First-order Statistics Research Articles

Radiomics-Based Diagnosis in Dentomaxillofacial Radiology: A Systematic Review.

Radiomics is a quantitative tool for digital image analysis. This systematic review aims to investigate the scientific articles to evaluate the potential implications of Radiomics analysis in Dentomaxillofacial Radiology (DMFR). Studies regarding Radiomics applications in DMFR and human samples, in vivo study, a case reports/series if ≧5 samples were included, while case reports/series if < 5 samples, articles other than in English, abstracts without full text, and studies published before 2015 were excluded. Fifty-one articles were selected from 3789 literatures. The QUADAS-2 tool was used for risk of bias assessment. The accuracy of predicting dentomaxillofacial pathologies was considered as the primary outcome, and the modeling type of Radiomics was considered as the secondary outcome. A meta-analysis could not be performed due to the lack of information and standardization among the reported accuracies. The reported accuracies were found between 0.66 and 99.65%. Logistic regression (n = 6) was found to be the most common Radiomics modeling type, followed by Support Vector Machine and Decision Tree (n = 5). Second-order statistics (n = 38) was the most common type of Radiomics application, followed by first-order (n = 26), higher-order (n = 20), and shape-based (n = 15) statistics. Further work is needed to increase standardization in the Radiomics workflow. Quantitative image analysis is an alternative tool for conventional visual radiographic evaluation. Radiomics systems depend on elements such as imaging modality, feature type, data mining, or statistical method. Radiomics applications do not justify digital transformation on their own, but the potential of its integration into the digital workflow is considerable.

Impact of Harmonization on MRI Radiomics Feature Variability Across Preprocessing Methods for Parkinson's Disease Motor Subtype Classification.

This study aimed to assess the reproducibility of MRI-derived radiomic features across multiple preprocessing methods for classifying Parkinson's disease (PD) motor subtypes and to evaluate the impact of ComBat harmonization on feature stability and machine learning performance. T1-weighted MRI scans from 140 PD patients (70 tremor-dominant and 70 postural instability gait difficulty) and 70 healthy controls were obtained from the Parkinson's Progression Markers Initiative (PPMI) database, acquired using different scanner models. Radiomic features were extracted from 16 brain regions using various preprocessing pipelines. ComBat harmonization was applied using a combined batch variable incorporating both scanner models and preprocessing methods. Intraclass correlation coefficients (ICC) and Kruskal-Wallis tests assessed feature reproducibility before and after harmonization. Feature selection was performed using Linear Support Vector Classifier with L1 regularization. Support vector machine classifiers were used for PD subtype classification. ComBat harmonization significantly improved feature reproducibility across all feature groups. The percentage of features showing excellent robustness (ICC ≥ 0.90) increased from 40.2 to 56.3% after harmonization. First-order statistic features showed the highest robustness, with 71.11% demonstrating excellent ICC after harmonization. The proportion of features significantly affected by preprocessing methods was reduced following harmonization. Classification accuracy improved dramatically, from a range of 34-75% before harmonization to 89-96% after harmonization across all preprocessing methods. AUC values similarly increased from 0.28-0.87 to 0.95-0.99 after harmonization. ComBat harmonization significantly enhanced the reproducibility of radiomic features across preprocessing methods and improved PD motor subtype classification performance. This study highlights the importance of harmonization in radiomics research for PD and suggests potential clinical applications in personalized treatment planning.

Fluid Dynamics Effects of a Porous Blockage on Laminar Flow in the Interior Subchannels of a 61-Pin Wire-Wrapped Rod Bundle

Flow blockages in Liquid Metal Fast Reactor (LMFR) fuel assemblies that can cause fuel pin cladding failures require further investigation for the development and optimization of these Generation IV reactor designs. The objective of this study was to experimentally evaluate the effects of a porous blockage accident scenario on laminar flow behavior in the interior subchannels of a prototypical 61-pin wire-wrapped rod bundle. The laminar flow condition is considered because of its importance to natural circulation and loss-of-coolant-accident operating scenarios as well as the limited availability of experimental data. The matched-index-of-refraction method was utilized to obtain two-dimension two-component time-resolved particle image velocimetry (TR-PIV) measurements at three planes centered on blocked and neighboring subchannel regions. Time-averaged first-order and second-order statistics, computed by Reynolds decomposition, were visualized including the mean and fluctuating streamwise and spanwise velocity components. Line profiles described the evolution of flow from upstream regions, to separated flow, and recombination downstream, while a modified Galilean decomposition distinguished differences between flow in the blocked measurement plane and flow in its neighboring counterpart region. Spatial-temporal cross-correlations were performed for the streamwise velocity fluctuations to characterize the convection velocity and decay of traveling vortices in the wake of the porous blockage. The isothermal TR-PIV measurements from this study provide high-fidelity experimental data sets for the validation of computational models and numerical studies to characterize complex fluid phenomena in wire-wrapped rod bundles during the potential accident scenario of a porous, interior flow blockage.

A Systematized Review of Quantitative Ultrasound Based on First-Order Speckle Statistics.

Since the late 1970s, the speckle interference patterns ubiquitous in pulse-echo ultrasound images have been used to characterize subresolution tissue structures. During this time, new models, estimation methods, and processing techniques have proliferated, offering a wealth of recommendations for the task of tissue characterization. A literature review was performed to draw attention to these various methods and to critically track assumptions and gaps in knowledge. A total of 388 articles were collected from a systematic search for first-order speckle statistics in diagnostic ultrasound in the NIH PubMed database and Elsevier's Scopus database. Articles were grouped by basic characteristics and evaluated for addressing fundamental assumptions. A sampling of models and methods is presented to reveal the state of the art in speckle statistics as well as sources of measurement error and other important considerations. While this body of literature emphasizes the value of speckle analysis in diagnostic ultrasound, it is shown that relatively little attention is devoted to basic assumptions such as the linearity of system response and scatterer geometry. Additionally, several areas of investigation are available to improve upon speckle statistics analysis, potentially leading to the advancement of this unique tool.

IMG-23. AN MRI-BASED RADIOMIC SIGNATURE TO PREDICT LONG-TERM SURVIVAL IN PATIENTS WITH DIFFUSE INTRINSIC PONTINE GLIOMAS

Abstract BACKGROUND Diffuse intrinsic pontine gliomas are characterized by a short life expectancy after diagnosis, with a mean overall survival of less than 12 months. To stratify patients and possibly better adapt therapy, we are proposing a radiomic signature that predicts, at the time of diagnosis, which patients might survive beyond 2 years. METHODS A retrospective database comprising 80 patients who had undergone 3D multimodal MRI at the time of diagnosis was compiled. All images were acquired at a single center, and the protocol included T1-weighted images acquired before and after contrast injection, T2-weighted and FLAIR images. Depending on the patient, some modalities were missing (around 10% per modality). After manual tumor delineation, a dedicated radiomic pipeline extracted 79 features (first-order statistics and texture parameters) per imaging modality. In addition, gender, age and 14 tumor shape features were collected without missing data. After feature selection (maximum of 4 features per model), five logistic regression models were defined, one per modality and another for clinical and shape variables. Each patient was eligible for 2 to 5 models, with an average score calculated over these models. RESULTS Based on a leave-one-out approach, the multi-model was able to define patients who survive more than 2 years correctly, (sensitivity=100%, specificity =79%). A prediction could be made for each patient despite missing MR modalities. Defining a radiomic signature based on the multi-model score (cut-off=0.5), Kaplan-Meier analysis shows a good separation of the two groups (p-value&amp;lt;0.001). This result is superior to that obtained with the TP53 mutation (available for 61 patients, with p-value&amp;lt;0.01). CONCLUSIONS We have demonstrated the value of multimodal 3D MRI in predicting which DIPG patients might survive beyond 2 years. The robustness of this radiomic signature has yet to be tested using multicentric images.

Emotion classification using electrocardiogram and machine learning: A study on the effect of windowing techniques

Automated emotion recognition using physiological signals has gained significant attention in recent years due to its potential applications in human–computer interaction, healthcare, and psychology. Electrocardiogram (ECG) signals are widely used due to their non-invasiveness, high temporal resolution, and direct relationship with the autonomic nervous system. In this study, we propose a novel approach for ECG-based emotion recognition using time-series to image encoding techniques and texture-based features combined with machine learning algorithms and deep learning architectures. The ECG data used in this study were obtained from the Continuously Annotated Signals of Emotion (CASE) and Wearable Stress and Affect Detection (WESAD) datasets. We categorized emotional states based on valence and arousal annotations into four classes: High-Valence High-Arousal, High-Valence Low-Arousal, Low-Valence High-Arousal, and Low-Valence Low-Arousal. The ECGs were segmented into 5 and 7-window segments and transformed into 2D representations using Gramian Angular Summation Field, Markov Transition Field, Recurrence Plot (RP), and the triple-channel fusion of all these images. Total of 85 textural features based on the Gray-Level Co-occurrence Matrix (GLCM), Gray-Level Run Length Matrix, Zernike’s Moments, Hu’s Moments, Fractal Dimension Texture Analysis (FDTA), and First-Order Statistics were extracted. Classifiers, namely Random Forest, Support Vector Machine (SVM), eXtreme Gradient Boosting (XGB), 1D Convolutional Neural Network (CNN), and Multi-head Attention Network were considered to classify the emotional states. The performance of the classifiers varied depending on the time-series to image encoding technique, segmentation approaches, and the classifier employed. We achieved the highest Weighted F-measure (F-m) of 94.91% (RP + XGB) and 86.78% (RP + SVM) using the 7-window and 5-window approaches, respectively. Our proposed 1D CNN architecture achieved the highest classification metrics (F-m = 92.52%, Balanced accuracy = 92.0%, Recall = 91.96%, and Precision = 93.16%) with RP images in a 7-window approach. GLCM and FDTA features made significant contributions to the classification of emotional states. Overall, our results suggest that the proposed method holds promise for developing more accurate and efficient emotion recognition systems.

Resolving spatial response heterogeneity in glioblastoma.

Spatial intratumoral heterogeneity poses a significant challenge for accurate response assessment in glioblastoma. Multimodal imaging coupled with advanced image analysis has the potential to unravel this response heterogeneity. Based on automated tumor segmentation and longitudinal registration with follow-up imaging, we categorized contrast-enhancing voxels of 61 patients with suspected recurrence of glioblastoma into either true tumor progression (TP) or pseudoprogression (PsP). To allow the unbiased analysis of semantically related image regions, adjacent voxels with similar values of cerebral blood volume (CBV), FET-PET, and contrast-enhanced T1w were automatically grouped into supervoxels. We then extracted first-order statistics as well as texture features from each supervoxel. With these features, a Random Forest classifier was trained and validated employing a 10-fold cross-validation scheme. For model evaluation, the area under the receiver operating curve, as well as classification performance metrics were calculated. Our image analysis pipeline enabled reliable spatial assessment of tumor response. The predictive model reached an accuracy of 80.0% and a macro-weighted AUC of 0.875, which takes class imbalance into account, in the hold-out samples from cross-validation on supervoxel level. Analysis of feature importances confirmed the significant role of FET-PET-derived features. Accordingly, TP- and PsP-labeled supervoxels differed significantly in their 10th and 90th percentile, as well as the median of tumor-to-background normalized FET-PET. However, CBV- and T1c-related features also relevantly contributed to the model's performance. Disentangling the intratumoral heterogeneity in glioblastoma holds immense promise for advancing precise local response evaluation and thereby also informing more personalized and localized treatment strategies in the future.

Evaluation of wind farm performance over heterogeneously rough terrain using large eddy simulation

We evaluate the effect of an abrupt change in the surface aerodynamic roughness height on a wind farm sited on it using the large eddy simulation (LES). Compared to a wind farm sited on a uniformly rough surface, the alteration in aerodynamic surface roughness from a rough to smooth value leads to substantial changes in the first-order and second-order turbulent statistics. Specifically, the rough-to-smooth surface roughness transition leads to an acceleration of the flow downstream of it, which affects the wake recovery and wind farm power production. Different velocity deficits are formulated considering different definitions of “upstream” velocity. The usual deficit, i.e., the difference between the overall wind farm upstream velocities and downstream of a turbine, attains negative values near the ground, rendering it difficult to model within the usual Gaussian radial-shape framework. An alternative definition, i.e., the difference in velocity at the same location with and without turbines on a heterogeneous surface, consistently yields positive values and is amenable to Gaussian shape-based modelling. The power generation decreases as the step change in surface roughness progressively moves into the wind farm. Maximum power is produced when all turbines are placed downstream of the surface roughness jump and minimum power is generated for a homogeneously rough surface when the entire wind farm is placed on the rough surface.

Two-Stage Training Framework Using Multicontrast MRI Radiomics for IDH Mutation Status Prediction in Glioma.

Purpose To develop a radiomics framework for preoperative MRI-based prediction of isocitrate dehydrogenase (IDH) mutation status, a crucial glioma prognostic indicator. Materials and Methods Radiomics features (shape, first-order statistics, and texture) were extracted from the whole tumor or the combination of nonenhancing, necrosis, and edema regions. Segmentation masks were obtained via the federated tumor segmentation tool or the original data source. Boruta, a wrapper-based feature selection algorithm, identified relevant features. Addressing the imbalance between mutated and wild-type cases, multiple prediction models were trained on balanced data subsets using random forest or XGBoost and assembled to build the final classifier. The framework was evaluated using retrospective MRI scans from three public datasets (The Cancer Imaging Archive [TCIA, 227 patients], the University of California San Francisco Preoperative Diffuse Glioma MRI dataset [UCSF, 495 patients], and the Erasmus Glioma Database [EGD, 456 patients]) and internal datasets collected from the University of Texas Southwestern Medical Center (UTSW, 356 patients), New York University (NYU, 136 patients), and University of Wisconsin-Madison (UWM, 174 patients). TCIA and UTSW served as separate training sets, while the remaining data constituted the test set (1617 or 1488 testing cases, respectively). Results The best performing models trained on the TCIA dataset achieved area under the receiver operating characteristic curve (AUC) values of 0.89 for UTSW, 0.86 for NYU, 0.93 for UWM, 0.94 for UCSF, and 0.88 for EGD test sets. The best performing models trained on the UTSW dataset achieved slightly higher AUCs: 0.92 for TCIA, 0.88 for NYU, 0.96 for UWM, 0.93 for UCSF, and 0.90 for EGD. Conclusion This MRI radiomics-based framework shows promise for accurate preoperative prediction of IDH mutation status in patients with glioma. Keywords: Glioma, Isocitrate Dehydrogenase Mutation, IDH Mutation, Radiomics, MRI Supplemental material is available for this article. Published under a CC BY 4.0 license. See also commentary by Moassefi and Erickson in this issue.

Development of a Benchmark for Thermal Striping in Gas-Cooled Fast Reactors

This paper presents the development of a benchmark for predicting thermal striping through simulation. This work utilized large eddy simulation and will be used to benchmark future models. The testing domain was created using both the STRUCT and the Reynolds-averaged Navier-Stokes turbulence models and is based on an earlier design of the General Atomics Fast Modular Reactor upper plenum. The plenum features two adjacent, identical hexagonal bundles each with a center-placed axial rod drive, with a hot left coolant stream and a cold right coolant stream. The simulation solves the nondimensional Navier-Stokes equations, with temperature accounted as a passive scalar. First- and second-order flow statistics were obtained after 600 convective time units of averaging. The first-order statistics reveal that the hot jet is damped by a recirculatory flow from the near wall. At the same location, the second-order statistics show strong oscillations both in velocity and temperature. The power spectral density was utilized to determine that a low-frequency oscillation occurs here that is within the range of interest for thermal striping. Furthermore, proper orthogonal decomposition was used to identify coherent structures that confirm the oscillatory behavior, indicative of thermal striping. Overall, this benchmark can aid in the development of future models for predicting thermal striping in nuclear reactors, potentially leading to improved reactor safety and performance.

Whole brain morphologic features improve the predictive accuracy of IDH status and VEGF expression levels in gliomas.

Glioma is a systemic disease that can induce micro and macro alternations of whole brain. Isocitrate dehydrogenase and vascular endothelial growth factor are proven prognostic markers and antiangiogenic therapy targets in glioma. The aim of this study was to determine the ability of whole brain morphologic features and radiomics to predict isocitrate dehydrogenase status and vascular endothelial growth factor expression levels. This study recruited 80 glioma patients with isocitrate dehydrogenase wildtype and high vascular endothelial growth factor expression levels, and 102 patients with isocitrate dehydrogenase mutation and low vascular endothelial growth factor expression levels. Virtual brain grafting, combined with Freesurfer, was used to compute morphologic features including cortical thickness, LGI, and subcortical volume in glioma patient. Radiomics features were extracted from multiregional tumor. Pycaret was used to construct the machine learning pipeline. Among the radiomics models, the whole tumor model achieved the best performance (accuracy 0.80, Area Under the Curve 0.86), while, after incorporating whole brain morphologic features, the model had a superior predictive performance (accuracy 0.82, Area Under the Curve 0.88). The features contributed most in predicting model including the right caudate volume, left middle temporal cortical thickness, first-order statistics, shape, and gray-level cooccurrence matrix. Pycaret, based on morphologic features, combined with radiomics, yielded highest accuracy in predicting isocitrate dehydrogenase mutation and vascular endothelial growth factor levels, indicating that morphologic abnormalities induced by glioma were associated with tumor biology.

Radiomics-Based Classification of Tumor and Healthy Liver on Computed Tomography Images.

Liver malignancies, particularly hepatocellular carcinoma and metastasis, stand as prominent contributors to cancer mortality. Much of the data from abdominal computed tomography images remain underused by radiologists. This study explores the application of machine learning in differentiating tumor tissue from healthy liver tissue using radiomics features. Preoperative contrast-enhanced images of 94 patients were used. A total of 1686 features classified as first-order, second-order, higher-order, and shape statistics were extracted from the regions of interest of each patient's imaging data. Then, the variance threshold, the selection of statistically significant variables using the Student's t-test, and lasso regression were used for feature selection. Six classifiers were used to identify tumor and non-tumor liver tissue, including random forest, support vector machines, naive Bayes, adaptive boosting, extreme gradient boosting, and logistic regression. Grid search was used as a hyperparameter tuning technique, and a 10-fold cross-validation procedure was applied. The area under the receiver operating curve (AUROC) assessed the performance. The AUROC scores varied from 0.5929 to 0.9268, with naive Bayes achieving the best score. The radiomics features extracted were classified with a good score, and the radiomics signature enabled a prognostic biomarker for hepatic tumor screening.

Flow analysis through a randomly packed pebble-bed geometry using computational fluid dynamics

This study presents a comprehensive analysis of the flow behavior in packed pebble-bed reactors using computational fluid dynamics (CFD) simulations. The pebble-bed geometry corresponds to an experimental facility located at the Texas A&amp;M Thermal-Hydraulics Research Laboratory. The unsteady Reynolds-averaged Navier–Stokes (URANS) k−ω shear stress transport (SST) and the large eddy simulation (LES) approaches were selected to model the turbulence at different Reynolds numbers. The numerical models were first validated by comparing the pressure drop results obtained from the simulations against established correlations, finding the simulation predictions in accurate agreement. Secondly, the velocity first-order statistics from the URANS k−ω SST and LES calculations were also contrasted with the available experimental particle image velocimetry data to validate the numerical models. Results were found in reasonable agreement as the mean absolute error achieved values smaller than 10% of the inlet velocity for most of the analyzed velocity profiles. A comprehensive turbulence characterization was performed, including second-order statistics, Reynolds stress anisotropy, and turbulent kinetic energy production. The proper orthogonal decomposition of the fluctuating velocity was examined in the current flow domain. The turbulence characterization revealed the complex nature of turbulence in packed pebble-bed geometries, which is further complicated by the presence of an enclosing wall. Overall, the findings of this study provide a solid foundation for the development of more accurate CFD-based methodologies for predicting the behavior of flow through packed pebble-bed reactors.

MicroCT and Histological Analysis of Clot Composition in Acute Ischemic Stroke : AComparative Study of MT-Retrieved Clots and Clot Analogs.

Assessing clot composition on prethrombectomy computed tomography (CT) imaging may help in stroke treatment planning. In this study we seek to use microCT imaging of fabricated blood clots to understand the relationship between CT radiographic signals and the biological makeup. Clots (n = 10) retrieved by mechanical thrombectomy (MT) were collected, and 6clot analogs of varying RBC composition were made. We performed paired microCT and histological image analysis of all 16clots using aScanCo microCT 100 (4.9 µm resolution) and standard H&E staining (imaged at 40×). From these data types, first order statistic (FOS) radiomics were computed from microCT, and percent composition of RBCs (%RBC) was computed from histology. Polynomial and linear regression (LR) were used to build statistical models based on retrieved thrombus microCT and %RBC that were evaluated for their ability to predict the %RBC of clot analogs from mean HU. Correlation analyses of microCT FOS with composition were completed for both retrieved clots and analogs. The LR model fits relating MT-retrieved clot %RBC with mean (R2 = 0.625, p = 0.006) and standard deviation (R2 = 0.564, p < 0.05) in HUs on microCT were significant. Similarly, LR models relating analog histological %RBC to analog protocol %RBC (R2 = 0.915, p = 0.003) and mean HUs on microCT (R2 = 0.872, p = 0.007) were also significant. When the LR model built using MT-retrieved clots was used to predict analog %RBC from mean HUs, significant correlation was observed between predictions and actual histological %RBC (R2 = 0.852, p = 0.009). For retrieved clots, significant correlations were observed for energy and total energy with %RBC and %FP (|R| > 0.7, q < 0.01). Analogs further demonstrated significant correlation between FOS energy, total energy, variance and %WBC (|R| > 0.9, q < 0.01). MicroCT can be used to build models that predict AIS clot composition from routine CT parameters and help us to better understand radiomic signatures associated with clot composition and first pass outcomes. In future work, such observations can be used to better infer clot composition and inform thrombectomy prognostics from pretreatment CTs.

Radiomics-Based Machine Learning Classification Strategy for Characterization of Hepatocellular Carcinoma on Contrast-Enhanced Ultrasound in High-Risk Patients with LI-RADS Category M Nodules.

Objective Accurate differentiation within the LI-RADS category M (LR-M) between hepatocellular carcinoma (HCC) and non-HCC malignancies (mainly intrahepatic cholangiocarcinoma [CCA] and combined hepatocellular and cholangiocarcinoma [cHCC-CCA]) is an area of active investigation. We aimed to use radiomics-based machine learning classification strategy for differentiating HCC from CCA and cHCC-CCA on contrast-enhanced ultrasound (CEUS) images in high-risk patients with LR-M nodules. Methods A total of 159 high-risk patients with LR-M nodules (69 HCC and 90 CCA/cHCC-CCA) who underwent CEUS within 1 month before pathologic confirmation from January 2006 to December 2019 were retrospectively included (111 patients for training set and 48 for test set). The training set was used to build models, while the test set was used to compare models. For each observation, six CEUS images captured at predetermined time points (T1, peak enhancement after contrast injection; T2, 30 seconds; T3, 45 seconds; T4, 60 seconds; T5, 1-2 minutes; and T6, 2-3 minutes) were collected for tumor segmentation and selection of radiomics features, which included seven types of features: first-order statistics, shape (2D), gray-level co-occurrence matrix, gray-level size zone matrix, gray-level run length matrix, neighboring gray tone difference matrix, and gray-level dependence matrix. Clinical data and key radiomics features were employed to develop the clinical model, radiomics signature (RS), and combined RS-clinical (RS-C) model. The RS and RS-C model were built using the machine learning framework. The diagnostic performance of these three models was calculated and compared. Results Alpha-fetoprotein (AFP), CA19-9, enhancement pattern, and time of washout were included as independent factors for clinical model (all p < 0.05). Both the RS and RS-C model performed better than the clinical model in the test set (area under the curve [AUC] of 0.698 [0.571-0.812] for clinical model, 0.903 [0.830-0.970] for RS, and 0.912 [0.838-0.977] for the RS-C model; both p < 0.05). Conclusions Radiomics-based machine learning classifiers may be competent for differentiating HCC from CCA and cHCC-CCA in high-risk patients with LR-M nodules.

On the dynamic residual measure of inaccuracy based on extropy in order statistics

AbstractIn this paper, we introduce a novel way to quantify the remaining inaccuracy of order statistics by utilizing the concept of extropy. We explore various properties and characteristics of this new measure. Additionally, we expand the notion of inaccuracy for ordered random variables to a dynamic version and demonstrate that this dynamic information measure provides a unique determination of the distribution function. Moreover, we investigate specific lifetime distributions by analyzing the residual inaccuracy of the first-order statistics. Nonparametric kernel estimation of the proposed measure is suggested. Simulation results show that the kernel estimator with bandwidth selection using the cross-validation method has the best performance. Finally, an application of the proposed measure on the model selection is provided.

Whole-brain traumatic controlled cortical impact to the left frontal lobe: Magnetic resonance image-based texture analysis.

This research assesses the capability of texture analysis (TA) derived from high-resolution (HR) T2-weighted magnetic resonance imaging to identify primary sequelae following 1-5 hours of controlled cortical impact mild or severe traumatic brain injury (TBI) to the left frontal cortex (focal impact) and secondary (diffuse) sequelae in the right frontal cortex, bilateral corpus callosum, and hippocampus in rats. The TA technique comprised first-order (histogram-based) and second-order statistics (including gray-level co-occurrence matrix, gray-level run length matrix, and neighborhood gray-level difference matrix). Edema in the left frontal impact region developed within 1 hour and continued throughout the 5-hour assessments. The TA features from HR images confirmed the focal injury. There was no significant difference among radiomics features between the left and right corpus callosum or hippocampus from 1 to 5 hours following a mild or severe impact. The adjacent corpus callosum region and the distal hippocampus region (s), showed no diffuse injury 1-5 hours after mild or severe TBI. These results suggest that combining HR images with TA may enhance detection of early primary and secondary sequelae following TBI.

Probing Synergistic High-Order Interaction for Multi-modal Image Fusion.

Multi-modal image fusion aims to generate a fused image by integrating and distinguishing the cross-modality complementary information from multiple source images. While the cross-attention mechanism with global spatial interactions appears promising, it only captures second-order spatial interactions, neglecting higher-order interactions in both spatial and channel dimensions. This limitation hampers the exploitation of synergies between multi-modalities. To bridge this gap, we introduce a Synergistic High-order Interaction Paradigm (SHIP), designed to systematically investigate spatial fine-grained and global statistics collaborations between the multi-modal images across two fundamental dimensions: 1) Spatial dimension: we construct spatial fine-grained interactions through element-wise multiplication, mathematically equivalent to global interactions, and then foster high-order formats by iteratively aggregating and evolving complementary information, enhancing both efficiency and flexibility. 2) Channel dimension: expanding on channel interactions with first-order statistics (mean), we devise high-order channel interactions to facilitate the discernment of inter-dependencies between source images based on global statistics. We further introduce an enhanced version of the SHIP model, called SHIP++ that enhances the cross-modality information interaction representation by the cross-order attention evolving mechanism, cross-order information integration, and residual information memorizing mechanism. Harnessing high-order interactions significantly enhances our model's ability to exploit multi-modal synergies, leading in superior performance over state-of-the-art alternatives, as shown through comprehensive experiments across various benchmarks in two significant multi-modal image fusion tasks: pan-sharpening, and infrared and visible image fusion. The source code is publicly available at https://github.com/manman1995/HOIF.

Machine learning and texture features based approach for classifying Alzheimer’s disease

In order to gain more insight into the nature of the disease and fill in numerous gaps in predictions and treatment, the problem of Alzheimer’s disease (AD) has been extensively studied. A lot of research has already been done to assist identify AD patients utilizing structural and textural features. In this work, texture features based machine learning models are proposed for multi-stage classification of Alzheimer’s disease. Multi-stage classification helps in identifying the disease at early stages and will reduce the suffering of the patients and their family members. The patients in this study are divided into four groups: CN (Cognitive Normal), AD (Alzheimer’s Disease), pMCI (Progressive Mild Cognitive Impairment), and sMCI (Static Mild Cognitive Impairment). The dataset is collected from the ADNI database and has about 15000 MRI raw images in ‘NIFTI’ format. In the first step, the data is preprocessed. Images are converted to the ‘jpg’ format after being stripped of their skulls. 8856 images are selected after preprocessing. For Feature Extraction, FOS (First-Order Statistical) and GLCM (Gray-Level Co-occurrence Matrix) features are utilized. Using normalization, the extracted features are scaled between 0 and 1. The normalized features are then used as input by SVM, Random Forest, Artificial Neural Network, and Decision Tree machine learning methods. Input features used for classification include GLCM, FOS features, and a combination of both. The performance is evaluated using accuracy, recall, F1-score, and precision parameters. The suggested method performs best when both GLCM and FOS features are provided as input. The most accurate method is Random Forest, which has a 66.1% accuracy rate. Decision tree, ANN, and SVM also performed well, with accuracies of 56.4, 58.5, and 59.2 respectively. The proposed approach has the potential for multi-stage classification of Alzheimer’s disease.

Estimating Per-Class Statistics for Label Noise Learning.

Real-world data may contain a considerable amount of noisily labeled examples, which usually mislead the training algorithm and result in degraded classification performance on test data. Therefore, Label Noise Learning (LNL) was proposed, of which one popular research trend focused on estimating the critical statistics (e.g., sample mean and sample covariance), to recover the clean data distribution. However, existing methods may suffer from the unreliable sample selection process or can hardly be applied to multi-class cases. Inspired by the centroid estimation theory, we propose Per-Class Statistic Estimation (PCSE), which establishes the quantitative relationship between the clean (first-order and second-order) statistics and the corresponding noisy statistics for every class. This relationship is further utilized to induce a generative classifier for model inference. Unlike existing methods, our approach does not require sample selection from the instance level. Moreover, our PCSE can serve as a general post-processing strategy applicable to various popular networks pre-trained on the noisy dataset for boosting their classification performance. Theoretically, we prove that the estimated statistics converge to their ground-truth values as the sample size increases, even if the label transition matrix is biased. Empirically, we conducted intensive experiments on various binary and multi-class datasets, and the results demonstrate that PCSE achieves more precise statistic estimation as well as higher classification accuracy when compared with state-of-the-art methods in LNL.