Mind the Gap: Learning Modality-Agnostic Representations With a Cross-Modality UNet.

Cross-modality recognition has many important applications in science, law enforcement and entertainment. Popular methods to bridge the modality gap include reducing the distributional differences of representations of different modalities, learning indistinguishable representations or explicit modality transfer. The first two approaches suffer from the loss of discriminant information while removing the modality-specific variations. The third one heavily relies on the successful modality transfer, could face catastrophic performance drop when explicit modality transfers are not possible or difficult. To tackle this problem, we proposed a compact encoder-decoder neural module (cmUNet) to learn modality-agnostic representations while retaining identity-related information. This is achieved through cross-modality transformation and in-modality reconstruction, enhanced by an adversarial/perceptual loss which encourages indistinguishability of representations in the original sample space. For cross-modality matching, we propose MarrNet where cmUNet is connected to a standard feature extraction network which takes as inputs the modality-agnostic representations and outputs similarity scores for matching. We validated our method on five challenging tasks, namely Raman-infrared spectrum matching, cross-modality person re-identification and heterogeneous (photo-sketch, visible-near infrared and visible-thermal) face recognition, where MarrNet showed superior performance compared to state-of-the-art methods. Furthermore, it is observed that a cross-modality matching method could be biased to extract discriminant information from partial or even wrong regions, due to incompetence of dealing with modality gaps, which subsequently leads to poor generalization. We show that robustness to occlusions can be an indicator of whether a method can well bridge the modality gap. This, to our knowledge, has been largely neglected in the previous works. Our experiments demonstrated that MarrNet exhibited excellent robustness against disguises and occlusions, and outperformed existing methods with a large margin (>10%). The proposed cmUNet is a meta-approach and can be used as a building block for various applications.

Reconstructing Randomly Masked Spectra Helps DNNs Identify Discriminant Wavenumbers.

Nondestructive detection methods, based on vibrational spectroscopy, are vitally important in a wide range of applications including industrial chemistry, pharmacy and national defense. Recently, deep learning has been introduced into vibrational spectroscopy showing great potential. Different from images, text, etc. that offer large labeled data sets, vibrational spectroscopic data is very limited, which requires novel concepts beyond transfer and meta learning. To tackle this, we propose a task-enhanced augmentation network (TeaNet). The key component of TeaNet is a reconstruction module that inputs randomly masked spectra and outputs reconstructed samples that are similar to the original ones, but include additional variations learned from the domain. These augmented samples are used to train the classification model. The reconstruction and prediction parts are trained simultaneously, end-to-end with back-propagation. Results on both synthetic and real-world datasets verified the superiority of the proposed method. In the most difficult synthetic scenarios TeaNet outperformed CNN by 17%. We visualized and analysed the neuron responses of TeaNet and CNN, and found that TeaNet's ability to identify discriminant wavenumbers was excellent compared to CNN. Our approach is general and can be easily adapted to other domains, offering a solution to more accurate and interpretable few-shot learning.

Effectiveness and safety of the Trevo® Retriever for mechanical thrombectomy in Chinese patients with acute ischemic stroke: Trevo Retriever China Registry.

BACKGROUND: To quantify the effectiveness and safety of the Trevo® Retriever for endovascular treatment of acute ischemic stroke (AIS) patients in China. METHODS: Trevo Retriever Registry (China) was a prospective, multicenter, non-comparative, open-label study of patients with AIS treated with the Trevo Retriever. The primary outcome was the proportion of patients achieving an expanded Thrombolysis in Cerebral Infarction (eTICI) score ≥2b at the end of endovascular treatment. Secondary outcomes included first-pass eTICI score ≥2b and 90-day modified Rankin Scale (mRS) score ≤2. RESULTS: The Trevo Retriever Registry (China) enrolled and followed 201 patients (62.1 ± 12.5 years-old; 70.6% male) at 11 centers. The pre-procedure NIHSS score and ASPECTS were 16 (interquartile range (IQR), 13-21) and 7 (IQR, 6-9), respectively, and 188 (93.5%) patients had an mRS score of 0 prior to the stroke. The main stroke etiology was large artery atherosclerosis, accounting for 71.6% (144/201) of patients. Post-procedure eTICI ≥2b was 98.4% (187/190). First-pass eTICI ≥2b was 74.7% (136/182). The 90-day good outcome (mRS ≤2) rate was 73.6% (148/201). The 90-day all-cause mortality was 5.5% (11/201). Neurological deterioration at 24 h post-procedure was observed in 7.7% (15/195) patients. Embolism in a new territory was seen in one patient (0.5%). Two (1.0%) procedure-related adverse events (AEs) occurred, which were intra-procedure cerebral artery embolism. No Trevo related AEs occurred. CONCLUSIONS: This real-world study of the Trevo Retriever in China demonstrated a high rate of revascularization and first-pass success that resulted in an overall high good function outcome rate and low mortality.

Ionic-Wind-Enhanced Raman Spectroscopy without Enhancement Substrates.

Raman spectra are often masked by strong fluorescence, which severely hinders the applications of Raman spectroscopy. Herein, for the first time, we report ionic-wind-enhanced Raman spectroscopy (IWERS) incorporated with photobleaching (PB) as a noninvasive approach to detect fluorescent and vulnerable samples without a substrate. In this study, ionic wind (IW) generated by needle-net electrodes transfers charges to the sample surface in air on the scale of millimeters rather than nanometers in surface-enhanced Raman spectroscopy. Density functional theory calculations reveal that the ionic particles in IW increase the susceptibility of the sample molecules, thus enhancing the Raman signals. Meanwhile, the incorporation of IW with PB yields a synergistic effect to quench fluorescence. Therefore, this approach can improve the signal-to-noise ratio of Raman peaks up to three times higher than that with only PB. At the same time, IWERS can avoid sample pollution and destruction without substrates as well as high laser power. For archeological samples and a red rock as an analogue to Mars geological samples, IWERS successfully identified weak but key Raman peaks, which were masked by strong florescence. It suggests that IWERS is a promising tool for characterizations in the fields of archeology, planetary science, biomedicine, and soft matter.

Towards an accurate and robust analysis pipeline for somatic mutation calling.

Accurate and robust somatic mutation detection is essential for cancer treatment, diagnostics and research. Various analysis pipelines give different results and thus should be systematically evaluated. In this study, we benchmarked 5 commonly-used somatic mutation calling pipelines (VarScan, VarDictJava, Mutect2, Strelka2 and FANSe) for their precision, recall and speed, using standard benchmarking datasets based on a series of real-world whole-exome sequencing datasets. All the 5 pipelines showed very high precision in all cases, and high recall rate in mutation rates higher than 10%. However, for the low frequency mutations, these pipelines showed large difference. FANSe showed the highest accuracy (especially the sensitivity) in all cases, and VarScan and VarDictJava outperformed Mutect2 and Strelka2 in low frequency mutations at all sequencing depths. The flaws in filter was the major cause of the low sensitivity of the four pipelines other than FANSe. Concerning the speed, FANSe pipeline was 8.8∼19x faster than the other pipelines. Our benchmarking results demonstrated performance of the somatic calling pipelines and provided a reference for a proper choice of such pipelines in cancer applications.

Strategies and techniques for improving heat resistance and mechanical performances of poly(lactic acid) (PLA) biodegradable materials.

Poly(lactic acid) (PLA) has attracted much attention as a substitute for petroleum-based plastics, but its low heat resistance limits its application range in packaging fields and disposable products. This paper summarizes the structural factors affecting the heat resistance of PLA and systematically summarizes methods to improve its heat resistance. PLA is a semi-crystalline polymer, and crystallinity, crystal size, and other factors are important factors affecting heat resistance. This paper systematically analyzes the means to control the crystallization behavior of PLA, and summarizes the effects of nucleating agents, cross-linking, grafting, and annealing processes on the crystallization behavior and heat resistance of PLA. The effects of PLA molecular chain cross-linking and grafting on the motility of PLA molecular chains and the heat resistance of PLA materials are further discussed from the perspective of PLA molecular chain segment movement. The research work on combining PLA with reinforcements such as high heat-resistant polymer materials, fiber, and nanoparticles to improve the mechanical properties and heat resistance of PLA by introducing rigid groups or structures is described in detail. Improving the heat resistance of PLA material is an important strategy to promote the application of biodegradable materials, and has broad research value and application prospects.

Ultra-fast fabrication of Bi2Te3 based thermoelectric materials by flash-sintering at room temperature combining with spark plasma sintering.

Highly crystalline Bi2Te3 based compounds with small grain size were successfully synthesized by flash sintering (FS) method in 10 s at room temperature under suitable current density using Bi, Te and Se powders. The instantaneously generated local Joule heat at grain boundary is regarded as the main reason for the rapid completion of chemical reaction and crystallization. By combining FS synthesis method with spark plasma sintering (SPS), Bi2Te3 based bulk materials with high relative density were fabricated in 10 min. Suitably prolonging sintering temperature and holding time in SPS process can decrease carrier concentration and phonon thermal conductivity, while increasing carrier mobility. Hence, the sample prepared at 753 K for 3 min shows 20% higher ZT value than that of the sample prepared at 723 K for 3 min. Compared with common zone melting or powder metallurgy methods taking several hours by complex operation, this method is time-saving and low cost.

Genetic U-Net: Automatically Designed Deep Networks for Retinal Vessel Segmentation Using a Genetic Algorithm.

Recently, many methods based on hand-designed convolutional neural networks (CNNs) have achieved promising results in automatic retinal vessel segmentation. However, these CNNs remain constrained in capturing retinal vessels in complex fundus images. To improve their segmentation performance, these CNNs tend to have many parameters, which may lead to overfitting and high computational complexity. Moreover, the manual design of competitive CNNs is time-consuming and requires extensive empirical knowledge. Herein, a novel automated design method, called Genetic U-Net, is proposed to generate a U-shaped CNN that can achieve better retinal vessel segmentation but with fewer architecture-based parameters, thereby addressing the above issues. First, we devised a condensed but flexible search space based on a U-shaped encoder-decoder. Then, we used an improved genetic algorithm to identify better-performing architectures in the search space and investigated the possibility of finding a superior network architecture with fewer parameters. The experimental results show that the architecture obtained using the proposed method offered a superior performance with less than 1% of the number of the original U-Net parameters in particular and with significantly fewer parameters than other state-of-the-art models. Furthermore, through in-depth investigation of the experimental results, several effective operations and patterns of networks to generate superior retinal vessel segmentations were identified. The codes of this work are available at https://github.com/96jhwei/Genetic-U-Net.

Laparoscopic Pancreaticoduodenectomy With Transanal Specimen Extraction for Periampullary Tumors.

OBJECTIVE: To explore the possibility of LPD with transanal specimen extraction for periampullary tumors. BACKGROUND: Natural orifice specimen extraction has been widely performed in colorectal surgery. But so far there is no report regarding natural orifice specimen extraction in LPD. METHODS: Data of 3 patients who underwent LPD with transanal specimen extraction by the same surgeon between July 2018 and March 2019 due to periampullary tumors were evaluated retrospectively. RESULTS: All patients underwent LPD with no conversion to open surgery. Specimens were all extracted transanally. No complications occurred except for delayed gastric emptying in 1 patient. The patients' anorectal function was intact and no opioids were required postoperatively. After a follow-up of 17-25 months, all patients were alive with no tumor recurrence. CONCLUSION: Treatment of periampullary tumors by LPD with transanal specimen extraction is feasible and safe.

Comparing the efficacy and safety of the Skyflow device with those of the Solitaire FR stent in patients with acute ischemic stroke: a prospective, multicenter, randomized, non-inferiority clinical trial.

BACKGROUND: Mechanical thrombectomy is the standard treatment for acute ischemic stroke (AIS) with large vessel occlusion (LVO) in the anterior circulation. This trial aimed to indicate whether Skyflow, a new thrombectomy device, could achieve the same safety and efficacy as Solitaire FR in the treatment of AIS. METHODS: This study was a prospective, multicenter, randomized, single blind, parallel, positive controlled, non-inferiority clinical trial. Patients with intracranial anterior circulation LVO within 8 hours from onset were included to receive thrombectomy treatment with either the Skyflow or Solitaire FR stent retriever. The primary endpoint was the rate of successful reperfusion (modified Treatment In Cerebral Infarction (mTICI) ≥2b) after the operation. The safety endpoints were the rate of symptomatic intracranial hemorrhage (sICH) and subarachnoid hemorrhage (SAH) at 24 hours after operation. RESULTS: A total of 95 and 97 patients were involved in the Skyflow group and Solitaire FR group, respectively. A successful reperfusion (mTICI ≥2b) was finally achieved in 84 (88.4%) patients in the Skyflow group and 80 (82.5%) patients in the Solitaire FR group. Skyflow was non-inferior to Solitaire FR in regard to the primary outcome, with the criterion of a non-inferiority margin of 12.5% (p=0.0002) after being adjusted for the combined center effect and the National Institutes of Health Stroke Scale (NIHSS) score. The rate of periprocedural sICH and SAH did not differ significantly between the two groups. CONCLUSION: Endovascular thrombectomy with the Skyflow stent retriever was non-inferior to Solitaire FR with regard to successful reperfusion in AIS due to LVO (with a pre-specified non-inferiority margin of 12.5%).

Recent progress of preparation of branched poly(lactic acid) and its application in the modification of polylactic acid materials.

Poly (lactic acid) (PLA) with branched structure has abundant terminal groups, high melt strength, good rheological properties, and excellent processability; it is a new research and application direction of PLA materials. This study mainly summarizes the molecular structure design, preparation methods, basic properties of branched PLA, and its application in modified PLA materials. The structure and properties of branched PLA prepared by ring-opening polymerization of monomer, functional group polycondensation, and chain extender in the processing process were introduced. The research progress of in situ formation of branched PLA by initiators, multifunctional monomers/additives through dynamic vulcanization, and irradiation induction was described. The effect of branched PLA on the structure and properties of linear PLA materials was analyzed. The role of branched PLA in improving the crystallization behavior, phase morphology, foaming properties, and mechanical properties of linear PLA materials was discussed. At the same time, its research progress in biomedicine and tissue engineering was analyzed. Branched PLA has excellent compatibility with PLA, which has important research value in regulating the structure and properties of PLA materials.

Effects of and prognostic factors affecting endovascular mechanical thrombectomy of acute vertebrobasilar artery occlusion.

PURPOSE: To investigate the clinical outcome and factors affecting the prognosis of endovascular mechanical thrombectomy of acute vertebrobasilar artery occlusion. MATERIALS AND METHODS: Eighty-three patients with acute vertebrobasilar artery occlusion were treated with endovascular mechanical thrombectomy, and the recanalization rate, clinical outcomes at three months, modified DWI-PC-ASPECTS, and MRA-BATMAN scores were analyzed. RESULTS: Following acute mechanical thrombectomy, the TICI 2B-3 score was achieved in all patients (100%). At three-month evaluation, 56 (67.5%) patients had good prognosis with the mRS score of 0-2, including 13 (23.2%) patients who had arterial occlusion caused by emboli and 43 (76.8%) who had atherosclerotic stenosis. In analyzing factors affecting the prognosis, a significant difference (P < 0.05) existed between patients with good (mRS 0-2) and poor (mRS 3-6) prognosis in the NIHSS (17.3 vs. 31.2, P = 0.000001), modified DWI-PC-ASPECTS (10.4 vs. 7.8, P = 0.021), and MRA-BATMAN (6.3 vs. 4.6, P = 0.003) scores. Univariate Logistic regression analysis demonstrated NIHS score ≥ 21, modified DWI-PC-ASPECTS score ≤ 8.5, and MRA-BATMAN score ≤ 6.5 to be the risk factors for poor prognosis. Multivariate Logistic regression analysis revealed NIHSS score ≥ 21 as an independent risk factor for poor prognosis. CONCLUSION: Endovascular mechanical thrombectomy is safe and effective in recanalizing occluded vertebrobasilar artery occlusion, and NIHS score ≥ 21, modified DWI-PC-ASPECTS score ≤ 8.5, and MRA-BATMAN score ≤ 6.5 are the risk factors for poor prognosis.

Characterisation of transient electromagnetic signals during fixed interference sources in tunnel structure.

The detection effect of the transient electromagnetic method is ambiguous in engineering applications due to the existence of interference sources, so explaining the influence of these fixed interference sources on is crucial. In this paper, the response characterisation of transient electromagnetic signals of fixed interference sources are thoroughly investigated. First, the secondary field generated by these interference sources is analyzed, and a typical fixed interference source is calculated. Then, a sensitivity analysis of the transient electromagnetic response curve is carried out. Finally, the mathematical superposition method for multiple field sources is proposed and verified. The results illustrate that the transient electromagnetic response curve of uniform full-space surrounding rock with a single fixed interference source has an apparent lifting phenomenon in the middle stage and presents an approximate horizontal change rule at the late stage. The transient electromagnetic response curves of multiple field sources separately illustrate the response characterisation of different field sources at different time stages. These research results can provide a valuable reference for the on-site interpretation of detection signals.

A Universal Positioning System for Coupling Characterization of SEM and AFM.

Hyphenated techniques, providing comprehensive information in various aspects such as constituent, structure, functional group, and morphology, play an important role in scientific research. Nowadays, coupling characterization of the same position in microscale is in great need in the field of nanomaterial research and exploration. In this article, a new hyphenated technique was developed to facilitate the coupling characterization of atomic force microscope (AFM) and scanning electron microscope (SEM) by designing a universal positioning system. The system consisted of a specimen holder with coordinate grids and a software for converting the coordinate values of the same point to fit SEM, specimen holder, and AFM system. In working condition, the coordinates of the labeled points and target position were firstly extracted from the SEM operation software, then converted into the numerical values adapted to the specimen holder itself, and finally transformed into the coordinates matching the AFM system. The experimental result showed that a retrieving rate of 96% was achieved for a spherical target with a diameter of 1 µm in a 30 µm × 30 µm square. The hyphenated technique is a universal, accurate, efficient, and financially feasible method in microanalysis field and has great application potential.

Heterostructured Ni3S2-Ni3P/NF as a Bifunctional Catalyst for Overall Urea-Water Electrolysis for Hydrogen Generation.

Urea oxidation reaction (UOR) has been proposed to replace the formidable oxygen evolution reaction (OER) to reduce the energy consumption for producing hydrogen from electrolysis of water owing to its much lower thermodynamic oxidation potential compared to that of the OER. Therefore, exploring a highly efficient and stable hydrogen evolution and urea electrooxidation bifunctional catalyst is the key to achieve economical and efficient hydrogen production. In this paper, we report a heterostructured sulfide/phosphide catalyst (Ni3S2-Ni3P/NF) synthesized via one-step thermal treatment of Ni(OH)2/NF, which allows the simultaneous occurrence of phosphorization and sulfuration. The obtained Ni3S2-Ni3P/NF catalyst shows a sheet structure with an average sheet thickness of ∼100 nm, and this sheet is composed of interconnected Ni3S2 and Ni3P nanoparticles (∼20 nm), between which there are a large number of accessible interfaces of Ni3S2-Ni3P. Thus, the Ni3S2-Ni3P/NF exhibits superior performance for both UOR and hydrogen evolution reaction (HER). For the overall urea-water electrolysis, to achieve current densities of 10 and 100 mA cm-2, cell voltage of only 1.43 and 1.65 V is required using this catalyst as both the anode and the cathode. Moreover, this catalyst also maintains fairly excellent stability after a long-term testing, indicating its potential for efficient and energy-saving hydrogen production. The theoretical calculation results show that the Ni atoms at the interface are the most efficient catalytically active site for the HER, and the free energy of hydrogen adsorption is closest to thermal neutrality, which is only 0.16 eV. A self-driven electron transfer at the interface, making the Ni3S2 sides become electron donating while Ni3P sides become electron withdrawing, may be the reason for the enhancement of the UOR activity. Therefore, this work shows an easy treatment for enhancing the catalytic activity of Ni-based materials to achieve high-efficiency urea-water electrolysis.

Proctotomy leak following laparoscopic total gastrectomy with transrectal specimen extraction for gastric cancer: a case report.

BACKGROUND: Despite increasing acceptance in colorectal surgery, natural orifice specimen extraction (NOSE) surgery for the treatment of gastric cancer is still in its infancy, especially via the transrectal approach, which was barely reported. So little is known about its complications. Here we report the first case of proctotomy leak after transrectal NOSE gastrectomy, and our experience in preventive interventions. CASE PRESENTATION: A 62-year-old male patient complaining of upper abdominal pain who underwent open distal gastrectomy for gastric cancer one year ago was diagnosed with recurrent gastric cancer by gastroscopic biopsy. We performed laparoscopic total gastrectomy with transrectal specimen extraction on the patient. The operation was completed in a total laparoscopic approach and the specimen was extracted through a 3 cm longitudinal incision in the anterior wall of the upper rectum, then interrupted sutures were used for full-thickness closure of the rectal incision. The operative time was 470 min and intra-operative blood loss was 100 mL. The postoperative pathological examination showed pT1bN0M0 gastric adenocarcinoma. The patient developed proctotomy leak on the 10th postoperative day. We analyzed the causes of this rare complication and put forward a series of technical improvements. After failure of conservative treatment, a diverting ileostomy was created and the patient eventually recovered. We successfully prevented proctotomy leak in the subsequent 20 transrectal NOSE gastrectomies using improved techniques. CONCLUSIONS: Proctotomy leak after transrectal specimen extraction should be considered among the complications of NOSE surgery and can be prevented by technical precautions.

Machine learning powered ellipsometry.

Ellipsometry is a powerful method for determining both the optical constants and thickness of thin films. For decades, solutions to ill-posed inverse ellipsometric problems require substantial human-expert intervention and have become essentially human-in-the-loop trial-and-error processes that are not only tedious and time-consuming but also limit the applicability of ellipsometry. Here, we demonstrate a machine learning based approach for solving ellipsometric problems in an unambiguous and fully automatic manner while showing superior performance. The proposed approach is experimentally validated by using a broad range of films covering categories of metals, semiconductors, and dielectrics. This method is compatible with existing ellipsometers and paves the way for realizing the automatic, rapid, high-throughput optical characterization of films.

Improved stability of ß-CsPbI3 inorganic perovskite using π-conjugated bifunctional surface capped organic cations for high performance photovoltaics.

Multiple-ring naphthalenediammonium is employed for the first time to overcome the intrinsic instability of ß-CsPbI3 perovskite via anchoring the ammonium groups occupying A-site vacancies. It improves charge transport and moisture stability giving out a champion power conversion efficiency of 16.69% for an excellent inorganic perovskite solar cell.

Relationship between the Young's Modulus and the Crystallinity of Cross-Linked Poly(ε-caprolactone) as an Immobilization Membrane for Cancer Radiotherapy.

Cancer is a leading cause of death in the world. In cancer radiotherapy, immobilization membranes composed of cross-linked poly(ε-caprolactone) (PCL) are utilized for patient positioning. A higher-dimensional stability of the membrane is urgently required to facilitate more accurate radiation dose delivery. It is extremely important to establish the relationship between the degree of crystallinity and the Young's modulus (E) because it determines the mechanical properties and can be modulated by crystallinity. When two components of the membrane with different strains are in contact, a gradient region adjacent to the interface is formed and confirmed by attenuated total reflection infrared microscopy. Atomic force microscopy (AFM) and Raman spectroscopy are used to scan the same area in the gradient region (14 µm × 14 µm) to characterize E and crystallinity (X Raman), respectively. This co-localized method ensures the accuracy of the relationship. Finally, 1764 AFM measurement data are processed and 49 pairs of E-X Raman data are obtained. The regression curve shows that E monotonically increases with X Raman. The nonlinearity of the curve may be attributed to the α-relaxation and cross-linking of PCL chains. The chemical structure of this material significantly impacts the mechanical properties, thus requiring future investigation.

Tailoring the physicochemical properties of chitosan-derived N-doped carbon by controlling hydrothermal carbonization time for high-performance supercapacitor application.

Although a few methods have been employed to fabricate N-doped porous carbons from various N-containing biomass resources, it is still a big challenge to obtain porous carbons with high supercapacitance performances. Herein, we demonstrate that aN-doped porous carbon with superior supercapacitance can be prepared from chitosan by properly controlling hydrothermal carbonization (HC). The physicochemical and supercapacitance properties of the HC-derived carbon are highly time-dependent and can be readily tailored. As compared with traditional direct pyrolysis, the proper control of HC time plays a very important role in promoting the supercapacitance performances of the N-doped carbon by increasing turbostratic structure, doped N content and active N species, specific surface area, and especially balancing micro- and mesoporosity. These synergistic effects produce a N-doped carbon with an ultrahigh specific capacitance of 406 ± 36 F g-1 in a three-electrode system, outstanding rate capability, and ultrahigh energy density (23.6 ± 3.1 W h kg-1).