquid. Here, we report the in situ spectroscopic observation of UV-induced photochemical reactions of aryl azides within a crystalline matrix in vacuum. The matrices are formed by attaching the reactive moieties to ditopic linkers, which are then assembled to yield metal-organic frameworks (MOFs) and surface-mounted MOFs (SURMOFs). These porous, crystalline frameworks are then used as model systems to study azide-related chemical processes under ultrahigh vacuum (UHV) conditions, where solvent effects can be safely excluded and in a large temperature regime. Infrared reflection absorption spectroscopy (IRRAS) allowed us to monitor the photoreaction of azide in SURMOFs precisely. The in situ IRRAS data, in conjunction with XRD, MS, and XPS, reveal that illumination with UV light first leads to forming a nitrene intermediate. In the second step, an intramolecular rearrangement occurs, yielding an indoloindole derivative. These findings unveil a novel pathway for precisely studying azide-related chemical transformations. Reference experiments carried out for solvent-loaded SURMOFs reveal a huge diversity of other reaction schemes, thus highlighting the need for model systems studied under UHV conditions.

The Interplay between Structure and Product Selectivity of CO2 Hydrogenation.

Identification of the active structure under reaction conditions is of great importance for the rational design of heterogeneous catalysts. However, this is often hampered by their structural complexity. The interplay between the surface structure of Co3 O4 and the CO2 hydrogenation is described. Co3 O4 with morphology-dependent crystallographic surfaces presents different reducibility and formation energy of oxygen vacancies, thus resulting in distinct steady-state composition and product selectivity. Co3 O4 -0 h rhombic dodecahedra were completely reduced to Co0 and CoO, which presents circa 85 % CH4 selectivity. In contrast, Co3 O4 -2 h nanorods were partially reduced to CoO, which exhibits a circa 95 % CO selectivity. The crucial role of the Co3 O4 structure in determining the catalytic performance for higher alcohol synthesis over CuCo-based catalysts is demonstrated. As expected, Cu/Co3 O4 -2 h shows nine-fold higher ethanol yield than Cu/Co3 O4 -0 h owing to the inhibition for methanation.

A comprehensive analysis of piRNAs from adult human testis and their relationship with genes and mobile elements.

BACKGROUND: Piwi-interacting RNAs (piRNAs) are a recently discovered class of small non-coding RNAs whose best-understood function is to repress mobile element (ME) activity in animal germline. To date, nearly all piRNA studies have been conducted in model organisms and little is known about piRNA diversity, target specificity and biological function in human. RESULTS: Here we performed high-throughput sequencing of piRNAs from three human adult testis samples. We found that more than 81% of the ~17 million putative piRNAs mapped to ~6,000 piRNA-producing genomic clusters using a relaxed definition of clusters. A set of human protein-coding genes produces a relatively large amount of putative piRNAs from their 3'UTRs, and are significantly enriched for certain biological processes, suggestive of non-random sampling by the piRNA biogenesis machinery. Up to 16% of putative piRNAs mapped to a few hundred annotated long non-coding RNA (lncRNA) genes, suggesting that some lncRNA genes can act as piRNA precursors. Among major ME families, young families of LTR and endogenous retroviruses have a greater association with putative piRNAs than other MEs. In addition, piRNAs preferentially mapped to specific regions in the consensus sequences of several ME (sub)families and some piRNA mapping peaks showed patterns consistent with the "ping-pong" cycle of piRNA targeting and amplification. CONCLUSIONS: Overall our data provide a comprehensive analysis and improved annotation of human piRNAs in adult human testes and shed new light into the relationship of piRNAs with protein-coding genes, lncRNAs, and mobile genetic elements in human.

From hub proteins to hub modules: the relationship between essentiality and centrality in the yeast interactome at different scales of organization.

Numerous studies have suggested that hub proteins in the S. cerevisiae physical interaction network are more likely to be essential than other proteins. The proposed reasons underlying this observed relationship between topology and functioning have been subject to some controversy, with recent work suggesting that it arises due to the participation of hub proteins in essential complexes and processes. However, do these essential modules themselves have distinct network characteristics, and how do their essential proteins differ in their topological properties from their non-essential proteins? We aimed to advance our understanding of protein essentiality by analyzing proteins, complexes and processes within their broader functional context and by considering physical interactions both within and across complexes and biological processes. In agreement with the view that essentiality is a modular property, we found that the number of intracomplex or intraprocess interactions that a protein has is a better indicator of its essentiality than its overall number of interactions. Moreover, we found that within an essential complex, its essential proteins have on average more interactions, especially intracomplex interactions, than its non-essential proteins. Finally, we built a module-level interaction network and found that essential complexes and processes tend to have higher interaction degrees in this network than non-essential complexes and processes; that is, they exhibit a larger amount of functional cross-talk than their non-essential counterparts.

All-solution-processed nonvolatile flexible nano-floating gate memory devices.

Organic semiconductors have great potential for future electronic applications owing to their inherent flexibility, low cost, light weight and ability to easily cover large areas. However, all of these advantageous material properties can only be harnessed if simple, cheap and low-temperature fabrication processes, which exclude the need for vacuum deposition and are compatible with flexible plastic substrates, are employed. There are a few solution-based techniques such as spin-coating and inkjet printing that meet the above criteria. In this paper, we describe a novel all-solution-processed nonvolatile memory device fabricated on a flexible plastic substrate. The source, drain and gate electrodes were printed using an inkjet printer with a conducting organic solution, while the semiconducting layer was spin-coated with an n-type polymer. The charge-trapping layer was composed of spin-coated reduced graphene oxide (rGO), which was prepared in the form of a solution using Hummer's method. The fabricated device was characterized in order to confirm the memory characteristics. Device parameters such as threshold voltage shift, retention/endurance characteristics, mechanical robustness and reliability upon bending were also analyzed.

Knowledge distillation of multi-scale dense prediction transformer for self-supervised depth estimation.

Depth estimation is an inverse projection problem that estimates pixel-level distances from a single image. Although, supervised methods have shown promising results, it has intrinsic limitations in requiring ground truth depth from an external sensor. On the other hand, self-supervised depth estimation relieves the burden for collecting calibrated training data, while there is still a large performance gap between supervised and self-supervised methods. The objective of this study is to reduce the performance gap between the supervised and self-supervised approaches. The loss function of previous self-supervised methods is mainly based on a photometric error, which is indirectly computed from synthesized images using depth and pose estimates. In this paper, we argue that direct depth cue is more effective to train a depth estimation network. To obtain the direct depth cue, we employed a knowledge distillation technique, which is a teacher-student learning framework. The teacher network was trained in a self-supervised manner based on a photometric error, and its predictions were utilized to train a student network. We constructed a multi-scale dense prediction transformer with Monte Carlo dropout, and multi-scale distillation loss was proposed to train the student network based on the ensemble of stochastic estimates. Experiments were conducted on the KITTI and Make3D datasets, and our proposed method achieved the state-of-the-art accuracy in self-supervised depth estimation. Our code is publicly available at https://github.com/ji-min-song/KD-of-MS-DPT .

A comparison of the safety and efficacy of minimally invasive surgery versus open surgery in treating cauda equina syndrome: A systematic review and meta-analysis.

PURPOSE: To compare the safety and efficacy of minimally invasive surgery (MIS) and open surgery (OS) in treating cauda equina syndrome (CES). METHODS: A systematic literature search was conducted, searching relevant databases for studies investigating MIS and/or OS in treating CES. Pooled outcomes and their 95% confidence intervals (CIs) were meta-analyzed via random-effects models. RESULTS: Ten studies were included in the meta-analysis. Pooled mean operation times were shorter for MIS (75.4 min; 95 %CI: 40.8, 110.0) than OS (155.1 min; 121.3, 188.9). Similarly, mean hospital stay was shorter for MIS (4.08 days; 2.77, 5.39 vs. 8.85 days; 6.56, 11.13). Mean blood loss was smaller for MIS (71.7 mL; 0, 154.5 vs. 366.5; 119.1, 614.0). Mean post-op lumbar/back visual analogue scale (VAS) score was lower for MIS (3.65; 2.75, 4.56 vs. 5.80; 4.55, 7.05). Mean post-op leg VAS score was 1.27 (0.41, 21.4) for MIS and 1.29 (0.47, 2.12) for OS. Mean complete bladder recovery rate was 81.0% (55.0%, 94.0%) for MIS and 75.0% (44.0%, 92.0%) for OS. Mean complete motor recovery rate was larger for MIS (70.0%; 48.0, 85.0 vs. 42.0%; 34.0, 51.0). Mean percentages of "excellent" patient outcomes were equal for MIS (64.0%; 48.0%, 77.0%) and OS (64.0%; 22.0%, 92.0%). CONCLUSION: MIS for CES was associated with reduced operative time, length of stay, and blood loss, compared to OS. MIS was also associated with better post-operative lumbar/back and leg VAS scores and complete motor and bladder recovery rates. MIS and OS produced an equal average percentage of "excellent" patient outcomes.

An Exploration of the Neural Network of Lance-Adams Syndrome: a Case Report.

Lance-Adams syndrome (LAS) is a rare neurological disorder that may occur after cardiopulmonary resuscitation. The LAS is usually caused by hypoxic changes. Neuroimaging studies show that the brain pathology of LAS patients is not uniform, and the pathophysiology of the myoclonus can vary from patient to patient. Our case study contributes to this etiological heterogeneity by neuroimaging and transcranial magnetic stimulation (TMS). In patients with rare brain conditions such as LAS, a combination of brain stimulation methods, such as TMS, and diffusion tensor imaging can provide insights into this condition's pathophysiology. These insights can facilitate the development of more effective therapies.

How and when should interactome-derived clusters be used to predict functional modules and protein function?

MOTIVATION: Clustering of protein-protein interaction networks is one of the most common approaches for predicting functional modules, protein complexes and protein functions. But, how well does clustering perform at these tasks? RESULTS: We develop a general framework to assess how well computationally derived clusters in physical interactomes overlap functional modules derived via the Gene Ontology (GO). Using this framework, we evaluate six diverse network clustering algorithms using Saccharomyces cerevisiae and show that (i) the performances of these algorithms can differ substantially when run on the same network and (ii) their relative performances change depending upon the topological characteristics of the network under consideration. For the specific task of function prediction in S.cerevisiae, we demonstrate that, surprisingly, a simple non-clustering guilt-by-association approach outperforms widely used clustering-based approaches that annotate a protein with the overrepresented biological process and cellular component terms in its cluster; this is true over the range of clustering algorithms considered. Further analysis parameterizes performance based on the number of annotated proteins, and suggests when clustering approaches should be used for interactome functional analyses. Overall our results suggest a re-examination of when and how clustering approaches should be applied to physical interactomes, and establishes guidelines by which novel clustering approaches for biological networks should be justified and evaluated with respect to functional analysis. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Changes in Pharyngeal Width Over Time as an Indicator of Dysphagia in Stroke Patients.

OBJECTIVE: To verify the pharyngeal width at rest as a measurement that could be used to assess changes in the degree of dysphagia over time in stroke patients. METHODS: In a cohort of stroke patients, we performed serial measurements of the pharyngeal width at the midpoints of the second (C2) and third (C3) cervical vertebral bodies using lateral neck X-rays while the patients were at rest. The JOSCYL width, a parameter named after the first initial of each developers' surname and defined as the average value of the upper and lower pharyngeal widths, was used to formulate the JOSCYL scale, which was calculated as the JOSCYL width × 100/neck circumference. All patients also underwent serial videofluoroscopic swallowing studies (VFSSs). The Spearman correlation analysis was used to detect correlations between the serial VFSS results, JOSCYL widths, and JOSCYL scale values. RESULTS: Over time, we observed significant positive and negative correlations of change in the JOSCYL width and scale with changes in the Penetration-Aspiration Scale and the Dysphagia Outcome and Severity Scale scores, respectively. CONCLUSION: The JOSCYL width and JOSCYL scale clearly reflected changes in dysphagia in stroke patients over time. These parameters may provide an easier method for evaluating whether post-stroke dysphagia has been alleviated.

Predictive Value of Pharyngeal Width at Rest (JOSCYL Width) for Aspiration in Elderly People.

OBJECTIVE: To develop a new tool for aspiration risk prediction based on pharyngeal width at rest in older adults with symptoms of aspiration. METHODS: Lateral cervical spine roentgenograms were obtained from 33 older adult patients who complained of dysphagia and from 33 healthy, age-matched controls. Pharyngeal width at rest was measured at two points. We named the average of these two pharyngeal widths 'JOSCYL Width', calculated 'JOSCYL Scale', and compared these parameters between dysphagia and control groups. Correlations of individual JOSCYL Width and JOSCYL Scale, with Penetration Aspiration Scale (PAS) and Dysphagia Outcome and Severity Scale (DOSS) scores were analyzed for the dysphagia group. To determine optimal cutoff points for predicting aspiration, a receiver operating characteristic curve analysis was performed on JOSCYL Width and JOSCYL Scale. RESULTS: Both JOSCYL Width and JOSCYL Scale of the dysphagia group were larger than those of the control group (p<0.001). The correlation between JOSCYL Width and severity of dysphagia was significant for the dysphagia group (PAS p=0.007; DOSS p=0.012). The correlation between JOSCYL Scale and the severity of dysphagia was also significant for the dysphagia group (PAS p=0.009; DOSS p=0.011). Optimal cutoffs for JOSCYL Width and JOSCYL Scale for predicting aspiration were 20.0 mm and 5.9, respectively. CONCLUSION: JOSCYL Width and JOSCYL Scale can be new indicators for predicting aspiration in older adults. They are both precise and easy to use.

Increased white matter diffusivity associated with phantom limb pain.

BACKGROUND: We utilized diffusion tensor imaging (DTI) to evaluate the cerebral white matter changes that are associated with phantom limb pain in patients with unilateral arm amputation. It was anticipated that this would complement previous research in which we had shown that changes in cerebral blood volume were associated with the cerebral pain network. METHODS: Ten patients with phantom limb pain due to unilateral arm amputation and sixteen healthy age-matched controls were enrolled. The intensity of phantom limb pain was measured by the visual analogue scale (VAS) and depressive mood was assessed by the Hamilton depression rating scale. Diffusion tensor-derived parameters, including fractional anisotropy, mean diffusivity, axial diffusivity (AD), and radial diffusivity (RD), were computed from the DTI. RESULTS: Compared with controls, the cases had alterations in the cerebral white matter as a consequence of phantom limb pain, manifesting a higher AD of white matter in both hemispheres symmetrically after adjusting for individual depressive moods. In addition, there were associations between the RD of white matter and VAS scores primarily in the hemispheres related to the missing hand and in the corpus callosum. CONCLUSIONS: The phantom limb pain after unilateral arm amputation induced plasticity in the white matter. We conclude that loss of white matter integrity, particularly in the hemisphere connected with the missing hand, is significantly correlated with phantom limb pain.

The Association Between Postburn Vitamin D Deficiency and the Biomechanical Properties of Hypertrophic Scars.

Fibroblasts, keratinocytes, mast cells, and other cells participate in hypertrophic scar formation and express the vitamin D receptor. We investigated the association between vitamin D deficiency and the biomechanical properties of hypertrophic burn scars. This cross-sectional study analyzed 486 participants enrolled from May 1, 2013 to April 30, 2017. When complete wound healing was agreed with by the two opinions, blood sampling and scar evaluation were performed. The values of melanin and erythema, trans-epidermal water loss (TEWL), and scar distensibility and elasticity were measured using pigment- and TEWL-measuring devices and a suction skin elasticity meter. 25(OH) vitamin D deficiency was defined as plasma level of <20 ng/ml. The vitamin D-deficient patients had significantly higher mean values of scar melanin and TEWL (P = .032, P = .007), whereas scar erythema level was similar. They also showed significantly lower values of Uf (final distensibility; P < .001), Ua/Uf (gross elasticity; P < .001) and Ur/Uf (biological elasticity; P = .014), and higher value of Uv/Ue (viscoelasticity or potency against interstitial fluid shift; P = .016). In multiple linear regression analysis, Uf, Ua/Uf, Uv/Ue, and Ur/Uf were significantly affected by 25(OH)-vitamin D level in deficient patients (Uf, P = .017; Ua/Uf, P = .045; Uv/Ue, P = .024; Ur/Uf, P = .021). Our results demonstrated that vitamin D deficiency was significantly related to increased pigmentation, decreased skin barrier function, low scar distensibility and elasticity, and slow interstitial fluid movement in burn patients.

Hydroxyl-mediated ethanol selectivity of CO2 hydrogenation.

Oxide-supported Rh nanoparticles have been widely used for CO2 hydrogenation, especially for ethanol synthesis. However, this reaction operates under high pressure, up to 8 MPa, and suffers from low CO2 conversion and alcohol selectivity. This paper describes the crucial role of hydroxyl groups bound on Rh-based catalysts supported on TiO2 nanorods (NRs). The RhFeLi/TiO2 NR catalyst shows superior reactivity (≈15% conversion) and ethanol selectivity (32%) for CO2 hydrogenation. The promoting effect can be attributed to the synergism of high Rh dispersion and high-density hydroxyl groups on TiO2 NRs. Hydroxyls are proven to stabilize formate species and protonate methanol, which is easily dissociated into *CH x , and then CO obtained from the reverse water-gas shift reaction (RWGS) is inserted into *CH x to form CH3CO*, followed by CH3CO* hydrogenation to ethanol.

Pharyngeal width and aspiration after stroke.

It is important to assess aspiration in stroke patients. The aim of this study was to develop a new additional tool to predict aspiration based on the pharyngeal width at rest in stroke patients with aspiration symptoms. The pharyngeal width was measured at the middle level of the second and third cervical vertebral bodies using a lateral neck roentgenogram in stroke patients and healthy controls. We named the average of the two pharyngeal widths as the JOSCYL width and calculated the 'JOSCYL width × 100 / neck circumference' as the JOSCYL scale. The correlation between the individual JOSCYL width, JOSCYL scale, and severity of dysphagia was analyzed in the stroke group using Spearman correlation analysis. The optimal cutoff point for predicting aspiration was determined by receiver-operating characteristic curve analysis of the JOSCYL width and JOSCYL scale. The JOSCYL width and JOSCYL scale of the stroke group were larger than those of the control group (P < 0.05). The correlation between the JOSCYL width, JOSCYL scale, and the severity of dysphagia was significant for the whole stroke group and the chronic stroke group (P < 0.05). The optimal cutoffs of the JOSCYL width and JOSCYL scale for predicting aspiration were approximately 18 mm and 50 in the stroke group. The JOSCYL width and JOSCYL scale are new indicators for predicting aspiration in stroke patients and are precise and easy to use. The accuracy of the JOSCYL width and JOSCYL scale for predicting aspiration was shown to be higher in the chronic stroke stage.

Self-assembled nanostructured resistive switching memory devices fabricated by templated bottom-up growth.

Metal-oxide-based resistive switching memory device has been studied intensively due to its potential to satisfy the requirements of next-generation memory devices. Active research has been done on the materials and device structures of resistive switching memory devices that meet the requirements of high density, fast switching speed, and reliable data storage. In this study, resistive switching memory devices were fabricated with nano-template-assisted bottom up growth. The electrochemical deposition was adopted to achieve the bottom-up growth of nickel nanodot electrodes. Nickel oxide layer was formed by oxygen plasma treatment of nickel nanodots at low temperature. The structures of fabricated nanoscale memory devices were analyzed with scanning electron microscope and atomic force microscope (AFM). The electrical characteristics of the devices were directly measured using conductive AFM. This work demonstrates the fabrication of resistive switching memory devices using self-assembled nanoscale masks and nanomateirals growth from bottom-up electrochemical deposition.

Spectacle: fast chromatin state annotation using spectral learning.

Epigenomic data from ENCODE can be used to associate specific combinations of chromatin marks with regulatory elements in the human genome. Hidden Markov models and the expectation-maximization (EM) algorithm are often used to analyze epigenomic data. However, the EM algorithm can have overfitting problems in data sets where the chromatin states show high class-imbalance and it is often slow to converge. Here we use spectral learning instead of EM and find that our software Spectacle overcame these problems. Furthermore, Spectacle is able to find enhancer subtypes not found by ChromHMM but strongly enriched in GWAS SNPs. Spectacle is available at https://github.com/jiminsong/Spectacle.

Variation in piRNA and transposable element content in strains of Drosophila melanogaster.

Transposable elements (TEs) are one of the most important features of genome architecture, so their evolution and relationship with host defense mechanisms have been topics of intense study, especially in model systems such as Drosophila melanogaster. Recently, a novel small RNA-based defense mechanism in animals called the Piwi-interacting RNA (piRNA) pathway was discovered to form an adaptive defense mechanism against TEs. To investigate the relationship between piRNA and TE content between strains of a species, we sequenced piRNAs from 16 inbred lines of D. melanogaster from the Drosophila Genetic Reference Panel. Instead of a global correlation of piRNA expression and TE content, we found evidence for a host response through de novo piRNA production from novel TE insertions. Although approximately 20% of novel TE insertions induced de novo piRNA production, the abundance of de novo piRNAs was low and did not markedly affect the global pool of ovarian piRNAs. Our results provide new insights into the evolution of TEs and the piRNA system in an important model organism.