VinDr-Mammo: A large-scale benchmark dataset for computer-aided diagnosis in full-field digital mammography.

Mammography, or breast X-ray imaging, is the most widely used imaging modality to detect cancer and other breast diseases. Recent studies have shown that deep learning-based computer-assisted detection and diagnosis (CADe/x) tools have been developed to support physicians and improve the accuracy of interpreting mammography. A number of large-scale mammography datasets from different populations with various associated annotations and clinical data have been introduced to study the potential of learning-based methods in the field of breast radiology. With the aim to develop more robust and more interpretable support systems in breast imaging, we introduce VinDr-Mammo, a Vietnamese dataset of digital mammography with breast-level assessment and extensive lesion-level annotations, enhancing the diversity of the publicly available mammography data. The dataset consists of 5,000 mammography exams, each of which has four standard views and is double read with disagreement (if any) being resolved by arbitration. The purpose of this dataset is to assess Breast Imaging Reporting and Data System (BI-RADS) and breast density at the individual breast level. In addition, the dataset also provides the category, location, and BI-RADS assessment of non-benign findings. We make VinDr-Mammo publicly available as a new imaging resource to promote advances in developing CADe/x tools for mammography interpretation.

PediCXR: An open, large-scale chest radiograph dataset for interpretation of common thoracic diseases in children.

Computer-aided diagnosis systems in adult chest radiography (CXR) have recently achieved great success thanks to the availability of large-scale, annotated datasets and the advent of high-performance supervised learning algorithms. However, the development of diagnostic models for detecting and diagnosing pediatric diseases in CXR scans is undertaken due to the lack of high-quality physician-annotated datasets. To overcome this challenge, we introduce and release PediCXR, a new pediatric CXR dataset of 9,125 studies retrospectively collected from a major pediatric hospital in Vietnam between 2020 and 2021. Each scan was manually annotated by a pediatric radiologist with more than ten years of experience. The dataset was labeled for the presence of 36 critical findings and 15 diseases. In particular, each abnormal finding was identified via a rectangle bounding box on the image. To the best of our knowledge, this is the first and largest pediatric CXR dataset containing lesion-level annotations and image-level labels for the detection of multiple findings and diseases. For algorithm development, the dataset was divided into a training set of 7,728 and a test set of 1,397. To encourage new advances in pediatric CXR interpretation using data-driven approaches, we provide a detailed description of the PediCXR data sample and make the dataset publicly available on https://physionet.org/content/vindr-pcxr/1.0.0/ .

Subclinical systolic dysfunction detected by 2D speckle tracking echocardiography in adults with diabetes mellitus: systematic review and meta-analysis of 6668 individuals with diabetes mellitus and 7218 controls.

PURPOSE: Speckle tracking echocardiography (STE) can help to identify subclinical features of diabetic cardiomyopathy (DCM). There is, however, significant heterogeneity in the reported strain values in literature. We performed a systematic review and meta-analysis to compare cardiac systolic strain values assessed by 2D-STE in asymptomatic adults with diabetes mellitus (DM) and healthy controls. METHODS: Five databases were searched, and a total of 41 valid studies (6668 individuals with DM and 7218 controls) were included for analysis. Pooled mean in each group and mean difference (MD) for left ventricular global longitudinal strain (LVGLS), LV global circumferential strain (LVGCS), LV global radial strain (LVGRS), LV longitudinal systolic strain rate (LVSR), left atrial reservoir strain (LARS) and right ventricular GLS (RVGLS) were assessed. RESULTS: Patients with DM had overall 2 units lower LVGLS than healthy subjects 17.5% [16.8, 18.3], vs 19.5 [18.7, 20.4], MD = - 1.96 [- 2.27, - 1.64]. Other strain values were also lower in patients with DM: LVGCS (MD = - 0.89 [- 1.26, - 0.51]); LVGRS (MD = - 5.03 [- 7.18, - 2.87]); LVSR (MD = - 0.06 [- 0.10, - 0.03]); LARS (MD = - 8.41 [- 11.5, - 5.33]); and RVGLS (MD = - 2.41 [- 3.60, - 1.22]). Meta-regression identified higher body mass index (BMI) as the single contributor to worse LVGLS, LVGCS and LVSR. Those with higher Hemoglobulin A1c had worse RVGLS. CONCLUSION: Myocardial strains were reduced in whole heart in patients with DM. The largest reduction was observed in LA reservoir strain, followed by RVGLS and LVGLS. Higher BMI in patients with DM is associated with worse LV strain values.

Learning to diagnose common thorax diseases on chest radiographs from radiology reports in Vietnamese.

Deep learning, in recent times, has made remarkable strides when it comes to impressive performance for many tasks, including medical image processing. One of the contributing factors to these advancements is the emergence of large medical image datasets. However, it is exceedingly expensive and time-consuming to construct a large and trustworthy medical dataset; hence, there has been multiple research leveraging medical reports to automatically extract labels for data. The majority of this labor, however, is performed in English. In this work, we propose a data collecting and annotation pipeline that extracts information from Vietnamese radiology reports to provide accurate labels for chest X-ray (CXR) images. This can benefit Vietnamese radiologists and clinicians by annotating data that closely match their endemic diagnosis categories which may vary from country to country. To assess the efficacy of the proposed labeling technique, we built a CXR dataset containing 9,752 studies and evaluated our pipeline using a subset of this dataset. With an F1-score of at least 0.9923, the evaluation demonstrates that our labeling tool performs precisely and consistently across all classes. After building the dataset, we train deep learning models that leverage knowledge transferred from large public CXR datasets. We employ a variety of loss functions to overcome the curse of imbalanced multi-label datasets and conduct experiments with various model architectures to select the one that delivers the best performance. Our best model (CheXpert-pretrained EfficientNet-B2) yields an F1-score of 0.6989 (95% CI 0.6740, 0.7240), AUC of 0.7912, sensitivity of 0.7064 and specificity of 0.8760 for the abnormal diagnosis in general. Finally, we demonstrate that our coarse classification (based on five specific locations of abnormalities) yields comparable results to fine classification (twelve pathologies) on the benchmark CheXpert dataset for general anomaly detection while delivering better performance in terms of the average performance of all classes.

A novel multi-view deep learning approach for BI-RADS and density assessment of mammograms.

Advanced deep learning (DL) algorithms may predict the patient's risk of developing breast cancer based on the Breast Imaging Reporting and Data System (BI-RADS) and density standards. Recent studies have suggested that the combination of multi-view analysis improved the overall breast exam classification. In this paper, we propose a novel multi-view DL approach for BI-RADS and density assessment of mammograms. The proposed approach first deploys deep convolutional networks for feature extraction on each view separately. The extracted features are then stacked and fed into a Light Gradient Boosting Machine (LightGBM) classifier to predict BI-RADS and density scores. We conduct extensive experiments on both the internal mammography dataset and the public dataset Digital Database for Screening Mammogra-phy (DDSM). The experimental results demonstrate that the proposed approach outperforms the single-view classification approach on two benchmark datasets by huge F1-score margins (+5% on the internal dataset and +10% on the DDSM dataset). These results highlight the vital role of combining multi-view information to improve the performance of breast cancer risk prediction.

VinDr-CXR: An open dataset of chest X-rays with radiologist's annotations.

Most of the existing chest X-ray datasets include labels from a list of findings without specifying their locations on the radiographs. This limits the development of machine learning algorithms for the detection and localization of chest abnormalities. In this work, we describe a dataset of more than 100,000 chest X-ray scans that were retrospectively collected from two major hospitals in Vietnam. Out of this raw data, we release 18,000 images that were manually annotated by a total of 17 experienced radiologists with 22 local labels of rectangles surrounding abnormalities and 6 global labels of suspected diseases. The released dataset is divided into a training set of 15,000 and a test set of 3,000. Each scan in the training set was independently labeled by 3 radiologists, while each scan in the test set was labeled by the consensus of 5 radiologists. We designed and built a labeling platform for DICOM images to facilitate these annotation procedures. All images are made publicly available in DICOM format along with the labels of both the training set and the test set.

Phase recognition in contrast-enhanced CT scans based on deep learning and random sampling.

PURPOSE: A fully automated system for interpreting abdominal computed tomography (CT) scans with multiple phases of contrast enhancement requires an accurate classification of the phases. Current approaches to classify the CT phases are commonly based on three-dimensional (3D) convolutional neural network (CNN) approaches with high computational complexity and high latency. This work aims at developing and validating a precise, fast multiphase classifier to recognize three main types of contrast phases in abdominal CT scans. METHODS: We propose in this study a novel method that uses a random sampling mechanism on top of deep CNNs for the phase recognition of abdominal CT scans of four different phases: noncontrast, arterial, venous, and others. The CNNs work as a slicewise phase prediction, while random sampling selects input slices for the CNN models. Afterward, majority voting synthesizes the slicewise results of the CNNs to provide the final prediction at the scan level. RESULTS: Our classifier was trained on 271 426 slices from 830 phase-annotated CT scans, and when combined with majority voting on 30% of slices randomly chosen from each scan, achieved a mean F1 score of 92.09% on our internal test set of 358 scans. The proposed method was also evaluated on two external test sets: CTPAC-CCRCC (N = 242) and LiTS (N = 131), which were annotated by our experts. Although a drop in performance was observed, the model performance remained at a high level of accuracy with a mean F1 scores of 76.79% and 86.94% on CTPAC-CCRCC and LiTS datasets, respectively. Our experimental results also showed that the proposed method significantly outperformed the state-of-the-art 3D approaches while requiring less computation time for inference. CONCLUSIONS: In comparison to state-of-the-art classification methods, the proposed approach shows better accuracy with significantly reduced latency. Our study demonstrates the potential of a precise, fast multiphase classifier based on a two-dimensional deep learning approach combined with a random sampling method for contrast phase recognition, providing a valuable tool for extracting multiphase abdomen studies from low veracity, real-world data.

Normal Ranges of Global Left Ventricular Myocardial Work Indices in Adults: A Meta-Analysis.

BACKGROUND: Recent studies have demonstrated that left ventricular myocardial work (MW) is incremental in diagnosis and prognostication compared with left ventricular ejection fraction and myocardial strain. The authors performed a meta-analysis of normal ranges of noninvasive MW indices including global work index, global constructive work, global wasted work, and global work efficiency and determined confounders that may contribute to variance in reported values. METHODS: Four databases (PubMed, Scopus, Embase, and the Cochrane Library) were searched through January 2021 using the key terms "myocardial work," "global constructive work," "global wasted work," "global work index," and "global work efficiency." Studies were included if the articles reported LV MW using two-dimensional transthoracic echocardiography in healthy normal subjects, either in a control group or comprising the entire study cohort. The weighted mean was estimated by using the random-effect model with a 95% CI. Heterogeneity across included studies was assessed using the I2 test. Funnel plots and the Egger regression test were used to assess potential publication bias. RESULTS: The search yielded 476 articles. After abstract and full-text screening, we included 13 data sets with 1,665 patients for the meta-analysis. The reported normal mean values of global work index and global constructive work among the studies were 2,010 mm Hg% (95% CI, 1,907-2,113 mm Hg%) and 2,278 mm Hg% (95% CI, 2,186-2,369 mm Hg%), respectively. Mean global wasted work was 80 mm Hg% (95% CI, 73-87 mm Hg%), and mean global work efficiency was 96.0% (95% CI, 96%-96%). Furthermore, gender significantly contributed to variations in normal values of global work index, global wasted work, and global work efficiency. No evidence of significant publication bias was observed. CONCLUSIONS: In this meta-analysis, the authors provide echocardiographic reference ranges for noninvasive indices of MW. These normal values could serve as a reference for clinical and research use.

Three-Dimensional Global Left Ventricular Myocardial Strain Reduced in All Directions in Subclinical Diabetic Cardiomyopathy: A Systematic Review and Meta-Analysis.

Background Three-dimensional (3D) speckle tracking echocardiography can identify subclinical diabetic cardiomyopathy without geometric assumption and loss of speckle from out-of-plane motions. There is, however, significant heterogeneity among the previous reports. We performed a systematic review and meta-analysis to compare 3D strain values between adults with asymptomatic, subclinical diabetes mellitus (ie, patients with diabetes mellitus without known clinical manifestations of cardiac disease) and healthy controls. Methods and Results After systematic review of 5 databases, 12 valid studies (544 patients with diabetes mellitus and 489 controls) were eligible for meta-analysis. Pooled means and mean difference (MD) using a random-effects model for 3D global longitudinal, circumferential, radial, and area strain were calculated. Patients with diabetes mellitus had an overall 2.31 percentage points lower 3D global longitudinal strain than healthy subjects (16.6%, 95% CI, 15.7-17.6 versus 19.0; 95% CI, 18.2-19.7; MD, -2.31, 95% CI, -2.72 to -2.03). Similarly, 3D global circumferential strain (18.9%; 95% CI, 17.5-20.3 versus 20.5; 95% CI, 18.9-22.1; MD, -1.50; 95% CI, -2.09 to -0.91); 3D global radial strain (44.6%; 95% CI, 40.2-49.1 versus 48.2; 95% CI, 44.7-51.8; MD, -3.47; 95% CI, -4.98 to -1.97), and 3D global area strain (30.5%; 95% CI, 29.2-31.8 versus 32.4; 95% CI, 30.5-34.3; MD, -1.76; 95% CI, -2.74 to -0.78) were also lower in patients with diabetes mellitus. Significant heterogeneity was noted between studies for all strain directions (inconsistency factor [I2], 37%-78%). Meta-regression in subgroup analysis of studies using the most popular vendor found higher prevalence of hypertension as a significant contributor to worse 3D global longitudinal strain. Higher hemoglobulin A1c was the most significant contributor to worse 3D global circumferential strain in patients with diabetes mellitus. Conclusions Three-dimensional myocardial strain was reduced in all directions in asymptomatic diabetic patients. Hypertension and hemoglobin A1c were associated with worse 3D global longitudinal strain and 3D global circumferential strain, respectively. Registration URL: https://www.crd.york.ac.uk/prospero; unique identifier: CRD42020197825.

Solvation force and adsorption isotherm of a fluid mixture in nanopores of complex geometry based on fundamental measure theory.

A novel method based on the fundamental measure theory is developed to calculate the solvation force and adsorption isotherm of a Lennard-Jones fluid mixture in complex geometries. Fast Fourier transform and 3D-voxel discretization are used for accurately computing the confined fluid densities in a closed pore of arbitrary geometry. Given the fluid densities, the solvation force distribution at the solid surface can be calculated using a new formulation from either mechanical or thermodynamic approach. Understanding the solvation force behavior, which depends on many factors such as pore geometry, confined density distribution, molecule size, is very important to analyze the pore deformation from a poromechanical point of view. Special attention in the numerical simulations is given to the adsorption problem of CH4and CO2gas mixture in ellipsoidal pore.

Reliability enhancement with dependable model predictive control.

Operational Technology (OT) systems are merging towards a conjoint architecture with the advances in communication networks and emerging standards such as IEC/IEEE 60802 for industrial automation, automotive, power and energy and other areas. In this paper, we present a Dependable Control System (DepCS) with Model Predictive Control (MPC) algorithm that works in such architectures using multiple MPC controllers (of a feedback control loop) to enhance the operational reliability. We termed this as Dependable Model Predictive Control (DepMPC) system. The reliability enhancement of a DepMPC system is achievable thanks to the fault-tolerance of multiple MPC controllers and the tractable information flows with Time-Sensitive Networking (TSN). Here, our discussion was focused only on the logical connectivity and not the hardware architecture. The numerical simulations are studied with three multi-variable plants that have control constraints. In this study, we introduced a Replacement Controller (RC) to improve the control performance of the DepMPC system. The combination of both the Replacement Controller and Dependable Model Predictive Control (RC-DepMPC) system proves a promising solution for actual implementations.

Unveiling prognostics biomarkers of tyrosine metabolism reprogramming in liver cancer by cross-platform gene expression analyses.

Tyrosine is mainly degraded in the liver by a series of enzymatic reactions. Abnormal expression of the tyrosine catabolic enzyme tyrosine aminotransferase (TAT) has been reported in patients with hepatocellular carcinoma (HCC). Despite this, aberration in tyrosine metabolism has not been investigated in cancer development. In this work, we conduct comprehensive cross-platform study to obtain foundation for discoveries of potential therapeutics and preventative biomarkers of HCC. We explore data from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), Gene Expression Profiling Interactive Analysis (GEPIA), Oncomine and Kaplan Meier plotter (KM plotter) and performed integrated analyses to evaluate the clinical significance and prognostic values of the tyrosine catabolic genes in HCC. We find that five tyrosine catabolic enzymes are downregulated in HCC compared to normal liver at mRNA and protein level. Moreover, low expression of these enzymes correlates with poorer survival in patients with HCC. Notably, we identify pathways and upstream regulators that might involve in tyrosine catabolic reprogramming and further drive HCC development. In total, our results underscore tyrosine metabolism alteration in HCC and lay foundation for incorporating these pathway components in therapeutics and preventative strategies.

Pooled summary of native T1 value and extracellular volume with MOLLI variant sequences in normal subjects and patients with cardiovascular disease.

T1 mapping by cardiac magnetic resonance (CMR) allows detection of abnormal myocardium. A number of myocardial abnormalities affects the signal captured in T1 mapping. We performed a systematic review and meta-analysis of native T1 and extracellular volume (ECV) in subjects with and without cardiac disease (1) to determine the normal ranges of T1 values and ECV by sequences as well as parameters influencing them, and (2) to summarize the differences in T1 values and ECV of the diseases relative to the normal ranges. Three databases (EMBASE, SCOPUS, and MEDLINE) were systematically searched for native T1 time and ECV. Only human studies with a sample size of ≥ 20 subjects were included. A random effect model was used to pool data. The 69 selected articles included 1954 healthy subjects and 3186 with disease. T1 of normal healthy was different among MOLLI variants: in 1.5T sequences, ShMOLLI had the shortest (944 ms [95% confidence interval 925, 963]), followed by MOLLI 3(3)3(3)5 flip-angle 50°, 967 [959, 975] and flip-angle 35°, 969 [951, 988]. 3T had longer T1 than 1.5T by approximately 100-200 ms. ECV of the normal healthy was consistent among the studies (ranging from 25 to 27%), irrespective of subjects' factors, sequences, vendors, and contrast type. Many diseases demonstrated longer native T1 than normal subjects, but T1 was shorter in Fabry disease and iron overload. In contrast, all disease states showed either normal or increased ECV. Diagnostic accuracy of native T1 time was minimally affected by the difference in the sequences. ECV is less influenced by methodology than T1 time among normal subjects. Different myocardial diseases are associated with shorter or longer T1 times, whereas ECV is consistently increased independent of the underlying pathophysiology.

Air pollution and telomere length: a systematic review of 12,058 subjects.

BACKGROUND: Over recent decades, adverse effects of ambient air pollution on the cardiovascular system have been clearly demonstrated. However, the underlying mechanisms are not fully elucidated. Air pollution may accelerates biological aging and thereby the susceptibility to cardiovascular diseases (CVDs). Telomeres are tandem repetitive DNA complexes that play a critical role in maintaining chromosome stability. There are, however, heterogeneities among the reported effects of air pollution on telomere. This study sought to evaluate the existing literature on the association between air pollution and telomere length (TL). METHODS: Two reviewers independently searched on electronic databases including PUBMED, EMBASE, SCOPUS, WEB OF SCIENCE and Ovid. The key terms were "air pollution" and "telomere" without language restriction. Articles relating to tobacco smoke were excluded. RESULTS: A total of 12,058 subjects from 25 articles remained for final review. All were observational studies: 14 cross-sectional, 6 cohort and 5 case-control studies. Nineteen (76%) assessed leukocyte telomere length (LTL) of which 15 found associations between air pollution and shorter TL, 2 with longer TL, 1 had mixed results, and a study of patients with type 2 diabetes found non-significant associations with TL. One found longer TL from saliva. The remaining studies were of placental cells, buccal cells or sperm and all reported shorter TL associated with air pollution. Particulate matter (PM) was investigated in 8 articles, and the remainder assessed black carbon (BC), benzene, lead, cadmium and polycyclic aromatic hydrocarbon (PAH). Geographically, 11 studies were conducted in Europe, with 10 in Asia and 4 in North America. While all followed Cawthon's protocol for TL assessment, discordance in the reporting formats did not allow us to perform a quantitative meta-analysis. CONCLUSIONS: Most of the studies support the association of shorter TL with air pollution. Uniform reporting format would be warranted for future studies to estimate true effect size of air pollution on TL.

CNN-Based Projected Gradient Descent for Consistent CT Image Reconstruction.

We present a new image reconstruction method that replaces the projector in a projected gradient descent (PGD) with a convolutional neural network (CNN). Recently, CNNs trained as image-to-image regressors have been successfully used to solve inverse problems in imaging. However, unlike existing iterative image reconstruction algorithms, these CNN-based approaches usually lack a feedback mechanism to enforce that the reconstructed image is consistent with the measurements. We propose a relaxed version of PGD wherein gradient descent enforces measurement consistency, while a CNN recursively projects the solution closer to the space of desired reconstruction images. We show that this algorithm is guaranteed to converge and, under certain conditions, converges to a local minimum of a non-convex inverse problem. Finally, we propose a simple scheme to train the CNN to act like a projector. Our experiments on sparse-view computed-tomography reconstruction show an improvement over total variation-based regularization, dictionary learning, and a state-of-the-art deep learning-based direct reconstruction technique.

MRI-Derived Myocardial Strain Measures in Normal Subjects.

OBJECTIVES: The aim of this study was to perform a systematic review and meta-analysis to estimate the normal ranges of magnetic resonance imaging (MRI)-based feature tracking (FT) and to identify sources of variations. Similar analyses were also performed for strain encoding, displacement encoding with stimulated echoes, and myocardial tagging. BACKGROUND: MRI-FT is a novel technique for quantification of myocardial deformation using MRI cine images. However, the reported 95% confidence intervals (CIs) from the 2 largest studies have no overlaps. METHODS: Four databases (EMBASE, SCOPUS, PUBMED, and Web of Science) were systematically searched for MRI strains of the left (LV) and right (RV) ventricles. The key terms for MRI-FT were "tissue tracking," "feature tracking," "cardiac magnetic resonance," "cardiac MRI," "CMR," and "strain." A random effects model was used to pool LV global longitudinal strain (GLS), global circumferential strain (GCS), global radial strain (GRS), and RVGLS. Meta-regressions were used to identify the sources of variations. RESULTS: 659 healthy subjects were included from 18 papers for MRI-FT. Pooled mean of LVGLS was -20.1% (95% CI: -20.9% to -19.3%), LVGCS -23% (95% CI: -24.3% to -21.7%), LVGRS 34.1% (95% CI: 28.5% to 39.7%), and RVGLS -21.8% (95% CI: -23.3% to -20.2%). Although there were no publication biases except for LVGCS, significant heterogeneities were found. Meta-regression showed that variation of LVGCS was associated with field strength (ß = 3.2; p = 0.041). Variations of LVGLS, LVGRS, and RVGLS were not associated with any of age, sex, software, field strength, sequence, LV ejection fraction, or LV size. LVGCS seems the most robust in MRI-FT. Among the MRI-derived strain techniques, the normal ranges were mostly concordant in LVGLS and LVGCS but varied substantially in LVGRS and RVGLS. CONCLUSIONS: The pooled means of 4 MRI-derived myocardial strain methods in normal subjects are demonstrated. Differences in field strength were attributed to variations of LVGCS.

Perturbed cooperative-state feedback strategy for model predictive networked control of interconnected systems.

A perturbed cooperative-state feedback (PSF) strategy is presented for the control of interconnected systems in this paper. The subsystems of an interconnected system can exchange data via the communication network that has multiple connection topologies. The PSF strategy can resolve both issues, the sensor data losses and the communication network breaks, thanks to the two components of the control including a cooperative-state feedback and a perturbation variable, e.g., ui=Kijxj+wi. The PSF is implemented in a decentralized model predictive control scheme with a stability constraint and a non-monotonic storage function (ΔV(x(k))≥0), derived from the dissipative systems theory. Numerical simulation for the automatic generation control problem in power systems is studied to illustrate the effectiveness of the presented PSF strategy.

Dependable control systems with Internet of Things.

This paper presents an Internet of Things (IoT)-enabled dependable control system (DepCS) for continuous processes. In a DepCS, an actuator and a transmitter form a regulatory control loop. Each processor inside such actuator and transmitter is designed as a computational platform implementing the feedback control algorithm. The connections between actuators and transmitters via IoT create a reliable backbone for a DepCS. The centralized input-output marshaling system is not required in DepCSs. A state feedback control synthesis method for DepCS applying the self-recovery constraint is presented in the second part of the paper.

Inverse rendering of Lambertian surfaces using subspace methods.

We propose a vector space approach for inverse rendering of a Lambertian convex object with distant light sources. In this problem, the texture of the object and arbitrary lightings are both to be recovered from multiple images of the object and its 3D model. Our work is motivated by the observation that all possible images of a Lambertian object lie around a low-dimensional linear subspace spanned by the first few spherical harmonics. The inverse rendering can therefore be formulated as a matrix factorization, in which the basis of the subspace is encoded in a spherical harmonic matrix S associated with the object's geometry. A necessary and sufficient condition on S for unique factorization is derived with an introduction to a new notion of matrix rank called nonseparable full rank. A singular value decomposition-based algorithm for exact factorization in the noiseless case is introduced. In the presence of noise, two algorithms, namely, alternating and optimization based are proposed to deal with two different types of noise. A random sample consensus-based algorithm is introduced to reduce the size of the optimization problem, which is equal to the number of pixels in each image. Implementations of the proposed algorithms are done on a real data set.

Density functional theory study on mechanisms of epoxy-phenol curing reaction.

A comprehensive picture on the mechanism of the epoxy-phenol curing reactions is presented using the density functional theory B3LYP/ 6-31G(d,p) and simplified physical molecular models to examine all possible reaction pathways. Phenol can act as its own promoter by using an addition phenol molecule to stabilize the transition states, and thus lower the rate-limiting barriers by 27.0-48.9 kJ/mol. In the uncatalyzed reaction, an epoxy ring is opened by a phenol with an apparent barrier of about 129.6 kJ/mol. In catalyzed reaction, catalysts facilitate the epoxy ring opening prior to curing that lowers the apparent barriers by 48.9-50.6 kJ/mol. However, this can be competed in highly basic catalysts such as amine-based catalysts, where catalysts are trapped in forms of hydrogen-bonded complex with phenol. Our theoretical results predict the activation energy in the range of 79.0-80.7 kJ/mol in phosphine-based catalyzed reactions, which agrees well with the reported experimental range of 54-86 kJ/mol.