High-quality Training Set Research Articles

DeSide: A unified deep learning approach for cellular deconvolution of tumor microenvironment

Cellular deconvolution via bulk RNA sequencing (RNA-seq) presents a cost-effective and efficient alternative to experimental methods such as flow cytometry and single-cell RNA-seq (scRNA-seq) for analyzing the complex cellular composition of tumor microenvironments. Despite challenges due to heterogeneity within and among tumors, our innovative deep learning-based approach, DeSide, shows exceptional accuracy in estimating the proportions of 16 distinct cell types and subtypes within solid tumors. DeSide integrates biological pathways and assesses noncancerous cell types first, effectively sidestepping the issue of highly variable gene expression profiles (GEPs) associated with cancer cells. By leveraging scRNA-seq data from six cancer types and 185 cancer cell lines across 22 cancer types as references, our method introduces distinctive sampling and filtering techniques to generate a high-quality training set that closely replicates real tumor GEPs, based on The Cancer Genome Atlas (TCGA) bulk RNA-seq data. With this model and high-quality training set, DeSide outperforms existing methods in estimating tumor purity and the proportions of noncancerous cells within solid tumors. Our model precisely predicts cellular compositions across 19 cancer types from TCGA and proves its effectiveness with multiple additional external datasets. Crucially, DeSide enables the identification and analysis of combinatorial cell type pairs, facilitating the stratification of cancer patients into prognostically significant groups. This approach not only provides deeper insights into the dynamics of tumor biology but also highlights potential therapeutic targets by underscoring the importance of specific cell type or subtype interactions.

Internal Thread Defect Generation Algorithm and Detection System Based on Generative Adversarial Networks and You Only Look Once.

In the field of industrial inspection, accurate detection of thread quality is crucial for ensuring mechanical performance. Existing machine-vision-based methods for internal thread defect detection often face challenges in efficient detection and sufficient model training samples due to the influence of mechanical geometric features. This paper introduces a novel image acquisition structure, proposes a data augmentation algorithm based on Generative Adversarial Networks (GANs) to effectively construct high-quality training sets, and employs a YOLO algorithm to achieve internal thread defect detection. Through multi-metric evaluation and comparison with external threads, high-similarity internal thread image generation is achieved. The detection accuracy for internal and external threads reached 94.27% and 93.92%, respectively, effectively detecting internal thread defects.

Hybrid deep learning for design of nanophotonic quantum emitter lenses

Inverse design of nanophotonic structures has allowed unprecedented control over light. These design processes however are accompanied with challenges, such as their high sensitivity to initial conditions, computational expense, and complexity in integrating multiple design constraints. Machine learning approaches, however, show complementary strengths, allowing huge sample sets to be generated nearly instantaneously, and with transfer learning, allowing modifications in design parameters to be integrated with limited retraining. Herein we investigate a hybrid deep learning approach, leveraging the accuracy and performance of adjoint-based topology optimization to produce a high-quality training set for a convolutional generative network. We specifically explore this in the context of 3D nanophotonic lenses, used for focusing light between plane-waves and single-point, single-wavelength sources such as quantum emitters. We demonstrate that this combined approach allows higher performance than adjoint optimization alone when additional design constraints are applied; can generate large datasets (which further allows faster iterative training to be performed); and can utilize transfer learning to be retrained on new design parameters with very few new training samples. This process can be used for general nanophotonic design, and is particularly beneficial when a range of design parameters and constraints would need to be applied.

Semantic-Oriented Visual Prompt Learning for Diabetic Retinopathy Grading on Fundus Images.

Diabetic retinopathy (DR) is a serious ocular condition that requires effective monitoring and treatment by ophthalmologists. However, constructing a reliable DR grading model remains a challenging and costly task, heavily reliant on high-quality training sets and adequate hardware resources. In this paper, we investigate the knowledge transferability of large-scale pre-trained models (LPMs) to fundus images based on prompt learning to construct a DR grading model efficiently. Unlike full-tuning which fine-tunes all parameters of LPMs, prompt learning only involves a minimal number of additional learnable parameters while achieving a competitive effect as full-tuning. Inspired by visual prompt tuning, we propose Semantic-oriented Visual Prompt Learning (SVPL) to enhance the semantic perception ability for better extracting task-specific knowledge from LPMs, without any additional annotations. Specifically, SVPL assigns a group of learnable prompts for each DR level to fit the complex pathological manifestations and then aligns each prompt group to task-specific semantic space via a contrastive group alignment (CGA) module. We also propose a plug-and-play adapter module, Hierarchical Semantic Delivery (HSD), which allows the semantic transition of prompt groups from shallow to deep layers to facilitate efficient knowledge mining and model convergence. Our extensive experiments on three public DR grading datasets demonstrate that SVPL achieves superior results compared to other transfer tuning and DR grading methods. Further analysis suggests that the generalized knowledge from LPMs is advantageous for constructing the DR grading model on fundus images.

EPI-SF: essential protein identification in protein interaction networks using sequence features.

Proteins are considered indispensable for facilitating an organism's viability, reproductive capabilities, and other fundamental physiological functions. Conventional biological assays are characterized by prolonged duration, extensive labor requirements, and financial expenses in order to identify essential proteins. Therefore, it is widely accepted that employing computational methods is the most expeditious and effective approach to successfully discerning essential proteins. Despite being a popular choice in machine learning (ML) applications, the deep learning (DL) method is not suggested for this specific research work based on sequence features due to the restricted availability of high-quality training sets of positive and negative samples. However, some DL works on limited availability of data are also executed at recent times which will be our future scope of work. Conventional ML techniques are thus utilized in this work due to their superior performance compared to DL methodologies. In consideration of the aforementioned, a technique called EPI-SF is proposed here, which employs ML to identify essential proteins within the protein-protein interaction network (PPIN). The protein sequence is the primary determinant of protein structure and function. So, initially, relevant protein sequence features are extracted from the proteins within the PPIN. These features are subsequently utilized as input for various machine learning models, including XGB Boost Classifier, AdaBoost Classifier, logistic regression (LR), support vector classification (SVM), Decision Tree model (DT), Random Forest model (RF), and Naïve Bayes model (NB). The objective is to detect the essential proteins within the PPIN. The primary investigation conducted on yeast examined the performance of various ML models for yeast PPIN. Among these models, the RF model technique had the highest level of effectiveness, as indicated by its precision, recall, F1-score, and AUC values of 0.703, 0.720, 0.711, and 0.745, respectively. It is also found to be better in performance when compared to the other state-of-arts based on traditional centrality like betweenness centrality (BC), closeness centrality (CC), etc. and deep learning methods as well like DeepEP, as emphasized in the result section. As a result of its favorable performance, EPI-SF is later employed for the prediction of novel essential proteins inside the human PPIN. Due to the tendency of viruses to selectively target essential proteins involved in the transmission of diseases within human PPIN, investigations are conducted to assess the probable involvement of these proteins in COVID-19 and other related severe diseases.

Machine Learning (ML)-Enabled Automation for High-Throughput Data Processing in Flow Cytometry

Introduction Flow cytometry is widely used in clinical and research laboratories for diagnostics, biomarker discovery, and immune system monitoring. Flow cytometry data processing still uses gating- and clustering-based approaches that are highly time-consuming and subjective. Data processing time increases with panel size and number of detected populations, posing challenges to the search for new biomarkers. Low reproducibility and method limitations have thus far hindered efforts to automate and standardize flow cytometry data processing; hence, these efforts have not yielded any significant advancements in data processing methods. Here we present a new ML-based algorithm for automated cell-type labeling. Our supervised ML approach allows us to classify every event in a flow cytometry data file solely based on the presence and absence of markers, without the need for prior knowledge or assumption about cell population content in the sample. This approach enables the detection of rare and/or new cell populations with a high average quality metric (f1-score). The rapid and high-quality analysis our algorithm can perform renders it applicable in clinical settings, particularly for detecting hematological abnormalities and cancers. Methods We processed 500 blood samples from a cohort of healthy donors and patients with various cancer diagnoses using 10 different 18-channel multicolor flow cytometry panels. We then used data from either the entire or a portion of these 500 samples in a 3:1 split for training:test datasets to train and test our algorithm on each cytometry panel. To do this, we manually matched cells with certain cellular phenotypes to create 10 high-quality training sets for supervised learning and 10 test datasets, one pair for each of the 10 panels. To train the cell type classifier, we set up a two-level boosting-based model. The first-level model filters out outliers, including dead cells, cellular debris, beads, and other undefined particles, in order to hone in on the target population. The second-level model for predicting cell types within a target population is defined by two approaches. The population-based approach detects major subpopulation types in a target population and predicts the precise population labels. This approach is useful for labeling a small number of previously known or predicted subpopulations. The marker-based approach is useful for target populations with large numbers of subpopulations, such as T cells harboring different combinations of cell-surface receptors. It predicts the presence or absence of specific markers on each cell to assign its phenotype. It also allows us to construct complex hierarchies in order to detect new populations that are challenging to identify manually. Figure 1 outlines our workflow. Results We validated our final set of 10 trained models on our test dataset. The summarized number of detected cell populations in the test dataset was 221, which corresponds to the number of unique cell types predicted by our models. Table 1 shows the evaluation metrics for our algorithm for populations with > 0.1% whole blood cells (WBCs).The average quality metric (f1-score) for all antibody panels used is 0.86. This value is the mean of all f1-scores calculated for all cell populations identified by our algorithm. Mean f1-score is the highest (0.96) for large populations, lower (0.87) for mid-sized populations, and lowest but acceptable (0.77) for small populations. Mean quality score for the marker-based models is also high (0.96). Compared to manual evaluation that took approximately 1 hour to analyze one data file, the algorithm completed analysis within 10 seconds. Conclusion Our new algorithm automates cell labeling and produces high-quality outputs that are comparable to manual processing, but with a much shorter turnaround time (TAT) and without the need for prior knowledge or expert competence from the user. Importantly, it allows us to effectively and accurately filter out outliers, identify the target population, and divide this target population into multiple cell subtypes including new and rare cell subpopulations, all without a priori assumptions about cell population content in the sample. Given its ability to perform high-quality cell population analysis and its short TAT, our algorithm provides rapid, unbiased, and precise cell typing that will have utility for the diagnosis of heme malignancies and immunoprofiling.

Fully automatic training sample collection for detecting multi-decadal inland/seaward urban sprawl

Accurate and timely monitoring of urban sprawl using remote sensing provides valuable insights for urban management and sustainable development. While existing studies have explored methods for large-scale monitoring of urban sprawl, most of them rely on the collection of manually labeled training samples for supervised classification. Additionally, previous studies generally focused on inland urban sprawl, neglecting the process of urban sprawl towards sea areas, known as seaward urban sprawl, which has profound ecological implications for coastal ecosystems. Here we develop a fully automatic algorithm for detecting urban sprawl, named Fully Automatic Detection of Urban Sprawl (FADUS), using the time-series Landsat imagery without manually collecting training samples or existing urban maps for reference. FADUS automatically generates initial samples through a sequential binary unsupervised clustering technique and then updates them by removing the spectral outliers to obtain a high-quality training set for subsequent supervised classification. A tidal module is included in FADUS to avoid pseudo-urban changes caused by tidal fluctuations so that both inland and seaward urban sprawl can be detected. By applying the algorithm to 75 coastal cities in China, the most rapidly urbanizing region in the world, we uncovered a neglected but dramatic seaward urban sprawl process that has converted sea areas to urban space since 1985. Our results indicate that 9904.36 km2 of coastal wetlands and nearshore seawater areas have been converted to land area, of which 44.83% is currently covered by built-up area. With our method, the fully automatic detection of urban dynamics is now possible for both coastal and inland cities worldwide. Our derived map of seaward urban sprawl in China provides valuable references for coastal land-cover change monitoring, wetland protection and restoration, and integrated coastal management.

Fatigue database of additively manufactured alloys

Fatigue is a process of mechanical degradation that is usually assessed based on empirical rules and experimental data obtained from standardized tests. Fatigue data of engineering materials are commonly reported in S-N (the stress-life relation), ε-N (the strain-life relation), and da/dN-ΔK (the relation between the fatigue crack growth rate and the stress intensity factor range) data. Fatigue and static mechanical properties of additively manufactured (AM) alloys, as well as the types of materials, parameters of AM, processing, and testing are collected from thousands of scientific articles till the end of 2022 using natural language processing, machine learning, and computer vision techniques. The results show that the performance of AM alloys could reach that of conventional alloys although data dispersion and system deviation are present. The database (FatigueData-AM2022) is formatted in compact structures, hosted in an open repository, and analyzed to show their patterns and statistics. The quality of data collected from the literature is measured by defining rating scores for datasets reported in individual studies and through the fill rates of data entries across all the datasets. The database also serves as a high-quality training set for data processing using machine learning models. The procedures of data extraction and analysis are outlined and the tools are publicly released. A unified language of fatigue data is suggested to regulate data reporting for the fatigue performance of materials to facilitate data sharing and the development of open science.

Mine local homogeneous representation by interaction information clustering with unsupervised learning in histopathology images

Background and objective: The success of data-driven deep learning for histopathology images often depends on high-quality training sets and fine-grained annotations. However, as tumors are heterogeneous and annotations are expensive, unsupervised learning approaches are desirable to obtain full automation.Methods: In this paper, an Interaction Information Clustering (IIC) method is proposed to extract locally homogeneous features in mutually exclusive clusters. Trained in an unsupervised paradigm, the framework learns invariant information from multiple spatially adjacent regions for improved classification. Additionally, an adaptive Conditional Random Field (CRF) model is incorporated to detect spatially adjacent image patches of high morphological homogeneity in an offset-constraint free manner.Results: Empirically, the proposed model achieves an observable improvement of 11.4% on the downstream patch-level classification accuracy, compared with state-of-the-art unsupervised learning approaches.Conclusion: Furthermore, evaluated with our clinically collected histopathology whole-slide images, the proposed model shows high consistency in tissue distribution compared with well-trained supervised learning, which is of important diagnostic significance in clinical practice.

Hyperspectral Remote Sensing Benchmark Database for Oil Spill Detection With an Isolation Forest-Guided Unsupervised Detector

Oil spill detection has attracted increasing attention in recent years since marine oil spill accidents severely affect environments, natural resources, and the lives of coastal inhabitants. Hyperspectral remote sensing images provide rich spectral information which is beneficial for the monitoring of oil spills in complex ocean scenarios. However, most of the existing approaches are based on supervised and semi-supervised frameworks to detect oil spills from hyperspectral images (HSIs), which require a massive amount of effort to annotate a certain number of high-quality training sets. In this study, we make the first attempt to develop an unsupervised oil spill detection method based on isolation forest for HSIs. First, a Gaussian statistical model is designed to remove the bands corrupted by severe noise. Then, kernel principal component analysis (KPCA) is employed to reduce the high dimensionality of the HSIs. Next, the probability of each pixel belonging to one of the classes of seawater and oil spills is estimated with the isolation forest, and a set of pseudo-labeled training samples is automatically produced using the clustering algorithm on the detected probability. Finally, an initial detection map can be obtained by performing the support vector machine (SVM) on the dimension-reduced data, and the initial detection result is further optimized with the extended random walker (ERW) model so as to improve the detection accuracy of oil spills. Experiments on hyperspectral oil spill data (HOSD) created by ourselves demonstrate that the proposed method obtains superior detection performance with respect to other state-of-the-art detection approaches. We will make HOSD and our developed library for oil spill detection publicly available at https://github.com/PuhongDuan/HOSD to further promote this research topic.

Classification of orbits in Poincaré maps using machine learning

Poincaré plots, also called Poincaré maps, are used by plasma physicists to understand the behavior of magnetically confined plasma in numerical simulations of a tokamak. These plots are created by the intersection of field lines with a two-dimensional poloidal plane that is perpendicular to the axis of the torus representing the tokamak. A plot is composed of multiple orbits, each created by a different field line as it goes around the torus. Each orbit can have one of four distinct shapes, or classes, that indicate changes in the topology of the magnetic fields confining the plasma. Given the (x, y) coordinates of the points that form an orbit, we want to assign a class to the orbit, a task that appears ideally suited for a machine learning approach. In this paper, we describe how we overcame two major challenges in solving this problem—creating a high-quality training set, with few mislabeled orbits, and converting the coordinates of the points into features that are discriminating, despite the variation within the orbits of a class and the apparent similarities between orbits of different classes. Our automated approach is not only more objective and accurate than visual classification, but is also less tedious, making it easier for plasma physicists to analyze the topology of magnetic fields from numerical simulations of the tokamak.

Re_Trans: Combined Retrieval and Transformer Model for Source Code Summarization

Source code summarization (SCS) is a natural language description of source code functionality. It can help developers understand programs and maintain software efficiently. Retrieval-based methods generate SCS by reorganizing terms selected from source code or use SCS of similar code snippets. Generative methods generate SCS via attentional encoder–decoder architecture. However, a generative method can generate SCS for any code, but sometimes the accuracy is still far from expectation (due to the lack of numerous high-quality training sets). A retrieval-based method is considered to have a higher accurac, but usually fails to generate SCS for a source code in the absence of a similar candidate in the database. In order to effectively combine the advantages of retrieval-based methods and generative methods, we propose a new method: Re_Trans. For a given code, we first utilize the retrieval-based method to obtain its most similar code with regard to sematic and corresponding SCS (S_RM). Then, we input the given code and similar code into the trained discriminator. If the discriminator outputs onr, we take S_RM as the result; otherwise, we utilize the generate model, transformer, to generate the given code’ SCS. Particularly, we use AST-augmented (Syntax Tree) and code sequence-augmented information to make the source code semantic extraction more complete. Furthermore, we build a new SCS retrieval library through the public dataset. We evaluate our method on a dataset of 2.1 million Java code-comment pairs, and experimental results show improvement over the state-of-the-art (SOTA) benchmarks, which demonstrates the effectiveness and efficiency of our method.

Short-term wind power prediction based on multidimensional data cleaning and feature reconfiguration

Wind power prediction decreases the uncertainty of the entire energy system, which is essential for balancing energy supply and demand. In order to improve the prediction accuracy, a short-term wind power prediction method based on data cleaning and feature reconfiguration is proposed. A large number of historical samples consisting of wind direction, wind speed, and wind power are mapped into a multidimensional sample space, and the distribution of wind data in different dimensions are analyzed in depth. By calculating the local density of each sample, outliers are effectively detected. The features of wind are reconfigured into a global information map combined with the time series information, which reflects the variation of the wind process in the short term. The features of the original data are greatly enriched, providing a high-quality training set for the prediction model. A redesigned convolutional neural network was used to predict short-term wind power, and the proposed methods were trained and tested based on a dataset of a real wind farm in China. Data cleaning and feature reconfiguration reduce the average single-point error by 1.38% and 2.56%, respectively, while the combined method reduced it by 6.24%. Plenty of experimental results show that the proposed methods achieve good performance and effectively improve the accuracy of short-term wind power prediction.

Pushing the limits of solubility prediction via quality-oriented data selection.

SummaryAccurate prediction of the solubility of chemical substances in solvents remains a challenge. The sparsity of high-quality solubility data is recognized as the biggest hurdle in the development of robust data-driven methods for practical use. Nonetheless, the effects of the quality and quantity of data on aqueous solubility predictions have not yet been scrutinized. In this study, the roles of the size and the quality of data sets on the performances of the solubility prediction models are unraveled, and the concepts of actual and observed performances are introduced. In an effort to curtail the gap between actual and observed performances, a quality-oriented data selection method, which evaluates the quality of data and extracts the most accurate part of it through statistical validation, is designed. Applying this method on the largest publicly available solubility database and using a consensus machine learning approach, a top-performing solubility prediction model is achieved.

Two-Level LSTM for Sentiment Analysis With Lexicon Embedding and Polar Flipping.

Sentiment analysis is a key component in various text mining applications. Numerous sentiment classification techniques, including conventional and deep-learning-based methods, have been proposed in the literature. In most existing methods, a high-quality training set is assumed to be given. Nevertheless, constructing a high-quality training set that consists of highly accurate labels is challenging in real applications. This difficulty stems from the fact that text samples usually contain complex sentiment representations, and their annotation is subjective. We address this challenge in this study by leveraging a new labeling strategy and utilizing a two-level long short-term memory network to construct a sentiment classifier. Lexical cues are useful for sentiment analysis, and they have been utilized in conventional studies. For example, polar and negation words play important roles in sentiment analysis. A new encoding strategy, that is, ρ -hot encoding, is proposed to alleviate the drawbacks of one-hot encoding and, thus, effectively incorporate useful lexical cues. Moreover, the sentimental polarity of a word may change in different sentences due to label noise or context. A flipping model is proposed to model the polar flipping of words in a sentence. We compile three Chinese datasets on the basis of our label strategy and proposed methodology. Experiments demonstrate that the proposed method outperforms state-of-the-art algorithms on both benchmark English data and our compiled Chinese data.

Integrating National Ecological Observatory Network (NEON) Airborne Remote Sensing and In-Situ Data for Optimal Tree Species Classification

Accurately mapping tree species composition and diversity is a critical step towards spatially explicit and species-specific ecological understanding. The National Ecological Observatory Network (NEON) is a valuable source of open ecological data across the United States. Freely available NEON data include in-situ measurements of individual trees, including stem locations, species, and crown diameter, along with the NEON Airborne Observation Platform (AOP) airborne remote sensing imagery, including hyperspectral, multispectral, and light detection and ranging (LiDAR) data products. An important aspect of predicting species using remote sensing data is creating high-quality training sets for optimal classification purposes. Ultimately, manually creating training data is an expensive and time-consuming task that relies on human analyst decisions and may require external data sets or information. We combine in-situ and airborne remote sensing NEON data to evaluate the impact of automated training set preparation and a novel data preprocessing workflow on classifying the four dominant subalpine coniferous tree species at the Niwot Ridge Mountain Research Station forested NEON site in Colorado, USA. We trained pixel-based Random Forest (RF) machine learning models using a series of training data sets along with remote sensing raster data as descriptive features. The highest classification accuracies, 69% and 60% based on internal RF error assessment and an independent validation set, respectively, were obtained using circular tree crown polygons created with half the maximum crown diameter per tree. LiDAR-derived data products were the most important features for species classification, followed by vegetation indices. This work contributes to the open development of well-labeled training data sets for forest composition mapping using openly available NEON data without requiring external data collection, manual delineation steps, or site-specific parameters.

Barrage Text Classification with Improved Active Learning and CNN

Traditional convolutional neural networks (CNNs) use a pooling layer to reduce the dimensionality of texts, but lose semantic information. To solve this problem, this paper proposes a convolutional neural network model based on singular value decomposition algorithm (SVD-CNN). First, an improved density-based center point clustering active learning sampling algorithm (DBC-AL) is used to obtain a high-quality training set at a low labelling cost. Second, the method uses the singular value decomposition algorithm for feature extraction and dimensionality reduction instead of a pooling layer, fuses the dimensionality reduction matrix, and completes the barrage text classification task. Finally, the partial sampling gradient descent algorithm (PSGD) is applied to optimize the model parameters, which accelerates the convergence speed of the model while ensuring stability of the model training. To verify the effectiveness of the improved algorithm, several barrage datasets were used to compare the proposed model and common text classification models. The experimental results show that the improved algorithm preserves the semantic features of the text more successfully, ensures the stability of the training process, and improves the convergence speed of the model. Further, the model’s classification performance on different barrage texts is superior to traditional algorithms.

Automated 4π radiotherapy treatment planning with evolving knowledge‐base

Non-coplanar 4π radiotherapy generalizes intensity modulated radiation therapy (IMRT) to automate beam geometry selection but requires complicated hyperparameter tuning to attain superior plan quality, which can be tedious and inconsistent. In this study, a fully automated 4π treatment planning was developed using evolving knowledge-base (EKB) planning guided by dose prediction. Twenty 4π lung and twenty 4π head and neck (HN) cases were included. A statistical voxel dose learning model was initially trained on low-quality plans created using generic hyperparameter templates without manual tuning. To improve the automated plan quality without being limited by the training data quality, a new 4π optimization problem was formulated to include a one-sided penalty on the organ-at-risk (OAR) dose deviation from the predicted dose. This directional OAR penalty encourages superior OAR sparing. The fast iterative shrinkage-thresholding algorithm (FISTA) was used to solve the large-scale beam orientation optimization problem. With the improved plans, new predictions were created to guide the next loop of EKB planning for a total of 10 loops. Plan quality was evaluated using a plan quality metric (PQM) points system based on clinical dose constraints and compared with automated planning approaches guided by manual high-quality plans using all non-coplanar beams, automated plans using individually evolved targeted dose, and manually created 4π plans. For the lung cases, the final EKB plans had significantly higher PQM than manually created 4π (+2.60%). The improvements plateaued after the third loop. The final HN EKB plans and manually created 4π plans had comparable PQMs, but had lower PQM compared to automated plans using a high-quality training set (-3.00% and -4.44%, respectively). The PQM consistently increased up to the sixth loop. Individually evolved plans were able to improve the plan quality from initial condition due to the one-sided cost function but the 60% of them were trapped in undesired local minima that were substantially worse than their corresponding EKB plans. Evolving knowledge-base planning is a novel automated planning technique guided by the predicted three-dimensional dose distribution, which can evolve from low-quality plans. EKB allows new beams to be used in the automated planning workflow for superior plan quality.

Low-frequency noise suppression for desert seismic data based on a wide inference network

Abstract The importance of seismic exploration has been recognized by geophysicists. At present, low-frequency noise usually exists in seismic exploration, especially in desert seismic records. This low-frequency noise shares the same frequency band with effective signals. This leads to the limitation or failure of traditional methods. In order to overcome the shortcomings of traditional denoising methods, we propose a novel desert seismic data denoising method based on a Wide Inference Network (WIN). The WIN aims to minimize the error between the prediction and target by residual learning during training, and it can obtain a set of optimal parameters, such as weights and biases. In this article, we construct a high-quality training set for a desert seismic record and this ensures the effective training of a WIN. In this way, each layer of the trained WIN can automatically extract a set of time–space characteristics without manual adjustment. These characteristics are transmitted layer by layer. Finally, they are utilized to extract effective signals. To verify the effectiveness of the WIN, we apply it to synthetic and real desert seismic records, respectively. In addition, we compare WIN with f – x deconvolution, variational mode decomposition (VMD) and shearlet transform. The results show that WIN has the best denoising performance in suppressing low-frequency noise and preserving effective signals.

Artificial intelligence at the intersection of pathology and radiology in prostate cancer.

Pathologic grading plays a key role in prostate cancer risk stratification and treatment selection, traditionally assessed from systemic core needle biopsies sampled throughout the prostate gland. Multiparametric magnetic resonance imaging (mpMRI) has become a well-established clinical tool for detecting and localizing prostate cancer. However, both pathologic and radiologic assessment suffer from poor reproducibility among readers. Artificial intelligence (AI) methods show promise in aiding the detection and assessment of imaging-based tasks, dependent on the curation of high-quality training sets. This review provides an overview of recent advances in AI applied to mpMRI and digital pathology in prostate cancer which enable advanced characterization of disease through combined radiology-pathology assessment.