Diffusion Model Research Articles

A tabular data generation framework guided by downstream tasks optimization

Recently, generative models have been gradually emerging into the extended dataset field, showcasing their advantages. However, when it comes to generating tabular data, these models often fail to satisfy the constraints of numerical columns, which cannot generate high-quality datasets that accurately represent real-world data and are suitable for the intended downstream applications. Responding to the challenge, we propose a tabular data generation framework guided by downstream task optimization (TDGGD). It incorporates three indicators into each time step of diffusion generation, using gradient optimization to align the generated fake data. Unlike the traditional strategy of separating the downstream task model from the upstream data synthesis model, TDGGD ensures that the generated data has highly focused columns feasibility in upstream real tabular data. For downstream task, TDGGD strikes the utility of tabular data over solely pursuing statistical fidelity. Through extensive experiments conducted on real-world tables with explicit column constraints and tables without explicit column constraints, we have demonstrated that TDGGD ensures increasing data volume while enhancing prediction accuracy. To the best of our knowledge, this is the first instance of deploying downstream information into a diffusion model framework.

Real-world image denoising via efficient diffusion model with controllable noise generation

Real-world image denoising via efficient diffusion model with controllable noise generation

Comprehensive Risk Assessment of Farmer Households in Flood and Earthquake Prone Areas Based on Multidimensional Information Diffusion Model

In recent years, the global practice of regional natural disaster reduction and comprehensive risk prevention has promoted the integration of multiple disciplines, and the theory and technology of comprehensive risk assessment for multiple natural disaster chains urgently need innovation. This article is based on multiple natural disaster chain scenarios and sets secondary indicators for multiple influencing factors based on the first level indicators composed of basic risk factors such as hazard factors, vulnerability of disaster bearing bodies, and disaster prevention and reduction capabilities. By using multidimensional information diffusion technology, multiple hazard factor hazard combination factors, vulnerability combination factors of disaster bearing bodies, and comprehensive disaster prevention and reduction capability combination factors reflecting various natural disaster scenarios are constructed, A household comprehensive risk assessment index system for multi-dimensional information diffusion assessment of natural disaster comprehensive risk was designed by quantifying risk factors such as the intensity of comprehensive disaster causing factors, vulnerability of disaster bearing bodies, and comprehensive disaster prevention and reduction capabilities in disaster scenarios. This paper expands the traditional information diffusion comprehensive risk assessment model, makes up for the lack of information on geographical units, establishes a multi impact factor multi dimension natural disaster risk comprehensive assessment model for multi risk factors of complex disaster scenarios with small samples, obtains the theoretical value of multi-dimensional information Diffusion model comprehensive risk assessment for complex disaster scenarios, breaks through the sample size limit of natural disaster risks, and makes up for the lack of information on small samples of small probability events, Improved the reliability and accuracy of comprehensive risk assessment for multiple natural disaster chains. Then, take the disaster scenario formed by the comprehensive risk of earthquake and flood in a place in the upper reaches of the Fujiang River basin as the analysis sample, calculate the comprehensive risk assessment value of flood and earthquake damage in the study area's micro samples (rural households in villages and towns). By comparison, it is found that the comprehensive risk value assessed by the multi-dimensional information Diffusion model has good reliability and high accuracy, which is significantly better than the traditional Kriging interpolation method and geographical weighted regression method. This article provides a comprehensive risk assessment of natural disasters at the micro household level, which can better reflect the diversity and regionality of risks, help clarify the hidden risks of rural natural disasters, and provide scientific basis for government emergency management decision-making.

P-182 High-resolution synthetic embryo images generated by AI

Abstract Study question How effectively can a latent diffusion model generate high-fidelity embryo images tailored to the specific contextual needs of researchers, based on the user’s text input? Summary answer The latent diffusion model successfully generated high-resolution embryo images, providing a novel approach to address data scarcity in embryo imaging. What is known already The use of AI in generating synthetic data has been increasingly explored to address the scarcity of real-world datasets in various fields. Particularly in medical imaging, AI-generated synthetic data offers a potential solution to overcome the limitations imposed by data privacy concerns and ethical considerations. Previous studies have shown that AI can replicate complex patterns in data, but its application in generating embryo images remains less explored. Study design, size, duration Single static images of 5,133 Day 5 blastocysts and 2,093 Day 3 cleavages were retrospectively collected from seven in vitro fertilization clinics between June 2011 and May 2022. The images were analyzed along with relevant metadata including clinical information and embryo grades. Day 3 embryo grading was based on the number of blastomeres, evenness, and fragmentation percentage. Day 5 grading criteria included the inner cell mass, trophectoderm, and blastocyst stage evaluations. Participants/materials, setting, methods An AI model using latent diffusion was developed to generate 10,051 Day 5 and 10,088 Day 3 synthetic embryo images. The authenticity was assessed through visual Turing tests, where embryologists discerned real from synthetic images. For the evaluation, 200 real (100 Day 5, 100 Day 3) and 200 synthetic (100 Day 5, 100 Day 3) images were randomly chosen from each dataset, ensuring a comprehensive test of the generated images’ realism. Main results and the role of chance The AI model’s efficacy in generating synthetic embryo images was assessed through visual Turing tests on Day 5 and Day 3 embryos, yielding accuracies of 0.59 and 0.57, respectively. These accuracies fall within the 99% confidence interval of near-random performance (0.41 to 0.59), highlighting the challenge in distinguishing synthetic from real images. For Day 5 embryos, the sensitivity and specificity were recorded at 0.52 and 0.66, indicating a moderate challenge in identifying synthetic images and a relatively higher ease in recognizing real ones. Day 3 embryos presented a lower sensitivity of 0.41, suggesting greater difficulty in detecting synthetic images, while the specificity increased to 0.73, indicating a stronger ability to identify real images. Collectively, with an overall accuracy of 0.58, sensitivity of 0.47, and specificity of 0.70, these findings confirm the synthetic images’ remarkable realism, closely emulating actual embryo features to the extent that differentiation by experts proved challenging. This level of realism underscores the synthetic images’ potential to significantly enrich embryological research datasets, promising advancements in the field. Limitations, reasons for caution The study’s limitation lies in the generated synthetic embryos being based on three embryo grade categories, limiting feature diversity. Additionally, there was no quantitative assessment method for these images, relying instead on expert evaluation, which required deploying a large number of embryologists. Wider implications of the findings The model’s generation of highly realistic embryo images counters data scarcity in embryological research, potentially elevating AI’s utility, enriching educational content, advancing reproductive medicine, and ensuring ethical data usage. Trial registration number not applicable

Pricing airbag option via first passage time approach

Airbag option is a new structured product that has been highly favored by investors in the over-the-counter derivatives market. It provides protections against the downside loss to investors like the ‘airbag’ in a car crash. In contrast with its popularity, there is little literature on the pricing of airbag options, especially from the mathematical finance point of view. In this paper, we study the pricing of airbag options using the first passage time approach. We propose two innovative new types of airbag options, namely airbag-TB and airbag-ER, which enhance the downside loss protections by integrating a time-varying barrier and early redemption features respectively in the payoffs. Explicit and closed-form pricing formulas for the original and the two new types of airbag options are derived under the mixed-exponential jump diffusion (MEJD) model, double exponential jump diffusion (DEJD) model and Black-Scholes (BS) model. In doing so, we significantly improve the Hybrid model proposed by Jaimungal and Sigloch (Incorporating risk and ambiguity aversion into a hybrid model of default. Math. Finance, 2012, 22, 57–81) by incorporating a stochastic redemption intensity model and more general underlying asset dynamics. We further demonstrate the performance of our pricing formulas and the two new types of airbag options with numerical examples based on Monte Carlo simulation, which verifies the accuracy and efficiency of our pricing formulas.

On the diffusive model of C<sub>60</sub> fullerene clusterization in liquids

Purpose of research. The purpose of the study is to explain the mechanisms of clustering of fullerene molecules in liquid media observed using various structural nuclear -physical methods, as well as the interpretation of experimental data within the framework of a microscopic diffusion model. Methods. The article gives a brief overview of the results of positron annihilation spectroscopy, small-angle neutron scattering, translucent electron microscopy, with which the geometric parameters of fullerene clusters in solutions were established. The theoretical method of research is the microscopic diffusion model “Limited diffusion aggregation”, which describes the kinetic processes of clusterization. Results. The diffusion limited aggregation model displays adequately the mechanism of formation of C60 molecules in the form of fractal units in the volume of fluid, observed in experiments on positron annihilation spectroscopy and small-angle neutron scattering. The structural indicators of aggregate fullerene particles in carbon disulfide, o-Xylene, toluene, water and other solvents are analyzed. The properties of various diffusion models (the reaction limited aggregation model and the diffusion limited aggregation model) are considered and their combination in relation to the assessmentsof the kinetics of fullerene aggregation in solutions. A quantitative comparison of the results of the discussed models was carried out using the example of carbon disulfide. Conclusion. The diffusion microscopic models adequately describe the phenomenon of fullerene aggregation in polar and non-polar solvents, which are recorded by various nuclear methods (positron annihilating spectroscopy and small-angle neutron scattering), the most reliable is the diffusion limited aggregation model, more than that, it is the basis of a numerical definition of structural structures parameters of units. Compared to neutron scattering, with the annihilation of positrons, in the aggregation of fullerene the [Ps–C60] molecular complexes participate in the clusterization, but this does not affect the change in the size of the cluster and the reliability of the results.

HTO and selenate diffusion through compacted Na-, Na–Ca-, and Ca-montmorillonite

Radionuclide transport in smectite clay barrier systems used for nuclear waste disposal is controlled by diffusion, with adsorption significantly retarding transport rates. While a relatively minor component of spent nuclear fuel, 79Se is a major driver of the safety case for spent fuel disposal due to its long half-life (3.3 × 105 yr) and its low adsorption to clay (KD < 10 L/kg), thus a thorough understanding of Se diffusion through clay is critical for understanding the long-term safety of spent fuel disposal systems. Through-diffusion experiments with tritiated water (HTO, conservative tracer) and Se(VI) were conducted with a well-characterized, purified montmorillonite source clay (SWy-2) under a constant ionic strength (0.1 M) and three different electrolyte compositions: Na+, Ca2+, and a Na + -Ca2+ mixture at pH 6.5 in order to probe the effects of electrolyte composition and interlayer cation composition on clay microstructure, Se(VI) aqueous speciation, and ultimately diffusion. The results were modeled using a reactive transport modeling approach to determine values of porosity (ε), De (effective diffusion coefficient), and KD (distribution coefficient for adsorption). HTO diffusive flux was higher in Ca-montmorillonite (De = 1.68 × 10−10 m2 s−1) compared to Na-montmorillonite (De = 7.83 × 10−11 m2 s−1). This increase in flux is likely due to a greater degree of clay layer stacking in the presence of Ca2+ compared to Na+, which leads to larger inter-particle pores. Overall, the Se(VI) flux was much lower than the HTO flux due to anion exclusion, with Se(VI) flux following the order Ca (De = 1.03 × 10−11 m2 s−1) > Na–Ca (De = 2.12 × 10−12 m2 s−1) > Na (De = 1.28 × 10−12 m2 s−1). These differences in Se(VI) flux are due to a combination of factors, including (1) larger accessible porosity in Ca-montmorillonite due to clay layer stacking and smaller electrostatic effects compared to Na-montmorillonite, (2) larger accessible porosity for neutral-charge CaSeO4 species which makes up 32% of aqueous Se(VI) in the pure Ca system, and (3) possibly higher Se(VI) adsorption for Ca-montmorillonite. Through a combination of experimental and modeling work, this study highlights the compounding effects that electrolyte and counterion compositions can have on radionuclide transport through clay. Diffusion models that neglect these effects are not transferable from laboratory experimental conditions to in situ repository conditions.

Synergistic adsorption of methylene blue using ternary composite of phosphoric acid geopolymer, calcium alginate, and sodium lauryl sulfate.

The removal of dyes from the aquatic ecosystem is necessary being a major threat to life. For enhanced remediation of methylene blue (MB) dye, a new ternary biopolymer-geopolymer-surfactant composite adsorbent is synthesized by combining phosphoric acid geopolymer (PAGP), calcium alginate (Alg), and sodium lauryl sulfate (SLS). During the synthesis of the composites, PAGP and SLS were mixed with the alginate matrix, producing porous hybrid beads. The PAGP-SLS-alginate (PSA) beads prepared were characterized using different analytical tools, i.e., scanning electron microscopy (SEM), Fourier transform infrared spectrophotometry (FTIR), X-ray diffractometry (XRD), surface area and porosimetery (SAP), and thermogravimetric analysis (TGA). To ascertain the ideal conditions for the adsorption process, a batch reactor procedure was used to investigate the effects of several parameters on MB adsorption, including pH (2, 4, 6, 8, 10), PSA adsorbent dosage (0.06-0.12 g), MB concentration (50-500 mg/L), contact time (15 to 300 min), and temperature (25, 35, and 45 °C). The SEM investigation indicated that ~ 1860 μm-sized PSA beads with 6-8 μm voids are generated. Based on XRD, FTIR, and SAP examinations, the material is amorphous, having numerous functional groups and an average pore size of 6.42 nm. Variation of pH has a little effect on the adsorption process, and the pH of 7.44 was found to be the pHpzc of the PSA beads. According to the findings of the batch study, equilibrium adsorption was obtained in 270-300 min, showing that the adsorption process was moderately slow-moving and effective. The dye adsorption linearly increased with initial dye concentration over concentration range of 50-500 mg/L and reciprocally decreased with rise in temperature. 0.06 g adsorbent dose, 25 °C, pH10, and 270 min were found to be the better conditions for adsorption experiments. Langmuir isotherm fitted well compared to Freundlich, Temkin, and Dubinin-Radushkevich (DR) isotherm models on the experimental data, and the maximum adsorption capacity(qmax) calculated was 1666.6 mg. g-1. Pseudo-second-order (PSO) kinetics model and multi steps (two) intra particle diffusion (IPD) model fitted well on the adsorption kinetics data. The system's entropy, Gibbs free energy, and change in enthalpy were measured and found to be -109.171 J. mol-1. K-1, - 8.198 to - 6.014 kJ. mol-1, and - 40.747 kJ. mol-1. Thermodynamics study revealed that adsorption process is exothermic, energetically favorable and resulting in the decrease in randomness. Chemisorption is found to be the dominant mechanism as confirmed by pH effect, Langmuir isotherm, PSO kinetics, IPD model, and thermodynamics parameters. PSA beads were successfully regenerated using ethanol in a course of 120 min and re-used for five times. To sum up, the PSA adsorbent's impressive adsorption capability of 1666.66 mg/g highlights its potential as a successful solution for methylene blue removal. The results of this study add to the expanding corpus of information on sophisticated adsorption materials and demonstrate PSA's potential for real-world uses in wastewater treatment and environmental clean-up.

Optical pulse-induced ultrafast antiferrodistortive transition in SrTiO3

The ultrafast dynamics of the antiferrodistortive phase transition in perovskite SrTiO3 is monitored via time-domain Brillouin scattering. Using femtosecond optical pulses, we initiate a thermally driven tetragonal-to-cubic structural transformation and detect the crystal phase through changes in the frequency of Brillouin oscillations (BO) induced by propagating acoustic phonons. Coupling the measured BO frequency with a spatiotemporal heat diffusion model, we demonstrate that, for a sample kept in the tetragonal phase, deposition of sufficient thermal energy induces a rapid transformation of the heat-affected region to the cubic phase. The initial phase change is followed by a slower reverse cubic-to-tetragonal phase transformation occurring on a timescale of hundreds of picoseconds. We attribute this ultrafast phase transformation in the perovskite to a structural resemblance between atomic displacements of the R-point soft optic mode of the cubic phase and the tetragonal phase, both characterized by anti-phase rotation of oxygen octahedra. The structural relaxation time exhibits a strong temperature dependence consistent with the prediction of the equation of motion describing collective oxygen octahedra rotation based on the energy landscape of the phenomenological Landau theory of phase transitions. Evidence of such a fast structural transition in perovskites can open up new avenues in information processing and energy storage sectors.

Reduction of organic contaminants from industrial effluent using the advanced oxidation process, chemical coagulation, and green nanotechnology

Municipal wastewater treatment systems use the chemical oxygen demand test (COD) to identify organic contaminants in industrial effluents that impede treatment due to their high concentration. This study reduced the COD levels in tannery wastewater using a multistage treatment process that included Fenton oxidation, chemical coagulation, and nanotechnology based on a synthetic soluble COD standard solution. At an acidic pH of 5, Fenton oxidation reduces the COD concentration by approximately 79%. It achieves this by combining 10 mL/L of H2O2 and 0.1 g/L of FeCl2. Furthermore, the author selected the FeCl3 coagulant for the coagulation process based on the best results of comparisons between different coagulants. At pH 8.5, the coagulation dose of 0.15 g/L achieved the maximum COD removal efficiency of approximately 56.7%. Finally, nano bimetallic Fe/Cu was used to complete the degradation and adsorption of the remaining organic pollutants. The XRD, SEM, and EDX analyses proved the formation of Fe/Cu nanoparticles. A dose of 0.09 g/L Fe/Cu NPs, 30 min of contact time, and a stirring rate of 200 rpm achieve a maximum removal efficiency of about 93% of COD at pH 7.5. The kinetics studies were analyzed using pseudo-first-order P.F.O., pseudo-second-order P.S.O., and intraparticle diffusion models. The P.S.O. showed the best fit among the kinetic models, with an R2 of 0.998. Finally, the authors recommended that technique for highly contaminated industrial effluents treatment for agriculture or industrial purposes.

Multiple diffusion models for predicting pathologic response of esophageal squamous cell carcinoma to neoadjuvant chemotherapy.

To assess the feasibility and diagnostic performance of the fractional order calculus (FROC), continuous-time random-walk (CTRW), diffusion kurtosis imaging (DKI), intravoxel incoherent motion (IVIM), mono-exponential (MEM) and stretched exponential models (SEM) for predicting response to neoadjuvant chemotherapy (NACT) in patients with esophageal squamous cell carcinoma (ESCC). This study prospectively included consecutive ESCC patients with baseline and follow up MR imaging and pathologically confirmed cT1-4aN + M0 or T3-4aN0M0 and underwent radical resection after neoadjuvant chemotherapy (NACT) between July 2019 and January 2023. Patients were divided into pCR (TRG 0) and non-pCR (TRG1 + 2 + 3) groups according to tumor regression grading (TRG). The Pre-, Post- and Delta-treatment models were built. 18 predictive models were generated according to different feature categories, based on six models by five-fold cross-validation. Areas under the curve (AUCs) of the models were compared by using DeLong method. Overall, 90 patients (71 men, 19 women; mean age, 64years ± 6 [SD]) received NACT and underwent baseline and Post-NACT esophageal MRI, with 29 patients in the pCR group and 61 patients in the non-pCR group. Among18 predictive models, The Pre-, Post-, and Delta-CTRW model showed good predictive efficacy (AUC = 0.722, 0.833 and 0.790). Additionally, the Post-FROC model (AUC = 0.907) also exhibited good diagnostic performance. Our study indicates that the CTRW model, along with the Post-FROC model, holds significant promise for the future of NACT efficacy prediction in ESCC patients.

Diffusion mechanism of fracture grouting in rock mass with flowing water

Grouting is a widely used geotechnical procedure known for its strong empirical properties. It has been extensively utilized for sealing subterranean water and strengthening soil and rock masses in their original locations in underground mining and underground space development. The diffusion mechanism of grouting in fractured rock is affected by the geological environment and slurry performance, which should be better revealed and characterized. In this paper, we take the surrounding rock of the roadway as the research object, consider the stress condition of the injected rock and time-dependent viscosity of the slurry, and establish fracture grouting diffusion model based on the rock composite fracture criterion. The grout strengthening mechanism is discussed based on a transparent fracture replica grouting test. According to the different transport characteristics of the slurry interacting with flowing water, it is divided into the following four areas grout-water balance area, grout loss area, stagnation area and water passage. The grout plugging mode is divided into two categories according to its final diffusion pattern: cross-section type water plugging mode and water-drop plugging mode. The pressure monitoring data of the test verified the two grout plugging modes. The diffusion distance of cis-reverse water derived in this study was in good agreement with the experimental results in grouting early stage. The research results can provide a theoretical basis for grouting engineering design.

SeisResoDiff: Seismic resolution enhancement based on a diffusion model

High resolution of post-stack seismic data assists in better interpretation of subsurface structures as well as high accuracy of impedance inversion. Therefore, geophysicists consistently strive to acquire higher resolution seismic images in petroleum exploration. Although there have been successful applications of conventional signal processing and machine learning for post-stack seismic resolution enhancement, there is limited reference to the seismic applications of the recent emergence and rapid development of generative artificial intelligence. Hence, we propose to apply diffusion models, among the most popular generative models, to enhance seismic resolution. Specifically, we apply the classic diffusion model—denoising diffusion probabilistic model (DDPM), conditioned on the seismic data in low resolution, to reconstruct corresponding high-resolution images. Herein the entire scheme is referred to as SeisResoDiff. To provide a comprehensive and clear understanding of SeisResoDiff, we introduce the basic theories of diffusion models and detail the optimization objective's derivation with the aid of diagrams and algorithms. For implementation, we first propose a practical workflow to acquire abundant training data based on the generated pseudo-wells. Subsequently, we apply the trained model to both synthetic and field datasets, evaluating the results in three aspects: the appearance of seismic sections and slices in the time domain, frequency spectra, and comparisons with the synthetic data using real well-logging data at the well locations. The results demonstrate not only effective seismic resolution enhancement, but also additional denoising by the diffusion model. Experimental comparisons indicate that training the model on noisy data, which are more realistic, outperforms training on clean data. The proposed scheme demonstrates superiority over some conventional methods in high-resolution reconstruction and denoising ability, yielding more competitive results compared to our previous research.

Comparison of diagnostic performance between diffusion models parameters and mono-exponential apparent diffusion coefficient in patients with prostate cancer: A systematic review and meta-analysis

Background: The importance of diffusion in prostate cancer (PCa) diagnosis has been widely proven. Several studies investigated diffusion models in PCa diagnosis. Materials and Methods: This systematic review and meta-analysis study was performed to evaluate the ability of three diffusion models to diagnose PCa from the scientific electronic databases Embase, PubMed, Scopus, and Web of Science (ISI) for the period up to March 2022 to identify all relevant articles. Results: Eighteen studies were included in the systematic review section (7 diffusion kurtosis imaging [DKI] studies, 4 diffusion tensor imaging [DTI] studies, 4 intravoxel incoherent motion [IVIM] studies, and 3 IVIM-DKI studies). Pooled sensitivity, specificity, accuracy, and summary area under each diffusion model’s curve (AUC) and 95% confidence intervals (CIs) were calculated. The pooled accuracy and 95% CI on detection (differentiation of tumor from normal tissue and benign prostatic hyperplasia/prostatitis) were obtained for apparent diffusion coefficient (ADC) at 87.97% (84.56%–91.38%) for DKI parameters (Gaussian diffusion [DK] 87.94% [78.71%–97.16%] and deviation from Gaussian diffusion [K] 86.84% [81.83%–91.85%]) and IVIM parameters (true molecular diffusion [DIVIM] 81.73% [72.54%–90.91%], perfusion-related diffusion [D*] 65% [48.47%–81.53%] and perfusion fraction [f] 80.36% [64.23%–96.48%]). The AUC values and 95% CI in the detection of PCa were obtained for ADC at 0.95 (0.92–0.97), for DKI parameters (DK 0.94 [0.89–0.99] and K 0.93 [0.90–0.96]) and for IVIM parameters (DIVIM 0.85 [0.80–0.91], D* 0.60 [0.43–0.77] and f 0.73 [0.63–0.84]). Two studies showed that the DTI accuracy values were 97.34% and 85%. For IVIM–kurtosis model in PCa detection, two studies stated that the DIVIM-K and KIVIM-K accuracy values were 85% and 84.44% (the pooled accuracy; 84.64% with 95% CI 75.78%–93.50%), and 72.50% and 71.11% (the pooled accuracy, 72.10% with 95% CI 64.73%–79.48%), respectively. Conclusion: Our findings showed that among the DKI, IVIM, and ADC parameters, it seems that ADC, Dk, DIVIM, and K are the most important, which can be used as an indicator to distinguish PCa from normal tissue. The DKI model probably has a higher ability to detect PCa from normal tissue than the IVIM model. DKI probably has the same diagnostic performance in PCa detection and grading compared to diffusion-weighted imaging and ADC.

Target-Guided Diffusion Models for Unpaired Cross-Modality Medical Image Translation.

In a clinical setting, the acquisition of certain medical image modality is often unavailable due to various considerations such as cost, radiation, etc. Therefore, unpaired cross-modality translation techniques, which involve training on the unpaired data and synthesizing the target modality with the guidance of the acquired source modality, are of great interest. Previous methods for synthesizing target medical images are to establish one-shot mapping through generative adversarial networks (GANs). As promising alternatives to GANs, diffusion models have recently received wide interests in generative tasks. In this paper, we propose a target-guided diffusion model (TGDM) for unpaired cross-modality medical image translation. For training, to encourage our diffusion model to learn more visual concepts, we adopted a perception prioritized weight scheme (P2W) to the training objectives. For sampling, a pre-trained classifier is adopted in the reverse process to relieve modality-specific remnants from source data. Experiments on both brain MRI-CT and prostate MRI-US datasets demonstrate that the proposed method achieves a visually realistic result that mimics a vivid anatomical section of the target organ. In addition, we have also conducted a subjective assessment based on the synthesized samples to further validate the clinical value of TGDM.

Comparative Analysis of Water-Induced Response in 3D-Printed SCF/ABS Composites under Controlled Diffusion

Additive manufacturing (AM) or 3D-printing of fiber-reinforced composites (FRCs) has garnered significant interests for its versatility in creating intricate parts and rapid prototyping due to cost-effectiveness. Although short fiber-reinforced thermoplastic composites are challenging to manufacture, their mechanical properties can be easily tailored by adjusting fiber type, orientation, and volume fraction. However, void formation during printing is a key issue, impacting mechanical properties and facilitating water ingression, affecting long-term durability. This work studies water diffusion characteristics and the associated hydro-aging of 3D-printed short carbon fiber (SCF)/acrylonitrile butadiene styrene (ABS) composites with controlled water diffusion. Effects of material type (ABS and SCF/ABS), 3D-printing path (horizontal and vertical filament orientation), and diffusion surface (uni-directional and bi-directional diffusion) on water diffusion coefficient and maximum water absorption level are characterized to ensure the long-term durability of 3D-printed ABS and SCF/ABS composites. Baseline representative volume element-based finite element (RVE-FE) diffusion models were developed based on micro-computed tomography (micro-CT) image analysis to understand water diffusion characteristics. This work proves that the SCF/ABS composite is more resistive to hydro-aging than neat ABS due to the SCFs’ hydrophobic nature. SCF/ABS composites, while providing distinct advantages over pure ABS in terms of mechanical properties, could also be more effective against water environments.

Sorption kinetics of vermiculite-K2CO3 in thermochemical energy storage: Model evaluation and mechanism exploration

Thermal energy storage technology is crucial for building heating and domestic hot water supply, playing an essential role in optimizing energy utilization. Hydrated salts have emerged as prominent materials for thermochemical energy storage (TCES), with their sorption kinetics directly impacting thermal energy conversion efficiency. This study experimentally analyzes the microstructure and sorption performance of the EV/K2CO3 composite sorbent. Numerical analysis methods are employed to evaluate the validity of various sorption kinetics models and to elucidate the sorption mechanism and potential limitations of the composite sorbent. The findings indicate that EV can disperse K2CO3 salt particles while preserving a porous structure, thereby improving the sorption kinetics efficiency and stability of the EV/K2CO3 composite sorbent. At low relative pressure, EV/K2CO3 demonstrates a chemical adsorption mechanism, with nucleation models accurately predicting its sorption behavior, primarily limited by the nucleation and growth of hydrated salt crystals. At moderate relative pressure, EV/K2CO3 exhibits a mechanism involving both chemical adsorption and solution absorption, with diffusion models and the first-order model showing higher prediction accuracy, primarily limited by diffusion within or outside the product layer. At high relative pressure, EV/K2CO3 primarily undergoes solution absorption, with geometric contraction models and the first-order model displaying good predictive performance, primarily limited by the reaction phase interface. This study enhances comprehension of the sorption mechanism of the EV/K2CO3 composite, providing a scientific foundation for the development of high-performance TCES materials suitable for building and environmental applications.

Automating computational design with generative AI

AbstractAI image generators based on diffusion models have recently garnered attention for their capability to create images from simple text prompts. However, for practical use in civil engineering they need to be able to create specific construction plans for given constraints. This paper investigates the potential of current AI generators in addressing such challenges, specifically for the creation of simple floor plans. We explain how the underlying diffusion‐models work and propose novel refinement approaches to improve semantic encoding and generation quality. In several experiments we show that we can improve validity of generated floor plans from 6% to 90%. Based on these results we derive future research challenges considering building information modeling. With this we provide: (i) evaluation of current generative AIs; (ii) propose improved refinement approaches; (iii) evaluate them on various examples; (iv) derive future directions for diffusion models in civil engineering.

Generation of realistic children's book images based on diffusion models

Generation of realistic children's book images based on diffusion models

Generative adversarial network-based reconstruction of healthy anatomy for anomaly detection in brain CT scans.

To help radiologists examine the growing number of computed tomography (CT) scans, automatic anomaly detection is an ongoing focus of medical imaging research. Radiologists must analyze a CT scan by searching for any deviation from normal healthy anatomy. We propose an approach to detecting abnormalities in axial 2D CT slice images of the brain. Although much research has been done on detecting abnormalities in magnetic resonance images of the brain, there is little work on CT scans, where abnormalities are more difficult to detect due to the low image contrast that must be represented by the model used. We use a generative adversarial network (GAN) to learn normal brain anatomy in the first step and compare two approaches to image reconstruction: training an encoder in the second step and using iterative optimization during inference. Then, we analyze the differences from the original scan to detect and localize anomalies in the brain. Our approach can reconstruct healthy anatomy with good image contrast for brain CT scans. We obtain median Dice scores of 0.71 on our hemorrhage test data and 0.43 on our test set with additional tumor images from publicly available data sources. We also compare our models to a state-of-the-art autoencoder and a diffusion model and obtain qualitatively more accurate reconstructions. Without defining anomalies during training, a GAN-based network was used to learn healthy anatomy for brain CT scans. Notably, our approach is not limited to the localization of hemorrhages and tumors and could thus be used to detect structural anatomical changes and other lesions.