Deep learning models semi-automatic training system for quality control of transthoracic echocardiography

Objective To explore the value of deep learning(DL)models semi-automatic training system for automatic optimization of clinical image quality control of transthoracic echocardiography(TTE).Methods Totally 1250 TTE videos from 402 patients were retrospectively collected,including 490 apical four chamber(A4C),310 parasternal long axis view of left ventricle(PLAX)and 450 parasternal short axis view of great vessel(PSAX GV).The videos were divided into development set(245 A4C,155 PLAX,225 PSAX GV),semi-automated training set(98 A4C,62 PLAX,90 PSAX GV)and test set(147 A4C,93 PLAX,135 PSAX GV)at the ratio of 5∶2∶3.Based on development set and semi-automatic training set,DL model of quality control was semi-automatically iteratively optimized,and a semi-automatic training system was constructed,then the efficacy of DL models for recognizing TTE views and assessing imaging quality of TTE were verified in test set.Results After optimization,the overall accuracy,precision,recall,and F1 score of DL models for recognizing TTE views in test set improved from 97.33%,97.26%,97.26%and 97.26%to 99.73%,99.65%,99.77%and 99.71%,respectively,while the overall accuracy for assessing A4C,PLAX and PSAX GV TTE as standard views in test set improved from 89.12%,83.87%and 90.37%to 93.20%,90.32%and 93.33%,respectively.Conclusion The developed DL models semi-automatic training system could improve the efficiency of clinical imaging quality control of TTE and increase iteration speed.

Design of a ZnO/Poly(vinylidene fluoride)inverse opal film for photon localization-assisted full solar spectrum photocatalysis

Owing to its photonic band gap(PBG)and slow light effects,aniline black(AB)-poly(vinylidene fluoride)(PVDF)inverse opal(IO)photonic crystal(PC)was constructed to promote the utility of light and realize photothermal synergetic catalysis.As a highly efficient reaction platform with the capability of restricting heat,a microreactor was introduced to further amplify the photothermal effects of near infrared(NIR)radiation.The photocatalytic efficiency of ZnO/0.5AB-PVDF IO(Z0.5A)increases 1.63-fold compared to that of pure ZnO film under a full solar spectrum,indicating the effectiveness of synergetic promotion by slow light and photothermal effects.Moreover,a 5.85-fold increase is achieved by combining Z0.5A with a microreactor compared to the film in a beaker.The photon localization effect of PVDF IO was further exemplified by finite-difference time-domain(FDTD)calculations.In conclusion,photonic crystal-microreactor enhanced photothermal catalysis has immense potential for alleviating the deteriorating water environment.

Multi-head attention-based long short-term memory model for speech emotion recognition

To fully make use of information from different representation subspaces,a multi-head attention-based long short-term memory(LSTM)model is proposed in this study for speech emotion recognition(SER).The proposed model uses frame-level features and takes the temporal information of emotion speech as the input of the LSTM layer.Here,a multi-head time-dimension attention(MHTA)layer was employed to linearly project the output of the LSTM layer into different subspaces for the reduced-dimension context vectors.To provide relative vital information from other dimensions,the output of MHTA,the output of feature-dimension attention,and the last time-step output of LSTM were utilized to form multiple context vectors as the input of the fully connected layer.To improve the performance of multiple vectors,feature-dimension attention was employed for the all-time output of the first LSTM layer.The proposed model was evaluated on the eNTERFACE and GEMEP corpora,respectively.The results indicate that the proposed model outperforms LSTM by 14.6%and 10.5%for eNTERFACE and GEMEP,respectively,proving the effectiveness of the proposed model in SER tasks.

Multimodal emotion recognition based on deep neural network

In order to increase the accuracy rate of emotion recognition in voiceand video,the mixed convolutional neural network(CNN)and recurrent neural network(RNN)ae used to encode and integrate the two information sources.For the audio signals,several frequency bands as well as some energy functions are extacted as low-level features by using a sophisticated audio technique,and then they are encoded w it a one-dimensional(I D)convolutional neural network to abstact high-level features.Finally,tiese are fed into a recurrent neural network for te sake of capturing dynamic tone changes in a temporal dimensionality.As a contrast,a two-dimensional(2D)convolutional neural network and a similar RNN are used to capture dynamic facial appearance changes of temporal sequences.The method was used in te Chinese Natral Audio-'Visual Emotion Database in te Chinese Conference on Pattern Recognition(CCPR)in2016.Experimental results demonstrate that te classification average precision of the proposed metiod is41.15%,which is increased by16.62%compaed with te baseline algorithm offered by the CCPR in2016.It is proved ta t te proposed method has higher accuracy in te identification of emotional information.

In situ nanobubble sizing by visualization particle tracking and image-based dynamic light scattering

A novel method combining visualization particle tracking with image-based dynamic light scattering was developed to achieve the in situ and real-time size measurement of nanobubbles(NBs).First,the in situ size distribution of NBs was visualized by dark-field microscopy.Then,real-time size during the preparation was measured using image-based dynamic light scattering,and the longitudinal size distribution of NBs in the sample cell was obtained in a steady state.Results show that this strategy can provide a detailed and accurate size of bubbles in the whole sample compared with the commercial ZetaSizer Nano equipment.Therefore,the developed method is a real-time and simple technology with excellent accuracy,providing new insights into the accurate measurement of the size distribution of NBs or nanoparticles in solution.

Dissociated neural potential for processing strategies between decimals and fractions

The entity and symbolic fraction comparison tasks separating identification and semantic access stages based on event-related potential technology were used to investigate neural differences between fraction and decimal strategies in magnitude processing of nonsymbolic entities and symbolic numbers.The experimental results show that continuous entities elicit stronger left-lateralized anterior N2 in decimals,while discretized ones elicit more significant right-lateralized posterior N2 in fractions during the identification stage.On the other hand,decimals elicit stronger N2 over the left-lateralized fronto-central sites while fractions elicit the more profound P2 over the right-lateralized fronto-central sites and N2 at biparietal regions during the semantic access stage.Hence,for nonsymbolic entity processing,alignments of decimals and continuous entities activate the phonological network,while alignments of fractions and discretized entities trigger the visuospatial regions.For symbolic numbers processing,exact strategies with rote arithmetic retrieval in verbal format are used in decimal processing,while approximate strategies with complex magnitude processing in a visuospatial format are used in fraction processing.

Sum Frequency Spectroscopy Studies on Cell Membrane Fusion Induced by Divalent Cations

Cell membrane fusion is a fundamental biological process involved in a number of cellular living functions. Regarding this, divalent cations can induce fusion of the lipid bilayers through binding and bridging of divalent cations to the charged lipids, thus leading to the cell membrane fusion. However, the elaborate mechanism of cell membrane fusion induced by divalent cations is still needed to be elucidated.Here, surface/interface sensitive sum frequency generation vibrational spectroscopy(SFG-VS) and dynamic light scattering(DLS) were applied in this research to study the responses of phospholipid monolayer to the exposure of divalent metal ions i.e.Ca^(2+)and Mg^(2+). According to the particle size distribution results measured by DLS experiments, it was found that Ca^(2+)could induce inter-vesicular fusion while Mg^(2+)could not. An octadecyltrichlorosilane self-assembled monolayer(OTS SAM)-lipid monolayer system was designed to model the cell membrane for the SFG-VS experiment. Ca^(2+)could interact with the lipid POO_(2)^(-)head groups more strongly, resulting in cell membrane fusion more easily, in comparison with Mg^(2+). No specific interaction between the two metal cations and the C=O groups was observed. However, the C=O orientations changed more after Ca^(2+)-PO2-binding than Mg^(2+)mediation on lipid monolayer. Meanwhile, Ca^(2+)could induce dehydration of the lipids(which should be related to the strong Ca^(2+)-PO_(2)^(-)interaction), leading to the reduced hindrance for cell membrane fusion.

Evolutionary selection on synonymous codons in RNA G-quadruplex structural region

To investigate how synonymous codons have been adapted to the formation of ribonucleic acid(RNA)G-quadruplex(rG4)structure,a computational searching algorithm G4Hunter was applied to detect rG4 structures in protein-coding sequences of mRNAs in five eukaryotic species.The native sequences forming rG4s were then compared with randomized sequences to evaluate selection on synonymous codons.Factors that may influence the formation of rG4 were also investigated,and the selection pressures of rG4 in different gene regions were compared to explore its potential roles in gene regulation.The results show universal selective pressure acts on synonymous codons in rG4 regions to facilitate rG4 formation in five eukaryotic organisms.While G-rich codon combinations are preferred in the rG4 structural region,C-rich codon combinations are selectively unfavorable for rG4 formation.Gene's codon usage bias,nucleotide composition,and evolutionary rate can account for the selective variations on synonymous codons among rG4 structures within a species.Moreover,rG4 structures in the translational initiation region showed significantly higher selective pressures than those in the translational elongation region.

Application of polypyrrole nanofibers in the selective extraction of methotrexate and its polyglutamate metabolites from hospital effluents

A novel solid-phase extraction(SPE)strategy based on polypyrrole(Ppy)nanofibers was developed for the determination of trace methotrexate and its polyglutamate metabolites(MTXs)in hospital effluents.Ppy was coated on the surface of electrospun polystyrene nanofibers by in situ oxidative polymerization to form Ppy electrospun nanofibers.The mechanism of adsorption on MTXs was explored through static adsorption studies.The MTX contents,after extraction,were determined by liquid chromatography-tandem mass spectrometry analysis.Results show that the physical/chemical adsorption of targets occurs on the surface of Ppy nanofibers,which is most likely dominated by multiple adsorptions and heterogeneous adsorption sites.Ppy nanofibers exhibit satisfying extraction performance.The content of targets detected in medical wastewater samples ranges from 21 to 2908 ng/L.The novel strategy based on Ppy nanofiber SPE can extract trace MTXs effectively,guarantee analytical accuracy,and circumvent the storage and transportation of water samples during on-site sampling operations.

Intracellular temperature measurement by a dual-thermocouple difference method

A dual thermocouple difference technique is developed to determine the accuracy and anti-interference ability in the process of intracellular temperature measurement.First,two micro-nano thermocouples(TC)and a high-precision signal acquisition module are used to measure the temperature difference between the cell and the culture medium(separated about 10μm from the cell).The cold junctions of two TCs are connected to eliminate the setting of the reference temperature and enhance the anti-interference ability.Then,a low-noise voltage amplifier and digital acquisition card are used to sample signals.In order to verify the feasibility of the dual thermocouple difference method,the temperature changes of U251 cells are detected.The calibration results of two TCs show that the Seebeck coefficient is about 5μV/℃,and the signal acquisition accuracy is 0.5μV in a low voltage range(0-15μV).With the dual thermocouple difference method,errors due to the cold junctions can be removed and the interference caused by environmental temperature fluctuation can be reduced.The phenomenon of cellular temperature increase proves that the dual thermocouple difference method can detect the tiny temperature change of a single cell.The method potentially is a highly powerful technique for studying local thermogenesis of cells and helps to explore the relationship between cellular thermogenesis and cellular processes.