Verification Of Accuracy Research Articles

Vibration isolation of infinitely periodic pile barriers for anti-plane shear waves: An exact series solution

The application of periodic pile barriers for isolating ambient vibrations has emerged as a significant research topic, which is essentially the problem of wave scattering by periodically distributed structures. This study presents an analytical series solution for anti-plane shear (SH) wave scattering by infinitely periodic pile barriers using the wave function expansion method, incorporating the wave field characteristics of infinitely periodic distributed scatterers subjected to plane waves. From the perspective of periodic theory, the wave fields around each scatterer are completely identical, differing only by a phase shift (frequency domain) or a time difference (time domain). As a result, as long as the boundary condition of one scatterer is satisfied, boundary conditions of the remaining scatterers can be satisfied simultaneously. This novel approach allows the selection of an arbitrary single pile as a reference in the infinitely periodic model to be solved, yielding accurate results without truncation errors and significantly reducing computational and storage requirements. A FORTRAN code is developed to achieve numerical evaluation of the theoretical formulation, followed by verification of the solution's accuracy using two degenerate examples. Numerical examples are conducted, focusing on the influence of pile stiffness, pile spacing, and pile type on the vibration isolation. The results demonstrate that decreasing the spacing between piles can effectively increase the bandgap width and reduce the displacement response. The screening effectiveness of solid piles is better than that of hollow pipe piles and filled pipe piles. In vibration mitigation design for engineering projects, the larger the shear stiffness ratio of the pile-soil is not always better, an economical and reasonable pile stiffness should be selected.

Influence of minimum segment width on intensity-modulated radiotherapy plan for left-sided breast cancer after breast conserving surgery.

To research the influence of the minimum segment width (MSW) on intensity-modulated radiotherapy (IMRT) plan for left breast cancer after breast conserving surgery and provide a reference for plan optimization. Ten patients with left-sided early-stage breast cancer after breast-conserving surgery (BCS) were selected for postoperative radiotherapy. The Monaco 5.11 treatment planning system (TPS) was used, and the optimization parameters of the plan were fixed, while MSW were set to 0.5, 1, 1.5, and 2 cm, and four 5 field-IMRT plans were designed for every patient. The plan quality and execution efficiency of each plan were compared, including planning target volume (PTV) dose, organs at risk (OAR) dose, segments number (SN), monitor units (MU), delivery time (DT), and dose verification results. With the increase of the MSW, the dose indexes of the OAR were similar (P > .05). The D2%, D98%, and D50% of the PTV gradually deviated from the prescription dose, and the uniformity gradually deteriorated (P < .05). The gamma passing rates (GPR) of dose verification and the dose deviation of the PTV were similar (P > .05). The SN, MU, and DT gradually decreased (P < .05), but the amplitude change caused by each increase of 0.5 cm gradually decreased, the decrease of 0.5 to 1 cm was the largest, reaching 16.41%, 21.25%, and 0.35 minutes, respectively, while the 3 values of 1.5 and 2 cm were close. When using Monaco 5.11 TPS to design IMRT plan for left breast cancer after BCS, 1.5 cm MSW could ensure better dose coverage and uniformity of the PTV, as little as possible exposure dose to the OAR, and high dose verification accuracy. The SN, MU, and DT were significantly reduced, and the treatment efficiency was improved.

Free Vibration Analysis of Arches with Interval-Uncertain Parameters

The dynamic characteristics of a structure serve as a crucial foundation for structural assessment, fault diagnosis, and ensuring structural safety. Therefore, it is imperative to investigate the impacts of uncertain parameters on the dynamic performance of structures. The dynamic characteristics of arches with uncertain parameters are analyzed in this paper. The uncertain parameters are regarded as non-probabilistic uncertainties and represented as interval variables. A model of an arch considering interval-uncertain parameters is built, and kinematic equations are established. The natural frequencies are obtained using the differential quadrature (DQ) method, and the relationships between natural frequency, radius, and central angle are also analyzed. On this basis, the Chebyshev polynomial surrogate (CPS) model is employed to solve the uncertain dynamic problem, and the natural frequencies are seen to be the objective functions of the CPS model. The accuracy verification of the model is achieved by comparison with the Monte Carlo simulation (MCS). Simulations are carried out considering different uncertainties, and the results show that the bounds of natural frequencies are influenced not only by the types of uncertain parameters, but also by their combinations.

BIOS-06. MACHINE LEARNING MADE EASY: DESIGN OF A PROPOSAL FOR A MEDICAL STUDENT EDUCATIONAL SESSION ON HIGH YIELD STRATEGIES FOR LEARNING AND APPLYING DATA SCIENCE

Abstract Machine learning is a burgeoning field in data science that has many applications in medicine, and it is especially useful in neurosurgery due to the often-complex and potentially emergent nature of day-to-day cases. Machine learning models have the potential to deliver accurate patient prognostication while offering the benefit of providing real-time AI supported clinical decision-making with high data granularity. These models can utilize numerical inputs for more straightforward applications and, with practice, can enable clinicians and trainees to incorporate advanced imaging sequences as inputs to optimize the accuracy of patient prognostication. In the future, these models will become much more commonplace in the clinical setting for neurosurgeons. As medical school curricula do not typically incorporate these statistical techniques into biomedical statistics courses, symposia centered on machine learning could prove essential to medical students seeking to learn these skills. This session would include a brief overview of machine learning models and the procedures for using the programming language of RStudio for the development of novel internet-based prognostication calculators. The benefit of RStudio is that it is user friendly and pre-designed templates are available that can enable students to perform facile customization in tailoring the code to the needs of their clinical investigation. There will be a brief overview of accuracy verification involving discussion of receiver operating curves with the use of training and testing datasets - the heart of machine learning. We have designed and outlined a curriculum that will enable students to identify and classify different techniques (support vector machines, logistic regression, clustering, nearest neighbor classifiers), and recognize the utility of machine learning approaches to neurosurgical outcomes prognostication. Our hope is that students will leave this symposium well-equipped with a newfound confidence and the baseline conceptual framework needed to begin exploring the awesome world of machine learning throughout their training.

Development of efficient techniques for ASR System for Speech Detection and Recognization system using Gaussian Mixture Model- Universal Background Model

Some practical uses of ASR have been implemented, including the transcription of meetings and the usage of smart speakers. It is the process by which speech waves are transformed into text that allows computers to interpret and act upon human speech. Scalable strategies for developing ASR systems in languages where no voice transcriptions or pronunciation dictionaries exist are the primary focus of this work. We first show that the necessity for voice transcription into the target language can be greatly reduced through cross-lingual acoustic model transfer when phonemic pronunciation lexicons exist in the new language. Afterwards, we investigate three approaches to dealing with languages that lack a pronunciation lexicon. Secondly, we have a look at the efficiency of graphemic acoustic model transfer, which makes it easy to build pronunciation dictionaries. Thesis problems can be solved, in part, by investigating optimization strategies for training on huge corpora (such as GA+HMM and DE+HMM). In the training phase of acoustic modelling, the suggested method is applied to traditional methods. Read speech and HMI voice experiments indicated that while each data augmentation strategy alone did not always increase recognition performance, using all three techniques together did. Power normalised cepstral coefficient (PNCC) features are tweaked somewhat in this work to enhance verification accuracy. To increase speaker verification accuracy, we suggest employing multiple “Gaussian Mixture Model-Universal Background Model (GMM-UBM) and SVM classifiers”. Importantly, pitch shift data augmentation and multi-task training reduced bias by more than 18% absolute compared to the baseline system for read speech, and applying all three data augmentation techniques during fine tuning reduced bias by more than 7% for HMI speech, while increasing recognition accuracy of both native and non-native Dutch speech.

FE modeling and simulation of the turning process considering the cutting induced hardening of workpiece materials

The accuracy of the cutting simulation model greatly depends on the constitutive models, thermophysical models, and friction models. However, accurate modeling of physical and mechanical relationships is not enough. The physical and mechanical behavior of the machined surface from the last cut should be modelled in the FE model. In this study, the cutting simulation model of S316L stainless steel was established. The above model consists of two subsequent simulated cuts. The first simulated cut was used to obtain the machined surface with the residual stress, and the second simulated cut was subsequent with the first cut to obtain the actual simulated results. The constitutive model was obtained by the split Hopkinson pressure bar (SHPB) and high-temperature hardness experiments. The specific heat capacity and thermal conductivity models were developed by laser thermal conductivity experiments with various temperatures. The friction model between the workpiece and the tool was established by orthogonal cutting experiments. The simulated cutting forces of the first and second cut were extracted and compared with the experimental results to verify the accuracy of the simulation models. The results showed that the average error of cutting forces for the first cut is 28.33 %, but that for the second cut is 8.02 %, which verifies the accuracy of the two-subsequent cutting simulation model. Additionally, the significant differences in the simulated cutting forces between the first and second cutting depict that the residual stress cannot be ignored for the accuracy verification of cutting simulation models.

Development of decision-making technique based on sentiment analysis of crowdsourcing data in medical social media resources

The object of the study is the decision-making modeling in the context of medical social media to increase the clinics’ effectiveness. The problem is to classify the patient reviews collected in the patient-clinic segment of the medical social media and to identify the situation related to the clinics’ activity by revealing the criteria characterizing the clinics’ activity out of the opinions. The proposed technique refers to lexicon-based sentiment analysis of opinions, the classification based on Valence Aware Dictionary and Sentiment Reasoner (VADER), the verification of the results accuracy with Multinomial Naive Bayes and Support Vector Machine, the manual sentiment analysis of opinions to detect criteria and the classification of opinions according to each criterion. Using this technique, out of 442587 patient reviews obtained from database cms_hospital_satisfaction_2020 of the Kaggle company generated on the basis of crowdsourcing of patient reviews on medical social media, 218914 patient reviews are classified as positive, 190360 – as neutral, and 33313 – as negative. The results accuracy is verified, and the clinics are rated by the «positive» opinions. 6 new criteria characterizing the clinics’ activity are discovered, and the identification of the situation related to the clinics’ activity based on the comparison of «positive» and «negative» opinions according to each criterion is presented. The possibility of using the results obtained from the identification to increase the clinics’ efficiency in making decisions is shown. The results obtained in this study can be used to improve the clinics’ performance according to public opinion. This opportunity involves the crowdsourcing of opinions about the clinic in the medical social media environment and the collection of opinions in a structured way.

Out-of-plane moving load response and vibrational behavior of sandwich curved beams with GPLRC face sheets and porous core

As a first endeavor, the out-of-plane free vibrational behaviors and moving load responses of sandwich curved beams with graphene platelets reinforced composite face sheets and porous core (GPLRC-FS-PC) are carried out. The governing equations are derived based on the first-order shear deformation theory (FSDT) using Hamilton’s principle, which are discretized by employing the differential quadrature method (DQM) and Newmark’s method in the spatial and time domains, respectively. To simulate the moving load using the DQM, the Heaviside function approach is utilized. The effective elastic properties of face sheets are estimated using Halpin-Tsai micromechanical model. The approach is validated by presenting convergence studies and accuracy verification followed by parametric studies. The numerical results indicate that by adding little amount of graphene platelets (GPLs) into the face sheets and core layer, the fundamental natural frequency and the displacement amplitudes under moving load significantly increases and decreases, respectively, regardless of the beam boundary conditions. It is revealed that the GPLs distribution patterns in face sheets and the core porosity distribution patterns affect the critical load velocities of the GPLRC-FS-PC sandwich curved beams. Also, it is shown that the frequency parameters strongly depend on the face sheet to core thickness ratio.

Crop Type Identification Using High-Resolution Remote Sensing Images Based on an Improved DeepLabV3+ Network

Remote sensing technology has become a popular tool for crop classification, but it faces challenges in accurately identifying crops in areas with fragmented land plots and complex planting structures. To address this issue, we propose an improved method for crop identification in high-resolution remote sensing images, achieved by modifying the DeepLab V3+ semantic segmentation network. In this paper, the typical crop area in the Jianghuai watershed is taken as the experimental area, and Gaofen-2 satellite images with high spatial resolutions are used as the data source. Based on the original DeepLab V3+ model, CI and OSAVI vegetation indices are added to the input layers, and MobileNet V2 is used as the backbone network. Meanwhile, the upper sampling layer of the network is added, and the attention mechanism is added to the ASPP and the upper sampling layers. The accuracy verification of the identification results shows that the MIoU and PA of this model in the test set reach 85.63% and 95.30%, the IoU and F1_Score of wheat are 93.76% and 96.78%, and the IoU and F1_Score of rape are 74.24% and 85.51%, respectively. The identification accuracy of this model is significantly better than that of the original DeepLab V3+ model and other related models. The proposed method in this paper can accurately extract the distribution information of wheat and rape from high-resolution remote sensing images. This provides a new technical approach for the application of high-resolution remote sensing images in identifying wheat and rape.

Online verification and management scheme of gateway meter flow in the power system by machine learning

Currently, the calibration of electric energy meters often involves manual meter reading, dismantling inspection, or regular sampling inspection conducted by professionals. To improve work efficiency and verification accuracy, this research integrates machine learning into the scheme of online verification and management of gateway meter flow in the power system. The approach begins by applying the Faster Region Convolutional Neural Network (Faster-RCNN) model and the Single Shot MultiBox Detector (SSD) model to the recognition system for dial readings. Then, the collected measurement data is pre-processed, excluding data collected under light load conditions. Next, an estimation error model and a solution equation for the electricity meter are established based on the pre-processed data. The operation error of the electricity meter is estimated, and the estimation accuracy is verified using the limited memory recursive least squares algorithm (LMRLSA). Furthermore, business assistant decision-making is carried out by combining the remote verification results with the estimation outcomes. The proposed dial reading recognition system is tested using 528 images of meter readings, achieving an accuracy of 98.49%. In addition, the influence of various parameters on the error results of the electricity meter is also explored. The results demonstrate that a memory length ranging from 600 to 1,200 and a line loss error of less than 5% yield the most suitable accuracy for estimating the electricity meter error. Meanwhile, it is advisable to remove measurement data collected under light load to avoid unnecessary checks. The experiments manifest that the proposed algorithm can properly eliminate the influence of old measurement data on the error parameter estimation, thereby enhancing the accuracy of the estimation. The adjustment of the memory length ensures real-time performance in estimating meter errors and enables online monitoring. This research has certain reference significance for achieving the online verification and management of gateway meter flow in the power system.

Predicting houses price by deep learning neural network

Deep learning is widely used in various fields, the article proposed a deep learning method to predict house prices through different characteristics of real estate, establish a prediction model, and carry out simulation experiments. First, extracting data from property transactions records, it is difficult to directly input the raw-data into the deep learning model, and there may be overfitting in the model training, so data will be pretreated. Second, Fully-Connected Neural Network is used to model different features influences to price. The sample data will be randomly divided into a training set and a validation set, of which 70% of the samples are used for building and training the model, and the remaining 30% are used for model accuracy verification. Experimental results show that the model can achieve a high accuracy in predicting houses price. The model can be used as a reference for the evaluation of housing prices.

The use of non-invasive stool tests for verification of Helicobacter pylori eradication and clarithromycin resistance.

Clarithromycin resistance of Helicobacter pylori (H. pylori) represents a major challenge in eradication therapy. In this study, we assessed if non-invasive stool tests can be used to verify successful H. pylori eradication and determine clarithromycin resistance. In this prospective study, patients undergoing urea breath testing (UBT) for confirmation of H. pylori eradication were asked to collect the stool as both a dry fecal sample and fecal immunochemical test (FIT). Stool H. pylori antigen testing (SAT) was performed on these samples and assessed for its accuracy in eradication verification. Type and duration of antibiotic treatment were retrospectively collected from patient records and compared with clarithromycin resistance determined by PCR of stool samples. H. pylori eradication information was available for a total of 145 patients (42.7% male, median age: 51.2). Successful eradication was achieved in 68.1% of patients. SAT on FIT samples had similar accuracy for eradication assessment compared to dry fecal samples, 72.1% [95% CI 61.4-81.2] versus 72.2% [95% CI 60.9-81.7]. Clarithromycin resistance rate was 13.4%. H. pylori antigen testing on FIT stool samples to verify H. pylori eradication is feasible and has similar accuracy as H. pylori antigen testing on dry stool samples. Dry stool, but not FIT, was suitable for non-invasive identification of H. pylori clarithromycin resistance by rt-PCR personalizing antibiotic treatment strategies without the need for invasive diagnostics is desirable, as the cure rate of first-line empirical H. pylori treatment remains low.

Quality control of Ganoderma lucidum by using C, H, O, and N stable isotopes and C and N contents for geographical traceability.

Ganoderma lucidum (G. lucidum) is a popular medicinal fungus that has been used in traditional medicine for decades, with its provenance influencing its medicinal and commercial worth. The amount of active ingredients and the price of G. lucidum from different origins vary significantly; hence, fraudulent labeling is common. Reliable techniques for G. lucidum geographic verification are urgently required to safeguard the interests of consumers, producers, and honest dealers. A stable isotope is widely acknowledged as a useful traceability technique and could be developed to confirm the geographical origin of G. lucidum. G. lucidum samples from various sources and in varying stages were identified by using δ 13C, δD, δ 18O, δ 15N, C, and N contents combined with chemometric tools. Chemometric approaches, including PCA, OPLS-DA, PLS, and FLDA models, were applied to the obtained data. The established models were used to trace the origin of G. lucidum from various sources or track various stages of G. lucidum. In the stage model, the δ 13C, δD, δ 18O, δ 15N, C, and N contents were considered meaningful variables to identify various stages of G. lucidum (bud development, growth, and maturing) using PCA and OPLS-DA and the findings were validated by the PLS model rather than by only four variables (δ 13C, δD, δ 18O, and δ 15N). In the origin model, only four variables, namely δ 13C, δD, δ 18O, and δ 15N, were used. PCA divided G. lucidum samples into four clusters: A (Zhejiang), B (Anhui), C (Jilin), and D (Fujian). The OPLS-DA model could be used to classify the origin of G. lucidum. The model was validated by other test samples (Pseudostellaria heterophylla), and the external test (G. lucidum) by PLS and FLDA models demonstrated external verification accuracy of up to 100%. C, H, O, and N stable isotopes and C and N contents combined with chemometric techniques demonstrated considerable potential in the geographic authentication of G. lucidum, providing a promising method to identify stages of G. lucidum.

Accuracy verification of fast adaptive galerkin finite volume solver for thermal and mechanical load crack propagation in triangular meshes

This study presents the details of the Galerkin finite volume method for triangular unstructured meshes and its application for the fracture analysis of 2D elasticity problems under simultaneous thermal and mechanical loading. The heat diffusion and elasticity equations are discretized using the Galerkin finite volume method (which, due to omitting matrix manipulation calculations, computations are considerably faster than FEM and XFEM Solvers). The maximum tangential stress criterion is presented after briefly describing the interaction integral formulation used to calculate the stress intensity factors in thermo-mechanical problems. The accuracy of the computed results of the present strategy under thermo-mechanical loads is demonstrated using some benchmark test cases. First, a thermal stress analysis of a plate with an inclined central crack under thermal boundary conditions is performed. Second, the development of a crack under mechanical stress boundary conditions is modeled. Third, crack propagation under both thermal and mechanical boundary conditions is simulated. The present modeling strategy's results are compared with those reported in previous numerical works to verify the accuracy. The stress intensity factors and predicted crack trajectories are utilized to assess the accuracy of computed results and investigate the quality of crack simulation by the proposed numerical method.

Susceptibility Assessment of Landslides in the Loess Plateau Based on Machine Learning Models: A Case Study of Xining City

Landslide susceptibility assessment can effectively predict the spatial distribution of potential landslides, which is of great significance in fields such as geological disaster prevention, urban planning, etc. Taking Xining City as an example, based on GF-2 remote sensing image data and combined with field survey data, this study delineated the spatial distribution range of developed landslides. Key factors controlling landslides were then extracted to establish a landslide susceptibility assessment index system. Based on this, the frequency ratio (FR), random forest (RF), support vector machine (SVM), and artificial neural network (ANN) models were applied to spatially predict landslide susceptibility with slope units as the basis. The main results are as follows: (1) The overall spatial distribution of landslide susceptibility classes in Xining City is consistent, but the differences between different landslide susceptibility classes are significant. (2) The high-susceptibility area predicted by the FR-RF model is the largest, accounting for 15.48% of the total study area. The prediction results of the FR-ANN and FR-SVM models are more similar, with high-susceptibility areas accounting for 13.96% and 12.97%, respectively. (3) The accuracy verification results show that all three coupled models have good spatial prediction capabilities in the study area. The order of landslide susceptibility prediction capabilities from high to low is FR-RF model &gt; FR-ANN model &gt; FR-SVM model. This indicates that in the study area, the FR-RF model is more suitable for carrying out landslide susceptibility assessment.

Influence of tree-based multi-layer node information on scoring accuracy and speed of speaker verification

In the speaker verification task based on Gaussian Mixture Model-Universal Background Model (GMM-UBM), by constructing the UBM as a tree structure, the kernel Gaussians suitable for different speakers can be quickly selected, which speeds up the modeling of speaker acoustic space by GMM. The tree-based kernel selection algorithm (TBKS) introduces a beam-width, which increases the candidate range of kernels and improves the kernel selection accuracy. In this paper, we improve the TBKS algorithm by introducing a recall rate to adjust the number of nodes recalled in each layer of the tree structure. This adjustment refines the quantity and resolution of Gaussian distributions in various subspaces within the acoustic space, compensating for the loss caused by discarding some significant Gaussians erroneously. Speaker verification experiments are carried out based on the Aishell2 dataset. The results reveal that the modified TBKS algorithm reduces EER by 7.5% relatively and increses computational reduction factor to 42.93, enhancing both recognition accuracy and speed. In addition, the test speech is spliced into different lengths and common environmental noise is added to verify the universality of the improved algorithm.

Development of a Sentence Verification Task to Measure Speech Comprehension and Listening Effort in Cochlear Implant Users.

Tests measuring speech comprehension and listening effort for cochlear implant (CI) users may reflect important aspects of real-world speech communication. In this study, we describe the development of a multiple-talker, English-language sentence verification task (SVT) for use in adult CI users to measure speech comprehension and listening effort. We also examine whether talker differences affect speech comprehension and listening effort. Thirteen experienced adult CI users participated in the study and underwent testing using a newly developed multiple-talker SVT. Participants were sequentially presented with audio recordings of unique sentences spoken in English by six different talkers. Participants classified each sentence as either true or false. Accuracy of classification and the response time (RT) for correct responses were used as measures of comprehension and listening effort, respectively. The effect of talker on the results was further analyzed. All 13 participants successfully completed the SVT. The mean verification accuracy for participants was 87.2% ± 8.8%. The mean RT for correct responses across participants was 1,050 ms ± 391 ms. When stratified by talker, verification accuracy ranged from 83.7% to 95.2% and mean RTs across participant ranged from 786 ms to 1,254 ms. Talker did not have a significant effect on sentence classification accuracy, but it did have a significant effect on RTs (p < .001). The SVT is an easily implemented test that can assess speech comprehension and listening effort simultaneously. CI users may experience increased effort for comprehending certain talkers' speech, despite showing similar levels of comprehension accuracy. https://doi.org/10.23641/asha.24126630.

Distribution of local curvature values as a structural feature for off-line handwritten signature verification

In the paper, a new feature for describing a digital image of a handwritten signature based on the frequency distribution of the values of the local curvature of the signature contours, is proposed. The calculation of this feature on the binary image of a signature is described in detail. A normalized histogram of distributions of local curvature values for 40 bins is formed. The frequency values recorded as a 40-dimensional vector are called the local curvature code of the signature.During verification, the proximity of signature pairs is determined by correlation between curvature codes and LBP codes described by the authors in [23]. To perform the signature verification procedure, a two-dimensional feature space is constructed containing images of the proximity of signature pairs. When verifying a signature with N authentic signatures of the same person, N(N-1)/2 patterns of the proximity of pairs of genuine signatures and N images of pairs of proximity of the analyzed signature with genuine signatures are presented in the feature space. The Support Vector Machine (SVM) is used as a classifier.Experimental studies were carried out on digitized images of genuine and fake signatures from two databases. The accuracy of automatic verification of signatures on the publicly available CEDAR database was 99,77 % and on TUIT was 88,62 %.

A dataset of GDP spatial distribution on the Qinghai-Tibet Plateau (2020)

Economic spatial distribution data serve as an indicator for disaster loss assessment. The economy of the Qinghai-Tibet Plateau is relatively backward and geological disasters occur frequently, so it is of great significance to study the impact of disasters on the economy in this area. Based on the method of zoning and industry simulation, this study establishes the corresponding relationship between the GDP1（GDP of Primary Industry）and land use types, DEM（Digital Elevation Model）, village points, roads and river buffers. Spatial models of night lighting, building site and points of interest data after the coupling of the GDP23（GDP of Secondary and Tertiary Industries）and multiple information are established in urban areas and counties. We used the random forest to determine the weights of various indicators in the secondary and tertiary industries, and calculated the grid value of GDP23. Finally, we superimposed the grid of GDP1 and GDP23 and created the spatial distribution map of GDP（Gross Domestic Product）of 100 m×100 m on the Qinghai-Tibet Plateau in 2020. According to the accuracy verification using county-level GDP statistical data, the Root Mean Square Error is 0.0385, which shows that the data can better reflect the spatial distribution of GDP on the Qinghai-Tibet Plateau. This dataset can provide GDP spatial data support for disaster risk assessment, disaster prevention and mitigation on the Qinghai-Tibet Plateau.

Hybrid-Supervised Deep Learning for Proton-Acoustic Reconstruction for 3D In Vivo Proton Dose Verification

Proton-acoustic (PA) image has shown great potential to provide real-time 3D dose verification of proton therapy. However, the PA image quality suffers from severe limited view artifacts, which significantly impairs its accuracy for dose verification. In this study, we developed a hybrid-supervised deep learning method for PA reconstruction to address the limited-view issues. Our method consists of two stages. In the first stage, a transformer-based network was proposed to reconstruct initial pressure maps from protoacoustic signals. The network was first trained using supervision by the iteratively reconstructed pressure map and then fine-tuned using transfer learning and self-supervision based on the data fidelity constraint. In the second stage, the PA image was further enhanced by a 3D U-net. The final PA images were converted to dose maps using conversion coefficients derived from CT images. Data from 126 prostate cancer patients treated by proton therapy were collected under an IRB protocol and were split into 86 and 40 patients for model training and testing, respectively. Data of each patient contains the planning CT scan, the corresponding clinical treatment plan, and the dose map calculated by commercial software. The radiofrequency signals were generated by performing proton acoustic simulation based on CT images and the ground truth pressure map derived from the treatment plan. An ultrasound detector matrix with 64 × 64 size and 500kHz central frequency was simulated under the perineum to acquire the signals in the prostate area. In the testing results, the method's accuracy was evaluated using Root-mean-squared-error (RMSE) and structural-similarity-index-measure (SSIM) between the reconstructed and ground truth pressure map and dose distribution. Testing results showed that the reconstructed pressure map achieved an average RMSE/SSIM of 0.0292/0.96, demonstrating excellent 3D information with details. Dose maps derived from the pressure map achieved an average RMSE/SSIM of 0.018/0.99 with a gamma index of 94.7% and 95.7% for 1%/3 mm and 1%/5 mm criteria compared to the ground truth dose maps. The reconstruction time was 6s, which can be further reduced using GPU. Our study achieves start-of-the-art performance in the challenging task of direct reconstruction from limited-view radiofrequency signals, demonstrating the great promise of PA imaging as a highly efficient and accurate tool for in-vivo 3D proton dose verification. Such high-precision 3D online dose verification can substantially reduce the range uncertainties of proton therapy to significantly improve its precision and outcomes.