First-order Statistical Features Research Articles

MRI-Based Radiomic Signature for PI3K/AKT-Activated Glioblastoma

Prior work has demonstrated that transcriptional signatures and immune cell infiltrates associated with phosphatidylinositol 3-kinase (PI3K) downregulation are able to distinguish glioblastoma patients who respond to immune checkpoint blockade. PI3K/Akt pathway is involved in tumorigenesis, immune microenvironment regulation, and chemo-radiation treatment resistance. We hypothesized that PI3K/Akt-activated glioblastoma may harbor imaging features that would allow for radiomic discrimination.Patients from The Cancer Genome Atlas and the Ivy Glioblastoma Atlas Project who underwent resection and received temozolomide-based chemoradiation from 1996-2003 were included. All tumors underwent RNA-sequencing and had pre-, post-operative and surveillance MRIs. Single sample Gene Set Enrichment Analysis (ssGSEA) was applied to gene expression data to identify samples with PI3K/Akt pathway activation. A programming environment-based machine learning software was used to extract unique quantitative image features from manually segmented pre-operative tumors from post-contrast T1 sequences. Prior to image analysis, all images underwent z-score normalization over all voxels. Extracted features included shape features, first-order statistical, intensity-histogram based, fractal, local intensity, and texture matrix-based features from both unfiltered and filtered images (wavelet decompositions). A threshold of P < 0.05 was used to identify significant radiomic features. Adaptive LASSO-penalized regression was used to develop a radiomic predictor for PI3K-Akt downregulation. Area under the ROC curve (AUC) assessed performance of the model with three-fold internal cross-validation.Twenty-two patients with glioblastoma were included in the analysis. Tumors from 8 of 22 patients (36.4%) had downregulation of PI3K/Akt signaling by ssGSEA. Among tumors with upregulation of PI3K/Akt signaling, 71% had upstream loss or deletion of PTEN. Tumors with downregulated PI3K/AKT signaling were also more likely to have intratumoral B cell lymphocytes identified by ssGSEA. Overall, 2800 radiomic features were extracted from pre-operative, post-contrast MRI images of 22 patients with glioblastoma. 49 features were significantly different between the two groups. Adaptive LASSO regression analysis was performed to develop a 6-feature radiomic signature with an AUC of 0.955 (95% Confidence Interval, 0.862-1.000).We identified a radiomic signature comprised of 6 features that classifies PI3K/Akt-activated glioblastoma, which is associated with increased immune cell infiltration and potential response to immunotherapy. This signature may help in identifying patients who may benefit from immunomodulation in glioblastoma.

Predicting Response of Lung Cancer with Definitive Radiation Therapy: A Radiomics-Based Prediction Model

We hypothesize that radiomic features provide information for predicting early treatment response of lung cancer with definitive radiation therapy (RT). This study aims to develop and validate a radiomics-based model for predicting early treatment response of lung cancer with definitive radiation therapy (RT).From January 2018 to March 2020, 128 lung cancer patients with definitive radiation therapy (RT) had enrolled retrospectively in the study and randomly divided into training set (n = 90) and test set(n = 38). Based on the Response Evaluation Criteria in Solid Tumors1.1 (RECIST1.1), response of lung cancer with definitive radiation therapy (RT) was dichotomized that treatment was effective if tumor volume decreased over 30༅ at the first follow-up CT after definitive radiation therapy (RT). Radiomic features were extracted using an open-source software development tool from the Gross Tumor Volume (GTV) of planning CT images. Specifically, a total of 107 radiomics features are extracted that include the first-order statistic features, the second-order texture features, and 2D-3D shape-based features. To reduce the feature dimensionality, we used the RFE (Recursive Feature Elimination) method to filter the features and obtained 13 out of 107 features that are most effective for final classification. By analyzing the data and comparing the experimental validation, we employed Random Forest classifier for the response prediction.All patients received definitive radiation therapy (56-70Gy) and no operation. 36 patients received Intensity-modulated radiation therapy (IMRT) and 92 patients received Volumetric Intensity Modulated Arc Therapy (VMAT) (P = 0.951). 103 patients are men and 25 patients are women (P = 0.969). In addition to the five-fold cross-validation, we evaluated the classification results using the independent test set. The AUC for five-fold cross validation is 0.798, and the AUC for the independent test set is 0.771.This study demonstrated that radiomic model shows a good prediction value of early treatment response of lung cancer with definitive radiation therapy (RT). Radiomic features extracted from the Gross Tumor Volume (GTV) of planning CT images may predict response of radiation and might be helpful for individual treatments. The result still needs to be further validated by increasing the amount of data, especially prospective data.

Fuhrman nuclear grade prediction of clear cell renal cell carcinoma: influence of volume of interest delineation strategies on machine learning-based dynamic enhanced CT radiomics analysis.

To investigate the influence of different volume of interest (VOI) delineation strategies on machine learning-based predictive models for discrimination between low and high nuclear grade clear cell renal cell carcinoma (ccRCC) on dynamic contrast-enhanced CT. This study retrospectively collected 177 patients with pathologically proven ccRCC (124 low-grade; 53 high-grade). Tumor VOI was manually segmented, followed by artificially introducing uncertainties as: (i) contour-focused VOI, (ii) margin erosion of 2 or 4mm, and (iii) margin dilation (2, 4, or 6mm) inclusive of perirenal fat, peritumoral renal parenchyma, or both. Radiomics features were extracted from four-phase CT images (unenhanced phase (UP), corticomedullary phase (CMP), nephrographic phase (NP), excretory phase (EP)). Different combinations of four-phasic features for each VOI delineation strategy were used to build 176 classification models. The best VOI delineation strategy and superior CT phase were identified and the top-ranked features were analyzed. Features extracted from UP and EP outperformed features from other single/combined phase(s). Shape features and first-order statistics features exhibited superior discrimination. The best performance (ACC 81%, SEN 67%, SPE 87%, AUC 0.87) was achieved with radiomics features extracted from UP and EP based on contour-focused VOI. Shape and first-order features extracted from UP + EP images are better feature representations. Contour-focused VOI erosion by 2mm or dilation by 4mm within peritumor renal parenchyma exerts limited impact on discriminative performance. It provides a reference for segmentation tolerance in radiomics-based modeling for ccRCC nuclear grading. • Lesion delineation uncertainties are tolerated within a VOI erosion range of 2mm or dilation range of 4mm within peritumor renal parenchyma for radiomics-based ccRCC nuclear grading. • Radiomics features extracted from unenhanced phase and excretory phase are superior to other single/combined phase(s) at differentiating high vs low nuclear grade ccRCC. • Shape features and first-order statistics features showed superior discriminative capability compared to texture features.

Machine Learning Classifiers for Surface Crack Detection in Fracture Experiments

Correctly determining the onset of fracture is crucial when performing mechanical experiments. Commonly carried out by visual inspection, here an image-based machine learning approach is proposed to classify cracked and un-cracked specimens. It yields the potential to objectify and automate crack detection, thereby removing sources of uncertainty and error from the post-processing of experiments. More than 30’000 speckle-pattern images obtained from 77 experiments on three specimen geometries are evaluated. They comprise uniaxial tension, notched tension as well as axisymmetric V-bending experiments. Statistical texture features are extracted from all images. They include both first-order (variance, skewness, kurtosis) and higher-order statistical texture features, i.e. Haralick features. The discriminatory power of the texture information is evaluated based on the Fisher's Discriminant Ratio and feature correlations are identified and quantified. Image texture feature subsets of high discriminatory power are used to parse neural network architectures of different complexities from simple perceptron to feed-forward and cascade neural networks. It is found that a small subset of the investigated texture features is highly significant for all experiments. Using this feature subset in conjunction with multi-layer, non-linear and low complexity feed-forward network architectures classification accuracies in the order of 99% are obtained. At the same time, it is shown that linear classifiers are not sufficient to robustly distinguish the state of the specimens, even when high discriminatory power features are used.Graphical Abstract: [Display omitted]

Bi-parametric magnetic resonance imaging based radiomics for the identification of benign and malignant prostate lesions: cross-vendor validation.

The purpose of this study was to develop Bi-parametric Magnetic Resonance Imaging (BP-MRI) based radiomics models for differentiation between benign and malignant prostate lesions, and to cross-vendor validate the generalization ability of the models. The prebiopsy BP-MRI data (T2-Weighted Image [T2WI] and the Apparent Diffusion Coefficient [ADC]) of 459 patients with clinical suspicion of prostate cancer were acquired using two scanners from different vendors. The prostate biopsies are the reference standard for diagnosing benign and malignant prostate lesions. The training set was 168 patients' data from Siemens (Vendor 1), and the inner test set was 70 patients' data from the same vendor. The external test set was 221 patients' data from GE (Vendor 2). The lesion Region of Interest (ROI) was manually delineated by experienced radiologists. A total of 851 radiomics features including shape, first-order statistical, texture, and wavelet features were extracted from ROI in T2WI and ADC, respectively. Two feature-ranking methods (Minimum Redundancy Maximum Relevance [MRMR] and Wilcoxon Rank-Sum Test [WRST]) and three classifiers (Random Forest [RF], Support Vector Machine [SVM], and the Least Absolute Shrinkage and Selection Operator [LASSO] regression) were investigated for their efficacy in building single-parametric radiomics signatures. A biparametric radiomics model was built by combining the optimal single-parametric radiomics signatures. A comprehensive diagnosis model was built by combining the biparametric radiomics model with age and Prostate Specific Antigen (PSA) value using multivariable logistic regression. All models were built in the training set and independently validated in the inner and external test sets, and the performances of models in the diagnosis of benign and malignant prostate lesions were quantified by the Area Under the Receiver Operating Characteristic Curve (AUC). The mean AUCs of the inner and external test sets were calculated for each model. The non-inferiority test was used to test if the AUC of model in external test was not inferior to the AUC of model in inner test. Combining MRMR and LASSO produced the best-performing single-parametric radiomics signatures with the highest mean AUC of 0.673 for T2WI (inner test AUC = 0.729 vs. external test AUC = 0.616, p = 0.569) and the highest mean AUC of 0.810 for ADC (inner test AUC = 0.822 vs. external test AUC = 0.797, p = 0.102). The biparametric radiomics model produced a mean AUC of 0.833 (inner test AUC = 0.867 vs. external test AUC = 0.798, p = 0.051). The comprehensive diagnosis model had an improved mean AUC of 0.911 (inner test AUC = 0.935 vs. external test AUC = 0.886, p = 0.010). The comprehensive diagnosis model for differentiating benign from malignant prostate lesions was accurate and generalizable.

Impact of feature selection methods and subgroup factors on prognostic analysis with CT-based radiomics in non-small cell lung cancer patients

BackgroundRadiomics is a new technology to noninvasively predict survival prognosis with quantitative features extracted from medical images. Most radiomics-based prognostic studies of non-small-cell lung cancer (NSCLC) patients have used mixed datasets of different subgroups. Therefore, we investigated the radiomics-based survival prediction of NSCLC patients by focusing on subgroups with identical characteristics.MethodsA total of 304 NSCLC (Stages I–IV) patients treated with radiotherapy in our hospital were used. We extracted 107 radiomic features (i.e., 14 shape features, 18 first-order statistical features, and 75 texture features) from the gross tumor volume drawn on the free breathing planning computed tomography image. Three feature selection methods [i.e., test–retest and multiple segmentation (FS1), Pearson's correlation analysis (FS2), and a method that combined FS1 and FS2 (FS3)] were used to clarify how they affect survival prediction performance. Subgroup analysis for each histological subtype and each T stage applied the best selection method for the analysis of All data. We used a least absolute shrinkage and selection operator Cox regression model for all analyses and evaluated prognostic performance using the concordance-index (C-index) and the Kaplan–Meier method. For subgroup analysis, fivefold cross-validation was applied to ensure model reliability.ResultsIn the analysis of All data, the C-index for the test dataset is 0.62 (FS1), 0.63 (FS2), and 0.62 (FS3). The subgroup analysis indicated that the prediction model based on specific histological subtypes and T stages had a higher C-index for the test dataset than that based on All data (All data, 0.64 vs. SCCall, 060; ADCall, 0.69; T1, 0.68; T2, 0.65; T3, 0.66; T4, 0.70). In addition, the prediction models unified for each T stage in histological subtype showed a different trend in the C-index for the test dataset between ADC-related and SCC-related models (ADCT1–ADCT4, 0.72–0.83; SCCT1–SCCT4, 0.58–0.71).ConclusionsOur results showed that feature selection methods moderately affected the survival prediction performance. In addition, prediction models based on specific subgroups may improve the prediction performance. These results may prove useful for determining the optimal radiomics-based predication model.

Classifying Lithofacies from Textural Features in Whole Core CT-Scan Images

Summary X-ray computerized tomography (CT) is a nondestructive method of providing information about the internal composition and structure of whole core reservoir samples. In this study we propose a method to classify lithology. The novelty of this method is that it uses statistical and textural information extracted from whole core CT images in a supervised learning environment. In the proposed approaches, first-order statistical features and textural grey-level co-occurrence matrix (GLCM) features are extracted from whole core CT images. Here, two workflows are considered. In the first workflow, the extracted features are used to train a support vector machine (SVM) to classify lithofacies. In the second workflow, a principal component analysis (PCA) step is added before training with two purposes: first, to eliminate collinearity among the features and second, to investigate the amount of information needed to differentiate the analyzed images. Before extracting the statistical features, the images are preprocessed and decomposed using Haar mother wavelet decomposition schemes to enhance the texture and to acquire a set of detail images that are then used to compute the statistical features. The training data set includes lithological information obtained from core description. The approach is validated using the trained SVM and hybrid (PCA + SVM) classifiers to predict lithofacies in a set of unseen data. The obtained results show that the SVM classifier can predict some of the lithofacies with high accuracy (up to 91% recall), but it misclassifies, to some extent, similar lithofacies with similar grain size, texture, and transport properties. The SVM classifier captures the heterogeneity in the whole core CT images more accurately compared with the core description, indicating that the CT images provide additional high-resolution information not observed by manual core description. Further, the obtained prediction results add information on the similarity of the lithofacies classes. The prediction results using the hybrid classifier are worse than the SVM classifier, indicating that low-power components may contain information that is required to differentiate among various lithofacies.

Attention-Based Second-Order Pooling Network for Hyperspectral Image Classification

Deep learning (DL) has exhibited huge potentials for hyperspectral image (HSI) classification due to its powerful nonlinear modeling and end-to-end optimization characteristics. Although the superior performance of DL-based methods has been witnessed, some limitations can still be found. On the one hand, existing DL frameworks usually resorted to first-order statistical features, whereas they rarely considered second-order or higher order statistical features. On the other hand, the optimization of complex hyperparameters (e.g., the layer number and convolutional kernel size) is time-consuming and a very tough task, making the designed DL framework unexplainable. To overcome these challenges, we propose a novel attention-based second-order pooling network (A-SPN). First, a first-order feature operator is designed to model the spectral–spatial information of HSI. Second, an attention-based second-order pooling (A-SOP) operator is designed to model discriminative and representative features. Finally, a fully connected layer with softmax loss is used for classification. The proposed framework can obtain second-order statistical features in an end-to-end manner. In addition, A-SPN is free of complex hyperparameters tuning, making it more explainable and easily equipped for classification tasks. Experimental results based on three common hyperspectral data sets demonstrate that A-SPN outperforms other traditional and state-of-the-art DL-based HSI classification methods in terms of generalization performance with limited training samples, classification accuracy, convergence rate, and computational complexity.

Feature-based automated segmentation of ablation zones by fuzzy c-mean clustering during low-dose computed tomography.

Intra-procedural monitoring and post-procedural follow-up is necessary for a successful ablation treatment. An imaging technique which can assess the ablation geometry accurately is beneficial to monitor and evaluate treatment. In this study, we developed an automated ablation segmentation technique for serial low-dose, noisy ablation computed tomography (CT) or contrast-enhanced CT (CECT). Low-dose, noisy temporal CT and CECT volumes were acquired during microwave ablation on normal porcine liver (four with non-contrast CT and eight with CECT). Highly constrained backprojection (HYPR) processing was used to recover ablation zone information compromised by low-dose noise. First-order statistic features and normalized fractional Brownian features (NBF) were used to segment ablation zones by fuzzy c-mean clustering. After clustering, the segmented ablation zone was refined by cyclic morphological processing. Automatic and manual segmentations were compared to gross pathology with Dice's coefficient (morphological similarity), while cross-sectional dimensions were compared by percent difference. Automatic and manual segmentations of the ablation zone were very similar to gross pathology (Dice Coefficients: Auto.-Path.=0.84±0.02; Manu.-Path.=0.76±0.03, P=0.11). The differences in ablation area, major diameter and minor diameter were 17.9±3.2%, 11.1±3.2% and 16.2±3.4%, respectively, when comparing automatic segmentation to gross pathology, which were lower than the differences of 32.9±16.8%, 13.0±9.8% and 21.8±5.8% when comparing manual segmentation to gross pathology. Manual segmentations tended to overestimate gross pathology when ablation area was less than 15 cm2 , but the automated segmentation tended to underestimate gross pathology when ablation zone is larger than 20 cm2 . Fuzzy c-means clustering may be used to aid automatic segmentation of ablation zones without prior information or user input, making serial CT/CECT has more potential to assess treatments intra-procedurally.

Mitigation of ocular artifacts for EEG signal using improved earth worm optimization-based neural network and lifting wavelet transform

An Electroencephalogram (EEG) is often tarnished by various categories of artifacts. Numerous efforts have been taken to improve its quality by eliminating the artifacts. The EEG involves the biological artifacts (ocular artifacts, ECG and EMG artifacts), and technical artifacts (noise from the electric power source, amplitude artifacts, etc.). From these physiological artifacts, ocular activities are one of the most well-known over other noise sources. Reducing the risks of this event and avoid it is practically very difficult, even impossible, as the ocular activities are involuntary tasks. To trim down the effect of ocular artifacts overlapping with EEG signal and overwhelm the subjected flaws, few intelligent approaches have to be developed. This proposal tempts to implement a novel method for detecting and preventing ocular artifacts from the EEG signal. The developed model involves two main phases: (a) Detection of Ocular artifacts and (b) Removal of ocular artifacts. For detecting the ocular artifacts, initially, the EEG is subjected to decomposition process using 5-level Discrete Wavelet Transform (DWT), and Empirical Mean Curve Decomposition (EMCD). Next to the decomposition process, the features like kurtosis, variance, Shannon’s entropy, and few first-order statistical features are extracted. These features will be helpful for the detection process in the classification side. For detecting the ocular artifacts from the decomposed signal, the extracted features are subjected to a machine learning algorithm called Neural Network (NN). As an improvement to the conventional NN, the training algorithm of ANN is improved by the improved Earth Worm optimization Algorithm (EWA) termed as Dual Positioned Elitism-based EWA (DPE-EWA), which updates the weight of NN to improve the performance. In the Removal phase, the optimized Lifting Wavelet Transform (LWT) is deployed, in which the improvement is made on optimizing the filter coefficients using the proposed DPE-EWA. Thus, the integration of optimized NN and optimized LWT suggests a potential possibility to accommodate the detection and removal of ocular artifacts that exist in the EEG signals.

ALGeNet: Adaptive Log-Euclidean Gaussian embedding network for time series forecasting

Time series prediction has attracted much attention as various issues can be formulated as such a task. It is one of the critical challenges to extract the intrinsic information for estimating future trends from historical data. Long Short-Term Memory Network (LSTM) shows excellent performance in this assignment. Probabilistic information extraction, which is demonstrated effective in object recognition in recent years, has not been introduced in time series prediction. To our knowledge, there has not been any work on plugging trainable probability distributions into LSTM as feature representation in an end-to-end manner. In this work we put forward an Adaptive Log-Euclidean Gaussian embedding Network (ALGeNet) to take one step further on solving this problem. The core of the network is capturing statistical information through the Gaussian Distribution with LSTM for end-to-end learning. As the space of Gaussian Distribution is a manifold, we try to embed Gaussian layer into LSTM through mapping Gaussian space into a linear space based on the Lie group and logarithm operation. We introduce four descriptors of Gaussian, two descriptors performing direct logarithm and the others performing indirect logarithm. All of them can extract first-order and second-order statistical features while utilizing the structures of geometry and smooth group of Gaussians. Furthermore, our adaptive mechanism merges the advantages of each descriptor and works well. Experimental results on a real-world wind speed dataset and a system-level electricity load dataset show that the proposed adaptive network outperforms some state-of-the-art algorithms.

Exploring technical issues in personalized medicine: NSCLC survival prediction by quantitative image analysis-usefulness of density correction of volumetric CT data.

The aim of this study was to apply density correction method to the quantitative image analysis of non-small cell lung cancer (NSCLC) computed tomography (CT) images, determining its influence on overall survival (OS) prediction of surgically treated patients. Clinicopathological (CP) data and preoperative CT scans, pre- and post-contrast medium (CM) administration, of 57 surgically treated NSCLC patients, were retrospectively collected. After CT volumetric density measurement of primary gross tumour volume (GTV), aorta and tracheal air, density correction was conducted on GTV (reference values: aortic blood and tracheal air). For each resulting data set (combining CM administration and normalization), first-order statistical and textural features were extracted. CP and imaging data were correlated with patients 1-, 3- and 5-year OS, alone and combined (uni-/multivariate logistic regression and Akaike information criterion). Predictive performance was evaluated using the ROC curves and AUC values and compared among non-normalized/normalized data sets (DeLong test). The best predictive values were obtained when combining CP and imaging parameters (AUC values: 1year 0.72; 3years 0.82; 5years 0.78). After normalization resulted an improvement in predicting 1-year OS for some of the grey level size zonebased features (large zone low grey level emphasis) and for the combined CP-imaging model, a worse performance for grey level co-occurrence matrix (cluster prominence and shade) and first-order statistical (range) parameters for 1- and 5-year OS, respectively. The negative performance of cluster prominence in predicting 1-year OS was the only statistically significant result (p value 0.05). Density corrections of volumetric CT data showed an opposite influence on the performance of imaging quantitative features in predicting OS of surgically treated NSCLC patients, even if no statistically significant for almost all predictors.

Prediction of transient tumor enlargement using MRI tumor texture after radiosurgery on vestibular schwannoma.

PurposeVestibular schwannomas (VSs) are uncommon benign brain tumors, generally treated using Gamma Knife radiosurgery (GKRS). However, due to the possible adverse effect of transient tumor enlargement (TTE), large VS tumors are often surgically removed instead of treated radiosurgically. Since microsurgery is highly invasive and results in a significant increased risk of complications, GKRS is generally preferred. Therefore, prediction of TTE for large VS tumors can improve overall VS treatment and enable physicians to select the most optimal treatment strategy on an individual basis. Currently, there are no clinical factors known to be predictive for TTE. In this research, we aim at predicting TTE following GKRS using texture features extracted from MRI scans.MethodsWe analyzed clinical data of patients with VSs treated at our Gamma Knife center. The data was collected prospectively and included patient‐ and treatment‐related characteristics and MRI scans obtained at day of treatment and at follow‐up visits, 6, 12, 24 and 36 months after treatment. The correlations of the patient‐ and treatment‐related characteristics to TTE were investigated using statistical tests. From the treatment scans, we extracted the following MRI image features: first‐order statistics, Minkowski functionals (MFs), and three‐dimensional gray‐level co‐occurrence matrices (GLCMs). These features were applied in a machine learning environment for classification of TTE, using support vector machines.ResultsIn a clinical data set, containing 61 patients presenting obvious non‐TTE and 38 patients presenting obvious TTE, we determined that patient‐ and treatment‐related characteristics do not show any correlation to TTE. Furthermore, first‐order statistical MRI features and MFs did not significantly show prognostic values using support vector machine classification. However, utilizing a set of 4 GLCM features, we achieved a sensitivity of 0.82 and a specificity of 0.69, showing their prognostic value of TTE. Moreover, these results increased for larger tumor volumes obtaining a sensitivity of 0.77 and a specificity of 0.89 for tumors larger than 6 cm3.ConclusionsThe results found in this research clearly show that MRI tumor texture provides information that can be employed for predicting TTE. This can form a basis for individual VS treatment selection, further improving overall treatment results. Particularly in patients with large VSs, where the phenomenon of TTE is most relevant and our predictive model performs best, these findings can be implemented in a clinical workflow such that for each patient, the most optimal treatment strategy can be determined.

Radiomics-Based Prediction of Long-Term Treatment Response of Vestibular Schwannomas Following Stereotactic Radiosurgery.

Stereotactic radiosurgery (SRS) is one of the treatment modalities for vestibular schwannomas (VSs). However, tumor progression can still occur after treatment. Currently, it remains unknown how to predict long-term SRS treatment outcome. This study investigates possible magnetic resonance imaging (MRI)-based predictors of long-term tumor control following SRS. Retrospective cohort study. Tertiary referral center. Analysis was performed on a database containing 735 patients with unilateral VS, treated with SRS between June 2002 and December 2014. Using strict volumetric criteria for long-term tumor control and tumor progression, a total of 85 patients were included for tumor texture analysis. All patients underwent SRS and had at least 2 years of follow-up. Quantitative tumor texture features were extracted from conventional MRI scans. These features were supplied to a machine learning stage to train prediction models. Prediction accuracy, sensitivity, specificity, and area under the receiver operating curve (AUC) are evaluated. Gray-level co-occurrence matrices, which capture statistics from specific MRI tumor texture features, obtained the best prediction scores: 0.77 accuracy, 0.71 sensitivity, 0.83 specificity, and 0.93 AUC. These prediction scores further improved to 0.83, 0.83, 0.82, and 0.99, respectively, for tumors larger than 5 cm. Results of this study show the feasibility of predicting the long-term SRS treatment response of VS tumors on an individual basis, using MRI-based tumor texture features. These results can be exploited for further research into creating a clinical decision support system, facilitating physicians, and patients to select a personalized optimal treatment strategy.

Analysis on Results Comparison of Feature Extraction Methods for Breast Cancer Classification

Feature extraction plays a vital role in image processing techniques for medical imaging. In this paper, the researchers proposeda breast cancer classification system by using image processing techniques to help the radiologists that the system can improvethe mammogram screening process and increase the life of cancer patients. Our breast cancer classification system based on thecombination of first-order Statistics features and second-order Gray Level Co-occurrence Matrix (GLCM) features and SupportVector Machine is used as a classifier. This system is composed of five stages. At first, preprocessing is carried out for removingnoise and detail artifacts from an image, reducing the size of the image by cropping, enhancing the image to show clearly theappearance of the image. Median Filters are used to remove the artifact, noise, high-frequency components and unwanted partsin the background of the mammogram images. Secondly, Otsu segmentation is used to extract the breast region from thebackground image. In a third stage, enhancement are applied on segmented result images to get efficient features for a higher classification accuracy rate. First-order statistics and second-order texture GLCM features are extracted form enhanced image.Support Vector Machine is used as a classifier for the classification of abnormal and normal images. Finally, performancecomparison of the first order, second order features and combination of first-order statistics and second-order GLCM features forbreast cancer detection system are done with classification accuracy scores. In this system, an input MIAS database are used forour breast cancer detection system.

No-reference quality index of depth images based on statistics of edge profiles for view synthesis

Virtual view synthesis has been increasingly popular due to the wide applications of multi-view and free-viewpoint videos. In view synthesis, texture images are rendered to generate the new viewpoint with the guidance of the depth images. The quality of depth images is vital for generating high-quality synthesized views. While the impact of texture image and the rendering process on the quality of the synthesized view has been extensively studied, the quality evaluation of depth images remains largely unexplored. With this motivation, this paper presents a no-reference image quality index for depth maps by modeling the statistics of edge profiles (SEP) in a multi-scale framework. The Canny operator is first utilized to locate the edges in depth images. Then the edge profiles are constructed, based on which the first-order and second-order statistical features are extracted for portraying the distortions in depth images. Finally, the random forest is employed for building the quality assessment model for depth maps. Experiments are conducted on two annotated view synthesis image/video quality databases. The experimental results and comparisons demonstrate that the proposed metric outperforms the relevant state-of-the-art quality metrics by a large margin. Furthermore, it has better generalization ability.

Preoperative Prediction of Axillary Lymph Node Metastasis in Breast Carcinoma Using Radiomics Features Based on the Fat-Suppressed T2 Sequence

To investigate the value of radiomics method based on the fat-suppressed T2 sequence for preoperative predicting axillary lymph node (ALN) metastasis in breast carcinoma. The data of 329 invasive breast cancer patients were divided into the primary cohort (n = 269) and validation cohort (n = 60). Radiomics features were extracted from the fat-suppressed T2-weighted images on breast MRI, and ALN metastasis-related radiomics feature selection was performed using Mann-Whitney U-test and support vector machines with recursive feature elimination; then a radiomics signature was constructed by linear support vector machine. The predictive models were constructed using a linear regression model based on the clinicopathologic factors and radiomics signature, and nomogram was used for a visual prediction of the combined model. The predictive performances are evaluated with the sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve. A total of 647 radiomics features were extracted from each patient. About 23 ALN metastasis-related radiomics features were selected to construct the radiomics signature, including 17 texture features, 5 first-order statistical features, and one shape feature; patient age, tumor size, HER2 status, and vascular cancer thrombus accompanied or not were selected to construct the cilinicopathologic feature model. The sensitivity, specificity, accuracy, and are under the curve value of radiomics signature, clinicopathologic feature model, and the nomogram were 65.22%, 81.08%, 75.00%, and 0.819 (95% confidence interval [CI]: 0.776-0.861), 30.44%, 81.08%, 61.67%, and 0.605 (95% CI: 0.571-0.624) and 60.87%, 89.19%, 78.33%, and 0.810 (95% CI: 0.761-0.855), respectively. Radiomics methods based on the fat-suppressed T2 sequence and the nomogram are helpful for preoperative accurate predicting ALN metastasis.

Preliminary value of CT radiomics in predicting anaplastic lymphoma kinase fusion gene expression in lung adenocarcinoma

Objective To explore the value of quantitative CT radiomics features in predicting the anaplastic lymphoma kinase (ALK) mutation status in lung adenocarcinoma patients. Methods This retrospective study reviewed one hundred and ninety-five lung adenocarcinoma patients (including 60 patients with ALK mutation) whose ALK genetic test results were available from Nov 2015 to May 2018 in PUMCH. VOIs were labeled by an automatic pulmonary nodule detection and segmentation algorithm and were later revised and confirmed by two senior radiologists. The PyRadiomics tools were used to resample the labeled regions, followed by image pre-processing (Wavelet filter or Laplacian of Gaussian (LoG) filter) and feature extraction. Normalized features were selected based on their representativeness on Dr. Wise research platform. Multivariate logistic regression was performed to develop prediction models of ALK mutation gene based on different image pre-processing techniques and different radiomics feature types. The results were validated by ten runs of five-fold cross validation. ROC curve analysis and Delong test were used to compare the predictive performance among models. Results Fifteen radiomics features with the highest representativeness were selected from the original 1 232 features. The prediction model based on these radiomics features showed good performance (AUC=0.88 in the training set and 0.78 in the validation set) and was not significantly different from the prediction models based on radiomics features of different pre-processing images (AUC=0.76, P=0.1, original CT images; AUC=0.75, P=0.3, Wavelet-filtered images; AUC=0.76, P=0.2, LoG-filtered images). Among the models built with radiomics features of different types, the one based on GLCM feature (a subtype of texture feature) showed the best performance in predicting ALK genetic status (AUC=0.83, accuracy=0.74, sensitivity=0.85 and specificity=0.69). The model based on first-order statistic features had an AUC of 0.80. Conclusion Quantitative CT radiomics features have a good potential to anticipate the expression of ALK fused gene in patients with lung adenocarcinoma. Key words: Lung neoplasm; Radiomics; Tomography, X-ray computed; Anaplastic lymphoma kinase

1932P - Evaluation of CT-based radiomics in patients with renal cell carcinoma

BackgroundAlthough the personalized medicine has always focused on the genomic or proteomic characterization of tumor, medical imaging is still one of the major factors to guide therapy and to monitor the progression of the tumor. Radiomics is an emerging field that converts the medical image data into the mineable quantitative features via the automatically algorithms, and can server as a bridge between medical image, genomics and clinical-parameters. Serval studies have demonstrated that the radiomic-based model can predict outcome of RCC, but the correlation between radiomic features and histological subtypes of RCC is still unknown. The aim of this letter is to focus on the ability of radiomics to identify the histological subtypes and metastasis of RCC. MethodsThis study included Forty-four patients diagnosed with renal tumor. For each renal lesion, The CT images of volume of interest (VOI) were obtained semi-automatically by two experienced nuclear medicine physician, 85 texture features were extracted from each VOI using the first-order statistics features, Shape Based Features, Gray Level Neighboring Gray Level Dependence Matrix and Neighboring Gray Tone Difference Matrix. ResultsTo investigate the value of radiomic features to capture phenotypic differences of RCC, we performed Unsupervised Clustering of patients with similar radiomic expression patterns. We analyzed the two main clusters of patients with clinical parameters, and found that the tumor clusters were statistically and significantly associated with primary tumor stage (P<0.001), M-stage (P=0.049) and benign (P=0.037), wherein high T-stages, M-stage and tumor group showed in cluster II. RCC histology and N-stage (lymph-node) did not reach statistical significance for their association with the radiomic expression patterns (P=0.165, 0.361, respectively). In addition, we analyzed the overall survival (OS) of the each radiomic features, and showed that P25, IMC1 and IMC2 were associated with OS (P=0.002, 0.002, 0.016, respectively, log-rank test). ConclusionsThe radiomic features from medical images could be helpful in deciphering T-stages, metastasis and benign of RCC and may have potential as imaging biomarker for prediction of RCC overall survival. Legal entity responsible for the studyZhejiang Cancer Hospital. FundingNational Natural Science Foundation of China (No 81402117, 81671775), Natural Science Foundation of Zhejiang Province (No LY17H160043) and Qianjiang talent plan of Zhejiang Province (No QJD1602025). DisclosureAll authors have declared no conflicts of interest.

Radiomics of small renal masses on multiphasic CT: accuracy of machine learning–based classification models for the differentiation of renal cell carcinoma and angiomyolipoma without visible fat

To investigate the discriminative capabilities of different machine learning-based classification models on the differentiation of small (< 4cm) renal angiomyolipoma without visible fat (AMLwvf) and renal cell carcinoma (RCC). This study retrospectively collected 163 patients with pathologically proven small renal mass, including 118 RCC and 45 AMLwvf patients. Target region of interest (ROI) delineation, followed by texture feature extraction, was performed on a representative slice with the largest lesion area on each phase of the four-phase CT images. Fifteen concatenations of the four-phasic features were fed into 224 classification models (built with 8 classifiers and 28 feature selection methods), classification performances of the 3360 resultant discriminative models were compared, and the top-ranked features were analyzed. Image features extracted from the unenhanced phase (UP) CT image demonstrated dominant classification performances over features from other three phases. The two discriminative models "SVM + t_score" and "SVM + relief" achieved the highest classification AUC of 0.90. The 10 top-ranked features from UP included 1 shape feature, 5 first-order statistics features, and 4 texture features, where the shape feature and the first-order statistics features showed superior discriminative capabilities in differentiating RCC vs. AMLwvf through the t-SNE visualization. Image features extracted from UP are sufficient to generate accurate differentiation between AMLwvf and RCC using machine learning-based classification model. • Radiomics extracted from unenhanced CT are sufficient to accurately differentiate angiomyolipoma without visible fat and renal cell carcinoma using machine learning-based classification model. • The highest discriminative models achieved an AUC of 0.90 and were based on the analysis of unenhanced CT, alone or in association with images obtained at the nephrographic phase. • Features related to shape and to histogram analysis (first-order statistics) showed superior discrimination compared with gray-level distribution of the image (second-order statistics, commonly called texture features).