Stochastic Clustering Research Articles

Robust and stochastic sparse subspace clustering

Sparse subspace clustering (SSC) has been widely employed in machine learning and pattern recognition, but it still faces scalability challenges when dealing with large-scale datasets. Recently, stochastic SSC (SSSC) has emerged as an effective solution by leveraging the dropout technique. However, SSSC cannot robustly handle noise, especially non-Gaussian noise, leading to unsatisfactory clustering performance. To address the above issues, we propose a novel robust and stochastic method called stochastic sparse subspace clustering with the Huber function (S3CH). The key idea is to introduce the Huber surrogate to measure the loss of the stochastic self-expression framework, thus S3CH inherits the advantage of the stochastic framework for large-scale problems while mitigating sensitivity to non-Gaussian noise. In algorithms, an efficient proximal alternating minimization (PAM)-based optimization scheme is developed. In theory, the convergence of the generated sequence is rigorously proved. Extensive numerical experiments on synthetic and six real datasets validate the advantages of the proposed method in clustering accuracy, noise robustness, parameter sensitivity, post-hoc analysis, and model stability.

A multi-symptomatic model of heroin use disorder in rats reveals distinct behavioral profiles and neuronal correlates of heroin vulnerability versus resiliency.

The behavioral and diagnostic heterogeneity within human opioid use disorder (OUD) diagnosis is not readily captured in current animal models, limiting translational relevance of the mechanistic research that is conducted in experimental animals. We hypothesize that a non-linear clustering of OUD-like behavioral traits will capture population heterogeneity and yield subpopulations of OUD vulnerable rats with distinct behavioral and neurocircuit profiles. Over 900 male and female heterogeneous stock rats, a line capturing genetic and behavioral heterogeneity present in humans, were assessed for several measures of heroin use and rewarded and non-rewarded seeking behaviors. Using a non-linear stochastic block model clustering analysis, rats were assigned to OUD vulnerable, intermediate and resilient clusters. Additional behavioral tests and circuit analyses using c-fos protein activation were conducted on the vulnerable and resilient subpopulations. OUD vulnerable rats exhibited greater heroin taking and seeking behaviors relative to those in the intermediate and resilient clusters. Akin to human OUD diagnosis, further vulnerable rat sub-clustering revealed subpopulations with different combinations of behavioral traits, including sex differences. Lastly, heroin cue-induced neuronal patterns of circuit activation differed between resilient and vulnerable phenotypes. Behavioral sex differences were recapitulated in patterns of circuitry activation, including males preferentially engaging extended amygdala stress circuitry, and females cortico-striatal drug cue-seeking circuitry. Using a non-linear clustering approach in rats, we captured behavioral diagnostic heterogeneity reflective of human OUD diagnosis. OUD vulnerability and resiliency were associated with distinct neuronal activation patterns, posing this approach as a translational tool in assessing neurobiological mechanisms underpinning OUD.

Simulations of radiation damage accumulation in Fe-9Cr under pulsed irradiation conditions representative of inertial fusion energy

Structural materials for laser-based inertial fusion energy (IFE) reactor concepts are expected to operate under pulsed irradiation conditions, with cycles consisting of microsecond-long neutron bursts followed by inter-pulse periods of up to one second in duration. During each laser shot, irradiation damage is introduced at dose rates that are up to six orders of magnitude higher than those in their magnetic fusion energy (MFE) counterparts. Under certain conditions, the inter-pulse periods may last an amount of time sufficient to anneal much of the damage introduced during each shot. This phenomenon is highly temperature dependent, with pulsed heating directly linked to the pulsed damage, large surface temperature spikes may also occur. As such, this intermittent mode of operation has the potential to lead to fundamental differences in how irradiation damage accumulates in structural reactor materials. However, damage to structural materials under IFE conditions has received comparatively much less attention than in MFE, as pulsed conditions add yet an extra dimension to the already extremely challenging problem of microstructural evolution under fusion neutron irradiation in structural materials. In this work we use the stochastic cluster dynamics (SCD) method to simulate the evolution with time of defect cluster concentrations under IFE conditions. We consider the Laser Inertial Fusion Energy (LIFE) reactor concept as the representative IFE design for our study, for which detailed spectral information is available, including gas transmutant production. We simulate several pulse frequencies and three different temperatures, and compare the results with continuous irradiation cases under identical average dose rates. The simulations are run in Fe-9Cr system as a model alloy for reduced-activation ferritic/martensitic (RAFM) steels, which are the leading structural material candidates for first-wall structures in MFE and IFE devices. We find that, in practically all scenarios, pulsed irradiation restricts the formation of helium-vacancy clusters relative to the levels seen under equivalent steady irradiation conditions. As well, although self-interstitial atom clusters do accumulate under pulsed operation, their number densities remain up to an order of magnitude lower than in continuous irradiation conditions. Based on the SCD results, we provide a temperature-pulse rate map to identify regions where pulsed irradiation may lead to larger defect accumulation than under continuous irradiation.

Stochastic cluster head selection model for energy balancing in IoT enabled heterogeneous WSN

Energy dissipation is the most important design limitation for Internet of Things (IoT) enabled Wireless Sensor Networks (WSNs). In order to prolong the life of WSNs, the energy of nodes must be used in an effective way. Clustering is a strategy that may effectively use the energy of the sensors, extending the life and scalability by managing the network load balance. The energy usage for network operation is reduced by using an evolutionary algorithm called Genetic Algorithm (GA). The Stochastic Cluster Head Selection Model (SCHSM) is described in the proposed protocol by taking the factors such as distance, node energy, density and capacity of nodes for developing the fitness function. The proposed protocol is designed for multiple movable sink nodes and this greatly improves the energy balancing factor in the network. For minimizing the communication gap among sensors and sinks, movable sinks can be placed carefully. Simulation results are analyzed for the system effectiveness.

Assembly structures of coastal woody species of eastern South America: Patterns and drivers

Tropical regions have provided new insights into how ecological communities are assembled. In dry coastal communities, water stress has been hypothesized to determine plant assembly structure by favoring preadapted lineages from neighboring ecosystems, consistent with functional clustering. However, it is unclear whether this hypothesis is sufficient to explain how coastal communities in tropical ecosystems are assembled. Here, we test whether water stress or other factors drive community assembly in woody plant communities across the coastal zone of Brazil, a tropical ecosystem. We characterized functional and phylogenetic structures of these communities and determined the underlying environmental factors (e.g., water stress, historical climate stability, edaphic constraints, and habitat heterogeneity) that drive their community assembly. Assemblages of coastal woody species show geographically varied patterns, including stochastic arrangements, clustering, and overdispersion of species relative to their traits and phylogenetic relatedness. Topographic complexity, water vapor pressure, and soil nutrient availability best explained the gradient in the functional structure. Water deficit, water vapor pressure, and soil organic carbon were the best predictors of variation in phylogenetic structure. Our results support the water-stress conservatism hypothesis on functional and phylogenetic structure, as well as the effect of habitat heterogeneity on functional structure and edaphic constraints on functional and phylogenetic structure. These effects are associated with increased phenotypic and phylogenetic divergence of woody plant assemblages, which is likely mediated by abiotic filtering and niche opportunities, suggesting a complex pattern of ecological assembly.

Data-driven allocation of smart grid-connected system based on ant colony optimization algorithm

In the era of advanced technology, integrating and distributing data are crucial in smart grid-connected systems. However, as energy loads continue to increase, practical implementation of these systems faces challenges in resource allocation and lacks efficient data collaboration. In this study, the ant colony optimization algorithm is further investigated for stochastic crossover systems and cluster nodes in intelligent path planning management. To improve the pheromone setting method in smart grid-connected systems, we propose an adaptive intelligent ant colony optimization algorithm called the Group Allocation Optimization Algorithm (GAOA). This algorithm expands the pheromone transmission rate of network nodes, establishes a multi-constrained adaptive model with data mining as the pheromone target, and analyzes the accuracy of resource allocation to import the optimal scheme for smart grid-connected systems. Through experimental results, we demonstrate that the optimized adaptive ant colony algorithm leads to effective improvements in grid-connected systems, pheromone evaluation, data throughput, convergence speed, and data load distribution. These findings provide evidence that the optimized ant colony algorithm is both feasible and effective for resource allocation in smart grid-connected systems.

A Massively Parallel Spatially Resolved Stochastic Cluster Dynamics Method for Simulations of Irradiated Materials

The spatially resolved stochastic cluster dynamics (SRSCD) method is one of the most important methods for simulating the time-evolution of spatially correlated microstructures in irradiated materials. It is a kinetic Monte Carlo-based method for stochastically evolving integer-valued populations of defects within multi finite volume elements according to reaction rates. However, the increasing spatial scale and complexity of the simulated systems have exceeded the capabilities of serial SRSCD. To extend SRSCD to simulate large-scale complex systems, we propose a massively parallel SRSCD method in this paper and implement the program named MISA-SLSCD. It contains four contributions: (1) a dynamic Defect-Reaction Tree data structure to efficiently store and update millions of defect species and reactions; (2) a double-grouping search strategy to speed up the search for defects and reactions; (3) an adaptive synchronous algorithm to balance the accuracy and efficiency of simulations; and (4) an on-demand communication strategy to eliminate communication redundancy. A series of numerical simulation results show that our method has high accuracy in simulating damage accumulation in irradiated materials and obtains a good performance of over 90% parallel efficiency on 32,000 CPU cores with a 32 million-volume-element system. Program summaryProgram Title: MISA-SLSCDCPC Library link to program files:https://doi.org/10.17632/ctmn5f5bzk.1Code Ocean capsule:https://codeocean.com/capsule/8351487Licensing provisions: BSD 3-clauseProgramming language: Fortran90Nature of problem: Behaviours of defects in irradiated materials are typically space-dependent long-term dynamical processes, and spanning multiple orders of magnitude in space (∼nm to m) and time (∼ μs to years). The SRSCD method is an important method for simulating the time-evolution of spatially correlated microstructures in irradiated materials. However, the increasing spatial scale and complexity of the simulated systems have exceeded the capabilities of serial and existing parallel versions. It is necessary to develop new massively parallel method and program to extend SRSCD to simulate large-scale complex systems.Solution method: To extend SRSCD to simulate large-scale complex systems, we propose a massively parallel SRSCD method in this paper and implement the program named MISA-SLSCD. It contains: (1) a dynamic Defect-Reaction Tree data structure to efficiently store and update millions of defect species and reactions; (2) a double-grouping search strategy to speed up the search for defects and reactions; (3) an adaptive synchronous algorithm to balance the accuracy and efficiency of simulations; and (4) an on-demand communication strategy to eliminate communication redundancy.

Banff scoring of kidney allograft biopsies: "Manual" application vs software-assisted sign-out.

Pathologists interpreting kidney allograft biopsies using the Banff system usually start by recording component scores (eg, i, t, cg) using histopathologic criteria committed to memory. Component scores are then melded into diagnoses using the same manual/mental processes. This approach to complex Banff rules during routine sign-out produces a lack of fidelity and needs improvement. We constructed a web-based "smart template" (software-assisted sign-out) system that uniquely starts with upstream Banff-defined additional diagnostic parameters (eg, infection) and histopathologic criteria (eg, percent interstitial inflammation) collectively referred to as feeder data that is then translated into component scores and integrated into final diagnoses using software-encoded decision trees. Software-assisted sign-out enables pathologists to (1) accurately and uniformly apply Banff rules, thereby eliminating human inconsistencies (present in 25% of the cohort); (2) document areas of improvement; (3) show improved correlation with function; (4) examine t-Distributed Stochastic Neighbor Embedding clustering for diagnosis stratification; and (5) ready upstream incorporation of artificial intelligence-assisted scoring of biopsies. Compared with the legacy approach, software-assisted sign-out improves Banff accuracy and fidelity, more closely correlates with kidney function, is practical for routine clinical work and translational research studies, facilitates downstream integration with nonpathology data, and readies biopsy scoring for artificial intelligence algorithms.

Correlation Analysis of Large-Span Cable-Stayed Bridge Structural Frequencies with Environmental Factors Based on Support Vector Regression

The dynamic characteristics of bridge structures are influenced by various environmental factors, and exploring the impact of environmental temperature and humidity on structural modal parameters is of great significance for structural health assessment. This paper utilized the Covariance-Driven Stochastic Subspace Identification method (SSI-COV) and clustering algorithms to identify modal frequencies from four months of acceleration data collected from the health monitoring system of the Jintang Hantan Twin-Island Bridge. Furthermore, a correlation analysis is conducted to examine the relationship between higher-order frequency and environmental factors, including temperature and humidity. Subsequently, a Support Vector Machine Regression (SVR) model is employed to analyze the effects of environmental temperature on structural modal frequencies. This study has obtained the following conclusions: 1. Correlation analysis revealed that temperature is the primary influencing factor in frequency variations. Frequency exhibited a strong linear correlation with temperature and little correlation with humidity. 2. SVR regression analysis was performed on frequency and temperature, and an evaluation of the fitting residuals was conducted. The model effectively fit the sample data and provided reliable predictive results. 3. The original structural frequencies underwent smoothing, eliminating the influence of temperature-induced frequency data generated by the SVR model. After eliminating the temperature effects, the fluctuations in frequency within a 24 h period significantly decreased. The data presented in this paper can serve as a reference for further health assessments of similar bridge structures.

Dynamic mixed models with heterogeneous covariance components using multivariate GARCH innovations and the Dirichlet process mixture

This paper presents innovative strategies for correlated data analysis using dynamic mixed effects models to account for the autoregressive conditional-heteroscedasticity of innovations and cross-sectional dependence. In a multivariate setting, the traditional modeling scheme was extended to time-varying components of variances and covariances by initiating the heterogeneous GARCH model. We propose a Bayesian semi-parametric approach by utilizing the centered Dirichlet process as priors for the heterogeneity effects involved in the mean structure. We present a stochastic clustering strategy to distinguish similar patterns of covariates. The Hamiltonian Monte Carlo technique is used to ease the computation of time-series equations for variances and covariances with higher efficiency. We conduct comprehensive simulation studies to examine model features, especially in cluster-wise settings. We adopt the maximal overlap discrete wavelet transforms to isolate short- and long-run sample information for practical applications. The transformation facilitates analyzing the effect of high-frequency or trend variation in macroeconomic factors. We illustrate the advantage of our proposed modeling methodology in the empirical studies by investigating the short-run impacts of macroeconomic events on economic growth and its prediction accounting for the unexpected volatility in the financial market of some selected countries.

Immune checkpoint blockade therapy mitigates systemic inflammation and affects cellular FLIP-expressing monocytic myeloid-derived suppressor cells in non-progressor non-small cell lung cancer patients

ABSTRACT Cancer cells favor the generation of myeloid cells with immunosuppressive and inflammatory features, including myeloid-derived suppressor cells (MDSCs), which support tumor progression. The anti-apoptotic molecule, cellular FLICE (FADD-like interleukin-1β-converting enzyme)-inhibitory protein (c-FLIP), which acts as an important modulator of caspase-8, is required for the development and function of monocytic (M)-MDSCs. Here, we assessed the effect of immune checkpoint inhibitor (ICI) therapy on systemic immunological landscape, including FLIP-expressing MDSCs, in non-small cell lung cancer (NSCLC) patients. Longitudinal changes in peripheral immunological parameters were correlated with patients’ outcome. In detail, 34 NSCLC patients were enrolled and classified as progressors (P) or non-progressors (NP), according to the RECIST evaluation. We demonstrated a reduction in pro-inflammatory cytokines such as IL-8, IL-6, and IL-1β in only NP patients after ICI treatment. Moreover, using t-distributed stochastic neighbor embedding (t-SNE) and cluster analysis, we characterized in NP patients a significant increase in the amount of lymphocytes and a slight contraction of myeloid cells such as neutrophils and monocytes. Despite this moderate ICI-associated alteration in myeloid cells, we identified a distinctive reduction of c-FLIP expression in M-MDSCs from NP patients concurrently with the first clinical evaluation (T1), even though NP and P patients showed the same level of expression at baseline (T0). In agreement with the c-FLIP expression, monocytes isolated from both P and NP patients displayed similar immunosuppressive functions at T0; however, this pro-tumor activity was negatively influenced at T1 in the NP patient cohort exclusively. Hence, ICI therapy can mitigate systemic inflammation and impair MDSC-dependent immunosuppression.

ESPRESSO: topological clustering of single-cell transcriptomics data to reveal informative genes for spatio–temporal architectures of cells

BackgroundBioinformatics capability to analyze spatio–temporal dynamics of gene expression is essential in understanding animal development. Animal cells are spatially organized as functional tissues where cellular gene expression data contain information that governs morphogenesis during the developmental process. Although several computational tissue reconstruction methods using transcriptomics data have been proposed, those methods have been ineffective in arranging cells in their correct positions in tissues or organs unless spatial information is explicitly provided.ResultsThis study demonstrates stochastic self-organizing map clustering with Markov chain Monte Carlo calculations for optimizing informative genes effectively reconstruct any spatio–temporal topology of cells from their transcriptome profiles with only a coarse topological guideline. The method, eSPRESSO (enhanced SPatial REconstruction by Stochastic Self-Organizing Map), provides a powerful in silico spatio–temporal tissue reconstruction capability, as confirmed by using human embryonic heart and mouse embryo, brain, embryonic heart, and liver lobule with generally high reproducibility (average max. accuracy = 92.0%), while revealing topologically informative genes, or spatial discriminator genes. Furthermore, eSPRESSO was used for temporal analysis of human pancreatic organoids to infer rational developmental trajectories with several candidate ‘temporal’ discriminator genes responsible for various cell type differentiations.ConclusionseSPRESSO provides a novel strategy for analyzing mechanisms underlying the spatio–temporal formation of cellular organizations.

Lancaster Stem Sammon Projective Feature Selection based Stochastic eXtreme Gradient Boost Clustering for Web Page Ranking

Web content mining retrieves the information from web in more structured forms. The page rank plays an essential part in web content mining process. Whenever user searches for any information on web, the relevant information is shown at top of list through page ranking. Many existing page ranking algorithms were developed and failed to rank the web pages in accurate manner through minimum time feeding. In direction to address the above mentioned issues, Lancaster Stem Sammon Projective Feature Selection based Stochastic eXtreme Gradient Boost Clustering (LSSPFS-SXGBC) Approach is introduced for page ranking based on user query. LSSPFS-SXGBC Approach has three processes for performing efficient web page ranking, namely preprocessing, feature selection and clustering. LSSPFS-SXGBC Approach in account of the numeral of operator request by way of an input. Lancaster Stemming Preprocessed Analysis is carried out in LSSPFS-SXGBC Approach for removing the noisy data from the input query. It eradicates the stem words, stop words and incomplete data for minimizing the time and space consumption. Sammon Projective Feature Selection Process is carried out in LSSPFS-SXGBC Approach to select the relevant features (i.e., keywords) based on user needs for efficient page ranking. Sammon Projection maps the high-dimensional space to lower dimensionality space to preserve the inter-point distance structure. After feature selection, Stochastic eXtreme Gradient Boost Page Rank Clustering process is carried out to cluster the similar keyword web pages based on their rank. Gradient Boost Page Rank Cluster is an ensemble of several weak clusters (i.e., X-means cluster). X-means cluster partitions the web pages into ‘x’ numeral of clusters where each reflection goes towards the cluster through adjacent mean value. For every weak cluster, selected features are considered as the training samples. Subsequently, all weak clusters are joined to form the strong cluster for attaining the webpage ranking results. By this way, an efficient page ranking is carried out through higher accurateness and minimum time consumption. The practical validation is carried out in LSSPFS-SXGBC Approach on factors such ranking accurateness, false positive rate, ranking time and space complexity with respect to numeral of user query.

BigNeuron: a resource to benchmark and predict performance of algorithms for automated tracing of neurons in light microscopy datasets.

BigNeuron is an open community bench-testing platform with the goal of setting open standards for accurate and fast automatic neuron tracing. We gathered a diverse set of image volumes across several species that is representative of the data obtained in many neuroscience laboratories interested in neuron tracing. Here, we report generated gold standard manual annotations for a subset of the available imaging datasets and quantified tracing quality for 35 automatic tracing algorithms. The goal of generating such a hand-curated diverse dataset is to advance the development of tracing algorithms and enable generalizable benchmarking. Together with image quality features, we pooled the data in an interactive web application that enables users and developers to perform principal component analysis, t-distributed stochastic neighbor embedding, correlation and clustering, visualization of imaging and tracing data, and benchmarking of automatic tracing algorithms in user-defined data subsets. The image quality metrics explain most of the variance in the data, followed by neuromorphological features related to neuron size. We observed that diverse algorithms can provide complementary information to obtain accurate results and developed a method to iteratively combine methods and generate consensus reconstructions. The consensus trees obtained provide estimates of the neuron structure ground truth that typically outperform single algorithms in noisy datasets. However, specific algorithms may outperform the consensus tree strategy in specific imaging conditions. Finally, to aid users in predicting the most accurate automatic tracing results without manual annotations for comparison, we used support vector machine regression to predict reconstruction quality given an image volume and a set of automatic tracings.

Stochastic hyperbola fitting, probabilistic inversion, reverse-time migration and clustering: A novel interpretation toolbox for in-situ planetary radar

Ground-penetrating radar (GPR) is becoming a mainstream tool in planetary exploration, and one of the few in-situ planetary geophysical methods. There are currently three missions (Perseverance, Tianwen-1, Chang’E-4) with GPR-equipped rovers, and two future missions (Chang’E-7, ExoMars) that will include GPR in their scientific payload. The large number of GPR data, combined with the novel setup of the measurements, creates the need for new data processing and interpretation techniques to address the unique challenges of in-situ planetary radar. The current paper proposes an interpretation pipeline that starts with a novel stochastic hyperbola fitting that estimates the probability kernel density of the bulk permittivity at different depths. Subsequently, the bulk permittivity distribution is transformed via a novel probabilistic inversion to a 1-dimensional (1D) permittivity profile. The inverted 1D permittivity profile is then used as an input to a bespoke reverse-time migration (RTM) using the finite-difference time-domain (FDTD) method. RTM using FDTD does not assume a clinical homogeneous half-space; instead, it accounts for the expected layered structure of the investigated medium. Lastly, the migrated radargram is clustered in order to identify subsurface targets and distinguish them from the background medium. Each of the processing steps has never been reported in planetary radar; and together act as a complete processing toolbox tuned for planetary science. The suggested interpretation pipeline is validated numerically in a 1D case study with a complex layered structure and multiple subsurface targets. The proposed processing scheme is then applied to the GPR data from the Chang’E-4 mission at the Von Kármán crater, revealing a previously unseen layered structure and a complex distribution of rocks/boulders.

Single-cell transcriptomic analysis of tumor heterogeneity and intercellular networks in human urothelial carcinoma.

Heterogeneity of tumor cells and the tumor microenvironment (TME) is significantly associated with clinical outcomes and treatment responses in patients with urothelial carcinoma (UC). Comprehensive profiling of the cellular diversity and interactions between malignant cells and TME may clarify the mechanisms underlying UC progression and guide the development of novel therapies. This study aimed to extend our understanding of intra-tumoral heterogeneity and the immunosuppressive TME in UC and provide basic support for the development of novel UC therapies. Seven patients with UC were included who underwent curative surgery at our hospital between July 2020 and October 2020. We performed single-cell RNA sequencing (scRNA-seq) analysis in seven tumors with six matched adjacent normal tissues and integrated the results with two public scRNA-seq datasets. The functional properties and intercellular interactions between single cells were characterized, and the results were validated using multiplex immunofluorescence staining, flow cytometry, and bulk transcriptomic datasets. All statistical analyses were performed using the R package with two-sided tests. Wilcoxon-rank test, log-rank test, one-way analysis of variance test, and Pearson correlation analysis were used properly. Unsupervised t-distributed stochastic neighbor embedding clustering analysis identified ten main cellular subclusters in urothelial tissues. Of them, seven urothelial subtypes were noted, and malignant urothelial cells were characterized with enhanced cellular proliferation and reduced immunogenicity. CD8 + T cell subclusters exhibited enhanced cellular cytotoxicity activities along with increased exhaustion signature in UC tissues, and the recruitment of CD4 + T regulatory cells was also increased in tumor tissues. Regarding myeloid cells, coordinated reprogramming of infiltrated neutrophils, M2-type polarized macrophages, and LAMP3 + dendritic cells contribute to immunosuppressive TME in UC tissues. Tumor tissues demonstrated enhanced angiogenesis mediated by KDR + endothelial cells and RGS5 + /ACTA2 + pericytes. Through deconvolution analysis, we identified multiple cellular subtypes may influence the programmed death-ligand 1 (PD-L1) immunotherapy response in patients with UC. Our scRNA-seq analysis clarified intra-tumoral heterogeneity and delineated the pro-tumoral and immunosuppressive microenvironment in UC tissues, which may provide novel therapeutic targets.

Large-scale in vitro microdosimetry via live cell microscopy imaging: implications for radiosensitivity and RBE evaluations in alpha-emitter radiopharmaceutical therapy

BackgroundAlpha-emitter radiopharmaceutical therapy (αRPT) has shown promising outcomes in metastatic disease. However, the short range of the alpha particles necessitates dosimetry on a near-cellular spatial scale. Current knowledge on cellular dosimetry is primarily based on in vitro experiments using cell monolayers. The goal of such experiments is to establish cell sensitivity to absorbed dose (AD). However, AD cannot be measured directly and needs to be modeled. Current models, often idealize cells as spheroids in a regular grid (geometric model), simplify binding kinetics and ignore the stochastic nature of radioactive decay. It is unclear what the impact of such simplifications is, but oversimplification results in inaccurate and non-generalizable results, which hampers the rigorous study of the underlying radiobiology.MethodsWe systematically mapped out 3D cell geometries, clustering behavior, agent binding, internalization, and subcellular trafficking kinetics for a large cohort of live cells under representative experimental conditions using confocal microscopy. This allowed for realistic Monte Carlo-based (micro)dosimetry. Experimentally established surviving fractions of the HER2 + breast cancer cell line treated with a 212Pb-labelled anti-HER2 conjugate or external beam radiotherapy, anchored a rigorous statistical approach to cell sensitivity and relative biological effectiveness (RBE) estimation. All outcomes were compared to a reference geometric model, which allowed us to determine which aspects are crucial model components for the proper study of the underlying radiobiology.ResultsIn total, 567 cells were measured up to 26 h post-incubation. Realistic cell clustering had a large (2x), and cell geometry a small (16.4% difference) impact on AD, compared to the geometric model. Microdosimetry revealed that more than half of the cells do not receive any dose for most of the tested conditions, greatly impacting cell sensitivity estimates. Including these stochastic effects in the model, resulted in significantly more accurate predictions of surviving fraction and RBE (permutation test; p < .01).ConclusionsThis comprehensive integration of the biological and physical aspects resulted in a more accurate method of cell survival modelling in αRPT experiments. Specifically, including realistic stochastic radiation effects and cell clustering behavior is crucial to obtaining generalizable radiobiological parameters.

Automated in-depth cerebral arterial labelling using cerebrovascular vasculature reframing and deep neural networks

Identifying the cerebral arterial branches is essential for undertaking a computational approach to cerebrovascular imaging. However, the complexity and inter-individual differences involved in this process have not been thoroughly studied. We used machine learning to examine the anatomical profile of the cerebral arterial tree. The method is less sensitive to inter-subject and cohort-wise anatomical variations and exhibits robust performance with an unprecedented in-depth vessel range. We applied machine learning algorithms to disease-free healthy control subjects (n = 42), patients with stroke with intracranial atherosclerosis (ICAS) (n = 46), and patients with stroke mixed with the existing controls (n = 69). We trained and tested 70% and 30% of each study cohort, respectively, incorporating spatial coordinates and geometric vessel feature vectors. Cerebral arterial images were analyzed based on the ‘segmentation-stacking’ method using magnetic resonance angiography. We precisely classified the cerebral arteries across the exhaustive scope of vessel components using advanced geometric characterization, redefinition of vessel unit conception, and post-processing algorithms. We verified that the neural network ensemble, with multiple joint models as the combined predictor, classified all vessel component types independent of inter-subject variations in cerebral arterial anatomy. The validity of the categorization performance of the model was tested, considering the control, ICAS, and control-blended stroke cohorts, using the area under the receiver operating characteristic (ROC) curve and precision-recall curve. The classification accuracy rarely fell outside each image’s 90–99% scope, independent of cohort-dependent cerebrovascular structural variations. The classification ensemble was calibrated with high overall area rates under the ROC curve of 0.99–1.00 [0.97–1.00] in the test set across various study cohorts. Identifying an all-inclusive range of vessel components across controls, ICAS, and stroke patients, the accuracy rates of the prediction were: internal carotid arteries, 91–100%; middle cerebral arteries, 82–98%; anterior cerebral arteries, 88–100%; posterior cerebral arteries, 87–100%; and collections of superior, anterior inferior, and posterior inferior cerebellar arteries, 90–99% in the chunk-level classification. Using a voting algorithm on the queued classified vessel factors and anatomically post-processing the automatically classified results intensified quantitative prediction performance. We employed stochastic clustering and deep neural network ensembles. Ma-chine intelligence-assisted prediction of vessel structure allowed us to personalize quantitative predictions of various types of cerebral arterial structures, contributing to precise and efficient decisions regarding the cerebrovascular disease.

Development and validation of a prognostic model based on a single-cell RNA-seq in Wilms tumor in children.

To analyze the heterogeneity between different cell types in pediatric Wilms tumor (WT) tissue, and identify the differentially expressed genes (DEGs) of malignant tumor cells, thereby establishing a prognostic model. The single-cell sequencing data of pediatric WT tissues were downloaded from the public database. Data filtration and normalization, principal component analysis, and T-distributed stochastic neighbor embedding cluster analysis were performed using the Seurat package of R language. Cells were divided into different clusters, malignant tumor cells were extracted, and DEGs were obtained. Then, the pseudo-time trajectory analysis was performed. Prognostic biomarkers were determined by univariate and multivariate COX regression analyses and LASSO regression analysis. Kaplan-Meier survival analysis and receiver operator characteristic curve analysis were performed. Combined with the prognostic biomarkers and clinical characteristics, a nomogram was generated to predict WT prognosis. The prognostic power was validated in the external datasets. Cells in the WT tissue were divided into 10 clusters. Three prognostic biomarkers that affected the survival time of patients were screened from 215 DEGs in malignant tumor cells, and a nomogram was constructed using the three genes and clinical characteristics. The area under the curve (AUC) values of 3- and 5-year disease-free survival were 0.756 and 0.734, respectively. In the external validation dataset, the AUC value of this nomogram model was 0.826. Based on the single-cell RNA-seq, we recognized cell clusters in the WT tissue of children, identified prognostic biomarkers in malignant tumor cells, and established a comprehensive prognostic model. Our findings might provide new ideas and methods for the diagnosis and treatment of WT.

Cluster analysis via random partition distributions

AbstractHierarchical and k‐medoids clustering are deterministic clustering algorithms defined on pairwise distances. We use these same pairwise distances in a novel stochastic clustering procedure based on a probability distribution. We call our proposed method CaviarPD, a portmanteau from cluster analysis via random partition distributions. CaviarPD first samples clusterings from a distribution on partitions and then finds the best cluster estimate based on these samples using algorithms to minimize an expected loss. Using eight case studies, we show that our approach produces results as close to the truth as hierarchical and k‐medoids methods, and has the additional advantage of allowing for a probabilistic framework to assess clustering uncertainty. The method provides an intuitive graphical representation of clustering uncertainty through pairwise probabilities from partition samples. A software implementation of the method is available in the CaviarPD package for R.