The Mars orbiter magnetometer of Tianwen-1:in-flight performance and first science results

The Mars Orbiter MAGnetometer(MOMAG)is a scientific instrument onboard the orbiter of China’s first mission for Mars—Tianwen-1.Since November 13,2021,it has been recording magnetic field data from the solar wind to the magnetic pile-up region surrounding Mars.Here we present its in-flight performance and first science results,based on its first one and one-half months’data.Comparing these early MOMAG observations to the magnetic field data in the solar wind from NASA’s Mars Atmosphere and Volatile EvolutioN(MAVEN)mission,we report that the MOMAG magnetic field data are at the same level in magnitude,and describe the same magnetic structures with similar variations in three components.We recognize 158 clear bow shock(BS)crossings in these MOMAG data;their locations match well statistically with the modeled average BS.We also identify and compare five pairs of datasets collected when Tianwen-1’s orbiter and the MAVEN probe made simultaneous BS crossings.These BS crossings confirm the global shape of modeled BS,as well as the south-north asymmetry of the Martian BS.Two cases presented in this paper suggest that the BS is probably more dynamic at flank than near the nose.So far,MOMAG performs well,and provides accurate magnetic field vectors.MOMAG is continuously scanning the magnetic field surrounding Mars.Data from MOMAG’s measurements complement data from MAVEN and will undoubt edly advance our understanding of the plasma environment of Mars.

Is sentinel lymph node biopsy necessary for the patients diagnosed with breast ductal carcinoma in situ using core needle biopsy or vacuum-assisted biopsy as the initial diagnostic method？

Objective: Axillary lymph node status is one of the most important prognostic indicator of survival for breast cancer, especially in ductal carcinoma in situ(DCIS). The purpose of this study was to investigate whether sentinel lymph node biopsy(SLNB) should be performed in patients with an initial diagnosis of DCIS. Methods: A retrospective study was performed of 124 patients with an initial diagnosis of DCIS between March 2000 and June 2014. The patients were treated with either SLNB or axillary node dissection during the surgery, and we compared the clinicopathologic characteristics, image features, and immunohistochemical results. Results: Eighty-two patients(66.1%) had pure DCIS and 25(20.2%) had DCIS with microinvasion(DCISM), 17(13.7%) updated to invasive breast cancer(IBC). 115 patients(92.7%) underwent SLNB, among them, 70 patients(56.5%) underwent axillary node dissection. 3 of 115 patients(2.6%) had a positive sentinel lymph node, only 1(1.4%) of 70 patients had axillary lymph node metastasis, in 84 patients(66.7%) who were diagnosed DCIS by core needle biopsy(CNB) and vacuum-assisted biopsy(VAB). 26 patients(31.0%) were upstaged into IBC or DCISM in the final histological diagnosis. The statistically significant factors predictive of underestimation were large tumor size, microcalcifications, comedo necrosis, positive Her-2 status, negative estrogen receptor status. Conclusion: The metastasis of sentinel lymph nodes in pure DCIS is very low, but the underestimation of invasive carcinoma in patients with an initial diagnosis of DCIS is an usual incident, especially in the cases when DCIS is diagnosed by CNB or VAB. Our findings suggest patients presenting with a preoperative diagnosis of DCIS associated with large tumor sizes, microcalcifications, comedo necrosis, positive Her-2 status, negative ER status are more likely to be DCISM and IBC in final diagnosis. SLNB should be performed in this part of patients.

Study on Microwave Pretreatment Technology to Improve the Effect of Shellac Impregnation of Fast-Growing Chinese Fir

To improve the mechanical properties of fast-growing Chinese fir(Cunnighamia lanceolate),expand its range of application,increase its value,and avoid the environmental pollution caused by impregnation with synthetic resin,Chinese fir was impregnated with a shellac solution.Since the shellac solution was difficult to penetrate into fast-growing Chinese fir,so microwave pretreatment was used to irradiate the wood to improve the permeability.This study investigated the effects of four factors,including the content of moisture in the wood before it was microwaved,the chamber pressure of microwave,the time of microwaving and the vacuum impregnation on the mechanical properties of Chinese fir wood.When the moisture content of wood before microwave was approximately 50%–60%,after microwaving and impregnation,the ultimate strength in static bending(modulus of rupture[MOR])and strength in compression perpendicular to the grain(SCPG)of the wood increased significantly.A microwave time of 100 seconds was more effective at improving the MOR and SCPG of the wood.If the wood was microwaved for too short or long period of time,the microwave pretreatment was not effective.When the samples were immersed in shellac for a longer period,the MOR and SCPG of Chinese fir gradually increased,but when the wood was impregnated for more than 12 hours,the increases were not significant.After the shellac penetrated the Chinese fir wood,it spread on the inner wall surface of tracheid to form a shellac film and easily formed plug-like deposits in microcapillaries.The use of Fourier-transform infrared spectroscopy,scanning electron microscopy and scanning electron microscopy-energy dispersion x-ray indicated that the microwave pretreatment can destroy the pit membranes on tracheids and facilitate the ability of shellac to penetrate the channel.

Effects of Different Modes of Returning Farmland to Forest on Physicochemical Properties of Soil in Bashang Area of Northern Hebei Province

[Objective]The research aimed to understand the change rules of physicochemical properties of soil in the different modes of returning farmland to forest.[Method]The physicochemical properties of soil in three different types of returning farmland to forest(Hippophae rhamnoides Linn.forest,Caragana korshinskii forest and Ulmus pumila L.forest)were analyzed by taking the enclosed grassland as comparison in Bashang area of northern Hebei Province.[Result]The results showed that there were significant differences among different types.Order of the bulk density of soil:H.rhamnoides Linn.forest(1.17 g/cm^3)<U.pumila L.forest(1.24 g/cm^3)<C.korshinskii forest(1.26 g/cm^3)<enclosed grassland(1.61 g/cm^3);order of the total porosity of soil:H.rhamnoides Linn.forest(55.29%)>U.pumila L.forest(52.55%)>C.korshinskii forest(51.93%)<enclosed grassland(38.85%);order of the organic matter content of soil:H.rhamnoides Linn.forest(3.191%)>C.korshinskii forest(3.128%)>U.pumila L.forest(2.804%)>enclosed grassland(2.536%).[Conclusion]After returning farmland to forest,vegetation played a positive role in physicochemical properties of soil.There were significant differences in physicochemical properties of soil among different modes of returning farmland to forest,and the improvement effect of H.rhamnoides Linn.forest on physicochemical properties of soil was the best.

A database of modeled gridded dry deposition velocities for 45 gaseous species and three particle size ranges across North America

The dry deposition process refers to the flux loss of an atmospheric pollutant due to uptake of the pollutant by the earth’s surfaces.Dry deposition flux of a chemical species is typically calculated as the product of its surface-layer concentration and its dry deposition velocity(V_(d)).Field measurement based V_(d) data are very scarce or do not exist for many chemical species considered in chemistry transport models.In the present study,gaseous and particulate dry deposition schemes were applied to generate a database of hourly V_(d) for 45 gaseous species and three particle size ranges for two years(2016–2017)at a 15 kmby 15 km horizontal resolution across North America.Hourly V_(d) of the 45 gaseous species ranged from<0.001 to 4.6 cm/sec across the whole domain,with chemical species-dependentmedian(mean)values being in the range of 0.018–1.37 cm/sec(0.05–1.43 cm/sec).The spatial distributions of the two-year average V_(d) showed values higher than 1–3 cm/sec for those soluble and reactive species over certain land types.Soluble species have the highest V_(d) over water surfaces,while insoluble but reactive species have the highest V_(d) over forests.Hourly V_(d) of PM_(2.5) across the whole domain ranged from 0.039 to 0.75 cm/sec with median(mean)value of 0.18(0.20)cm s^(−1),while the mean V_(d) for PM_(2.5)–10 is twice that of PM_(2.5).Uncertainties in the modeled V_(d) are typically on the order of a factor of 2.0 or larger,which needs to be considered when applying the dataset in other studies.

Prevalence,variation,and transmission patterns of human respiratory syncytial virus from pediatric patients in Hubei,China during 2020–2021

Human respiratory syncytial virus(RSV)is a severe threat to children and a main cause of acute lower respiratory tract infections.Nevertheless,the intra-host evolution and inter-regional diffusion of RSV are little known.In this study,we performed a systematic surveillance in hospitalized children in Hubei during 2020–2021,in which 106 RSV-positive samples were detected both clinically and by metagenomic next generation sequencing(mNGS).RSV-A and RSV-B groups co-circulated during surveillance with RSV-B being predominant.About 46 high-quality genomes were used for further analyses.A total of 163 intra-host nucleotide variation(iSNV)sites distributed in 34 samples were detected,and glycoprotein(G)gene was the most enriched gene for iSNVs,with non-synonymous substitutions more than synonymous substitutions.Evolutionary dynamic analysis showed that the evolutionary rates of G and NS2 genes were higher,and the population size of RSV groups changed over time.We also found evidences of inter-regional diffusion from Europe and Oceania to Hubei for RSV-A and RSV-B,respectively.This study highlighted the intra-host and inter-host evolution of RSV,and provided some evi-dences for understanding the evolution of RSV.

Prognostic Significance of Tumor-infiltrating CD8^＋ or CD3^＋ T Lymphocytes and Interleukin-2 Expression in Radically Resected Non-small Cell Lung Cancer

Background：Altered immunoresponse is associated with tumorigenesis and cancer progression.This study assessed the levels of tumor-infiltrating CD3^＋ or CD8^＋ T lymphocytes and interleukin-2 （IL-2） protein in radically resected non-small cell lung cancer （NSCLC） tissues to predict overall survival （OS） of the patients.Methods：Paraffin-embedded tissue specimens from 129 NSCLC patients were retrospectively collected for immunostaining of CD8^＋,CD3^＋,and IL-2 expression.Clinicopathological and survival data were collected and analyzed using the Chi-squared test,Kaplan-Meier curves,and the log-rank test or the Cox regression model.Results：The data showed a significant inverse association between CD8^＋ T lymphocyte levels and IL-2 expression （r =-0.927; P =0.000） and between the levels of CD8^＋ and CD3^＋ T lymphocytes （r =-0.722; P =0.000）,but a positive association between CD3^＋ T lymphocyte levels and IL-2 expression （r =0.781; P =0.000） in NSCLC tissues.Furthermore,the levels of CD3^＋ and CD8^＋ T lymphocytes and IL-2 expression were associated with tumor stage （P =0.023,0.006,and 0.031,respectively） and the level of CD8^＋ T lymphocytes was associated with the patient gender （P =0.024）.In addition,the levels of CD8^＋ T lymphocytes were associated with an unfavorable 5-year OS,whereas patients with high levels of CD3^＋ T lymphocytes in tumor lesions and IL-2-expressing tumors had significantly better 5-year OS rates than patients with low levels.Conclusions：The levels of CD8^＋ T cells in tumor lesions and IL-2 expression were both independent predictors of OS for these NSCLC patients.Thus,the detection of tumor-infiltrating CD3^＋ or CD8^＋ T lymphocytes and IL-2 expression could be useful to predict the prognosis of radically resected NSCLC patients.

Long-term trends of fine particulate matter and chemical composition in the Pearl River Delta Economic Zone （PRDEZ）, China

Understanding essential for formulating the trends in PM2.5 levels is clean air plans. This paper analyzes PM2.5 data from various published sources for the years 2000 to 2010 in the Pearl River Delta Economic Zone （PRDEZ）. The long-term variation in PM2.5 mass concentration is analyzed. Results show that PM2.5, organic carbon （OC）, elemental carbon （EC）, and SO24 show a similar trend, increasing before 2005 and then decreasing slightly. The annual average PM2.5 concentra- tion ranges from 49.1 μg·m-3 in 2000 to 64.3 μg·m-3 in 2010, with a peak of 84.1 μg· m3 in 2004. None of these 11 years meets the new National Ambient Air Quality standard （NAAQS） for PM2.5 （35 μg· m-3）. Overall average concentrations of OC, EC, and SO2/4 are 13.0, 6.5, and 11.8μg·m-3, respectively. NO3 and NH＋ respectively have concentrations of 1.5 μg·m-3 and 2.9 μg·m-3 in 2000 and 6.4 μg·m 3 and 5.3μg·m-3 in 2010, with a statistically significant average annual trend of＋ 0.2 μg·m-3 ·yr-1 and ＋ 0.1 μg· m-3. yr-1. In certain geographic regions, OC and EC contribute most of the PM2.5, while in other regions secondary water-soluble ions are more important. In general, OC and SO2/4- are the dominant components of PM2.5, contributing 20.6% and 18.6%, respectively. These results provide, for the first time, a better understanding of the long-term PM2.5 characteristics and trends, on a species-by-species basis, in the PRDEZ. The results indicate that PM2.5 abatement needs to prioritize secondary species.

NEXT:a neural network framework for next POI recommendation

The task of next POI recommendations has been studied extensively in recent years.However,developing a unified recommendation framework to incorporate multiple factors associated with both POIs and users remains challenging,because of the heterogeneity nature of these information.Further,effective mechanisms to smoothly handle cold-start cases are also a difficult topic.Inspired by the recent success of neural networks in many areas,in this paper,we propose a simple yet effective neural network framework,named NEXT,for next POI recommendations.NEXT is a unified framework to learn the hidden intent regarding user's next move,by incorporating different factors in a unified manner.Specifically,in NEXT,we incorporate meta-data information,e.g.,user friendship and textual descriptions of POIs,and two kinds of temporal contexts(i.e.,time interval and visit time).To leverage sequential relations and geographical influence,we propose to adopt DeepWalk,a network representation learning technique,to encode such knowledge.We evaluate the effectiveness of NEXT against other state-of-the-art alternatives and neural networks based solutions.Experimental results on three publicly available datasets demonstrate that NEXT significantly outperforms baselines in real-time next POI recommendations.Further experiments show inherent ability of NEXT in handling cold-start.

WRF/CHEM modeling of impacts of weather conditions modified by urban expansion on secondary organic aerosol formation over Pearl River Delta

In this paper, the online Weather Research and Forecasting and Chemistry （WRF/CHEM） model, coupled with urban canopy （UCM） and biogenic-emission models, is used to explore impacts of urban expansion on secondary organic aerosols （SOA） formation. Two scenarios of urban maps are used in WRF/CHEM to represent early 1990s （pre-urbanization） and current urban distribution in the Pearl River Delta （PRD）. Month-long simulation results using the above land-use scenarios for March 2001 show： （1） urbanization can increase monthly averaged temperatures by about 0.63 ℃, decrease monthly averaged 10-m wind speeds by 38%, increase monthly averaged boundary-layer depths by 80 m, and decrease monthly aver- aged water mixing ratio by 0.2g/kg. （2） Changes in meteorological conditions can result in detectable concentration changes of NOx, VOC, O3 and NO3 radicals. Urbanization decreases surface NOx and VOC concentrations by a maximum of 4 ppbv and 1.5 ppbv, respectively. Surface O3 and NO3 radical concentrations over major cities increase by about 2-4 ppbv and 4-12 pptv, respectively; areas with increasing O3 and NO3 radical concentrations generally coincide with the areas of temperature increase and wind speed reduction where NOx and VOC decrease. （3） Urbanization can induce 9% increase of SOA in Foshan, Zhongshan and west Guangzhou and 3% decrease in Shenzhen and Dongguan. Over PRD major cities, SOA from Aitken mode reduces by 30% but with more than 70% SOA from accumulate mode. Urbanization has stronger influence on SOA formation from Aitken mode. （4） Over the PRD, 55-65% SOA comes from aromatics precursors. Urbanization has strongest influence on aromatics precursors to produce SOA （14% increase）, while there is less influence on alkane precursors. Alkene precursors have negative contribution to SOA formation under urbanization situation.

Microstructure-based multiphysics modeling for semiconductor integration and packaging

Semiconductor technology and packaging is advancing rapidly toward system integration where the packaging is co-designed and co-manufactured along with the wafer fabrication.However,materials issues,in particular the mesoscale microstructure,have to date been excluded from the integrated product design cycle of electronic packaging due to the myriad of materials used and the complex nature of the material phenomena that require a multiphysics approach to describe.In the context of the materials genome initiative,we present an overview of a series of studies that aim to establish the linkages between the material microstructure and its responses by considering the multiple perspectives of the various multiphysics fields.The microstructure was predicted using thermodynamic calculations,sharp interface kinetic models,phase field,and phase field crystal modelingtechniques.Based on the predicted mesoscale microstructure,linear elastic mechanical analyses and electromigration simulations on the ultrafine interconnects were performed.The microstructural index extracted by a method based on singular value decomposition exhibits a monotonous decrease with an increase in the interconnect size.An artificial neural network-based fitting revealed a nonlinear relationship between the microstructure index and the average von Mises stress in the ultrafine interconnects.Future work to address the randomness of microstructure and the resulting scatter in the reliability is discussed in this study.

Wearable multichannel pulse condition monitoring system based on flexible pressure sensor arrays

Pulse diagnosis is an irreplaceable part of traditional Chinese medical science.However,application of the traditional pulse monitoring method was restricted in the modernization of Chinese medical science since it was difficult to capture real signals and integrate obscure feelings with a modern data platform.Herein,a novel multichannel pulse monitoring platform based on traditional Chinese medical science pulse theory and wearable electronics was proposed.The pulse sensing platform simultaneously detected pulse conditions at three pulse positions(Chi,Cun,and Guan).These signals were fitted to smooth surfaces to enable 3-dimensional pulse mapping,which vividly revealed the shape of the pulse length and width and compensated for the shortcomings of traditional single-point pulse sensors.Moreover,the pulse sensing system could measure the pulse signals from different individuals with different conditions and distinguish the differences in pulse signals.In addition,this system could provide full information on the temporal and spatial dimensions of a person’s pulse waveform,which is similar to the true feelings of doctors’fingertips.This innovative,cost-effective,easily designed pulse monitoring platform based on flexible pressure sensor arrays may provide novel applications in modernization of Chinese medical science or intelligent health care.

A one-way coupled atmospheric-hydrological modeling system with combination of high-resolution and ensemble precipitation forecasting

Coupled hydrological and atmospheric model- ing is an effective tool for providing advanced flood forecasting. However, the uncertainties in precipitation forecasts are still considerable. To address uncertainties, a one-way coupled atmospheric-hydrological modeling sys- tem, with a combination of high-resolution and ensemble precipitation forecasting, has been developed. It consists of three high-resolution single models and four sets of ensemble forecasts from the THORPEX Interactive Grande Global Ensemble database. The former provides higher forecasting accuracy, while the latter provides the range of forecasts. The combined precipitation forecasting was then implemented to drive the Chinese National Flood Forecasting System in the 2007 and 2008 Huai River flood hindcast analysis. The encouraging results demonstrated that the system can clearly give a set of forecasting hydrographs for a flood event and has a promising relative stability in discharge peaks and timing for warning purposes. It not only gives a deterministic prediction, but also generates probability forecasts. Even though the signal was not persistent until four days before the peak discharge was observed in the 2007 flood event, the visualization based on threshold exceedance provided clear and concise essential warning information at an early stage. Forecasters could better prepare for the possibility of a flood at an early stage, and then issue an actual warning if the signal strengthened. This process may provide decision support for civil protection authorities. In future studies, different weather forecasts will be assigned various weight coefficients to represent the covariance of predictors and the extremes of distributions.

Photochemical indicators of ozone sensitivity： application in the Pearl River Delta, China

Surface 03 production has a highly nonlinear relationship with its precursors. The spatial and temporal heterogeneity of O3-NOx-VOC-sensitivity regimes complicates the control- decision making. In this paper, the indicator method was used to establish the relationship between 03 sensitivity and assessment indicators. Six popular ratios indicating ozone-precursor sensitivity, HCHO/NOy, H2O2/ HNO3, O3/NOy, O3/NOz, O3/HNO3, and H2O2/NOz, were evaluated based on the distribution of NOx- and VOC-sensitive regimes. WRF-Chem was used to study a serious ozone episode in fall over the Pearl River Delta （PRD）. It was found that the south-west of the PRD is characterized by a VOCsensitive regime, while its north-east is NO y-sensitive, with a sharp transition area between the two regimes. All indicators produced good representations of the elevated ozone hours in the episode on 6 November 2009, with H202/HNO3 being the best indicator. The threshold sensitivity levels for HCHO/NOy, H2O2/HNO3, O3/NOy, O3/NOz, O3/HNO3, and H2O2/NOz were estimated to be 0.41, 0.55, 10.2, 14.0, 19.1, and 0.38, respectively. Threshold intervals for the indicators H2O2/HNO3, O3/ NOy, O3/NOz, O3/HNO3, and H202/NOz were able to identify more than 95% of VOC- and NOx-sensitive grids. The ozone episode on 16 November 16 2008 was used to independently verify the results, and it was found that only H2O2/HNO3 and H2O2/NOz were able to differentiate the ozone sensitivity regime well. Hence, these two ratios are suggested as the most appropriate indicators for identifying fall ozone sensitivity in the PRD. Since the species used for indicators have seasonal variation, the utility of those indicators for other seasons should be investigated in the future work.

Modeling Pitch Contour of Chinese Mandarin Sentences with the PENTA Model

In continuous speech, the pitch contour of the same syllable may vary much due to its contextual information. The Parallel Encoding and Target Approximation （PENTA） model is applied here to Mandarin speech synthesis with a method to predict pitch contours for Chinese syllables with different contexts by combining the Classification And Regression Tree （CART） with the PENTA model to improve its prediction accuracy. CART was first used to cluster the syllables＇ normalized pitch contours according to the syllables contextual information and the distances between pitch contours. The average pitch contour was used to train the PENTA model with the average contour for each cluster. The initial pitch is required with the PENTA model to predict a continuous pitch contour. A Pitch Discontinuity Model （PDM） was used to predict the initial pitches at positions with voiceless consonants and prosodic boundaries. Initial tests on a Chinese four-syllable word corpus containing 2048 words were extended to tests with a continuous speech corpus containing 5445 sentences. The results are satisfactory in terms of the Root Mean Square Error （RMSE） comparing the predicted pitch contour with the original contour. This method can model pitch contours for Mandarin sentences with any text for speech synthesis.

Effects of combined procedure and devascularization alone on hyperdynamics of the portal venous system in patients with portal hypertension

Shunts and devascularizations have totally dif-ferent effects on the hemodynamics of the portal venous system.The actual results of pericardial devascularization(PCDV)alone and conventional splenorenal shunt com-bined with pericardial devascularization(combined pro-cedure,CP)should be determined by more clinical observations.This study aimed to evaluate effects on hemodynamics in the portal venous system after CP and PCDV only.In 20 patients who received CP and 18 who received PCDV,hemodynamic parameters of the portal venous system were studied by magnetic resonance angio-graphy 1 week before and 2 weeks after operation.Free portal pressure(FPP)was continuously detected by a transducer during the operations.Compared to the pre-operative data,a decreased flow in the portal vein(PVF)[(563.12±206.42)mL/min vs(1080.63±352.85)mL/min,P,0.05],a decreased portal vein diameter(PVD)[(1.20±0.11)cm vs(1.30±0.16)cm,P<0.01],a decreased FPP[(21.50±2.67)mmHg vs(29.88±2.30)mmHg,P<0.01]and an increased flow in the superior mesenteric vein(SMVF)[(1105.45±309.03)mL/min vs(569.13±178.46)mL/min,P<0.05]were found in the CP group after operation;a decreased PVD[(1.27±0.16)cm vs(1.40±0.23)cm,P<0.05],a decreased PVF[(684.60±165.73)mL/min vs(1175.64±415.09)mL/min,P<0.05],a decreased FPP[(24.40±3.78)mmHg vs(28.80±3.56)mmHg,P<0.05]and an increased SMVF[(697.91±121.83)mL/min vs(521.30±115.82)mL/min,P<0.05]were observed in the PCDV group.After operation,PVF in the CP group[(563.12±206.42)mL/min vs(684.60±165.73)mL/min,P<0.05]had no significant decrease,while FPP[(21.50±2.67)mmHg vs(24.40±3.78)mmHg,P<0.01]had a significant decrease as compared with that in the PCDV group.PVF and FPP could be decreased by both surgical procedures,but the effect of decreasing FPP was much better in the combined procedure than in PCDV alone.Further,there was no significant difference in PVF between the two groups.It is suggested that the combined surgical procedure could integrate the advan-tages of shunting with those of devascularization,as well as maintaining the normal anatomic structure of hepatic portal system,thus it should be one of the best choices for patients with portal hypertension when surgical interven-tions are considered.

In-situ artificial retina with all-in-one reconfigurable photomemristor networks

Despite that in-sensor processing has been proposed to remove the latency and energy consumption during the inevitable data transfer between spatial-separated sensors,memories and processors in traditional computer vision,its hardware implementation for artificial neural networks(ANNs)with all-in-one device arrays remains a challenge,especially for organic-based ANNs.With the advantages of biocompatibility,low cost,easy fabrication and flexibility,here we implement a self-powered in-sensor ANN using molecular ferroelectric(MF)-based photomemristor arrays.Tunable ferroelectric depolarization was intentionally introduced into the ANN,which enables reconfigurable conductance and photoresponse.Treating photoresponsivity as synaptic weight,the MFbased in-sensor ANN can operate analog convolutional computation,and successfully conduct perception and recognition of white-light letter images in experiments,with low processing energy consumption.Handwritten Chinese digits are also recognized and regressed by a large-scale array,demonstrating its scalability and potential for low-power processing and the applications in MF-based in-situ artificial retina.