scholarly journals Radiomics and Artificial Intelligence Analysis with Textural Metrics Extracted by Contrast-Enhanced Mammography in the Breast Lesions Classification

Diagnostics ◽  
2021 ◽  
Vol 11 (5) ◽  
pp. 815
Author(s):  
Roberta Fusco ◽  
Adele Piccirillo ◽  
Mario Sansone ◽  
Vincenza Granata ◽  
Maria Rosaria Rubulotta ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Feature Selection ◽  
Breast Biopsy ◽  
Univariate Analysis ◽  
Dual Energy ◽  
Support Vector ◽  
Textural Features ◽  
Surgical Specimen ◽  
Contrast Enhanced ◽  
Malignant Lesions

The aim of the study was to estimate the diagnostic accuracy of textural features extracted by dual-energy contrast-enhanced mammography (CEM) images, by carrying out univariate and multivariate statistical analyses including artificial intelligence approaches. In total, 80 patients with known breast lesion were enrolled in this prospective study according to regulations issued by the local Institutional Review Board. All patients underwent dual-energy CEM examination in both craniocaudally (CC) and double acquisition of mediolateral oblique (MLO) projections (early and late). The reference standard was pathology from a surgical specimen for malignant lesions and pathology from a surgical specimen or fine needle aspiration cytology, core or Tru-Cut needle biopsy, and vacuum assisted breast biopsy for benign lesions. In total, 104 samples of 80 patients were analyzed. Furthermore, 48 textural parameters were extracted by manually segmenting regions of interest. Univariate and multivariate approaches were performed: non-parametric Wilcoxon–Mann–Whitney test; receiver operating characteristic (ROC), linear classifier (LDA), decision tree (DT), k-nearest neighbors (KNN), artificial neural network (NNET), and support vector machine (SVM) were utilized. A balancing approach and feature selection methods were used. The univariate analysis showed low accuracy and area under the curve (AUC) for all considered features. Instead, in the multivariate textural analysis, the best performance considering the CC view (accuracy (ACC) = 0.75; AUC = 0.82) was reached with a DT trained with leave-one-out cross-variation (LOOCV) and balanced data (with adaptive synthetic (ADASYN) function) and a subset of three robust textural features (MAD, VARIANCE, and LRLGE). The best performance (ACC = 0.77; AUC = 0.83) considering the early-MLO view was reached with a NNET trained with LOOCV and balanced data (with ADASYN function) and a subset of ten robust features (MEAN, MAD, RANGE, IQR, VARIANCE, CORRELATION, RLV, COARSNESS, BUSYNESS, and STRENGTH). The best performance (ACC = 0.73; AUC = 0.82) considering the late-MLO view was reached with a NNET trained with LOOCV and balanced data (with ADASYN function) and a subset of eleven robust features (MODE, MEDIAN, RANGE, RLN, LRLGE, RLV, LZLGE, GLV_GLSZM, ZSV, COARSNESS, and BUSYNESS). Multivariate analyses using pattern recognition approaches, considering 144 textural features extracted from all three mammographic projections (CC, early MLO, and late MLO), optimized by adaptive synthetic sampling and feature selection operations obtained the best results (ACC = 0.87; AUC = 0.90) and showed the best performance in the discrimination of benign and malignant lesions.

scholarly journals Radiomic and Artificial Intelligence Analysis with Textural Metrics, Morphological and Dynamic Perfusion Features Extracted by Dynamic Contrast-Enhanced Magnetic Resonance Imaging in the Classification of Breast Lesions

2021 ◽  
Vol 11 (4) ◽  
pp. 1880
Author(s):  
Roberta Fusco ◽  
Adele Piccirillo ◽  
Mario Sansone ◽  
Vincenza Granata ◽  
Paolo Vallone ◽  
...  
Keyword(s):  
Magnetic Resonance Imaging ◽  
Artificial Intelligence ◽  
Feature Selection ◽  
Resonance Imaging ◽  
Dynamic Features ◽  
Surgical Specimen ◽  
Dce Mri ◽  
Dynamic Contrast Enhanced ◽  
Contrast Enhanced ◽  
Malignant Lesions

Purpose: The aim of the study was to estimate the diagnostic accuracy of textural, morphological and dynamic features, extracted by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) images, by carrying out univariate and multivariate statistical analyses including artificial intelligence approaches. Methods: In total, 85 patients with known breast lesion were enrolled in this retrospective study according to regulations issued by the local Institutional Review Board. All patients underwent DCE-MRI examination. The reference standard was pathology from a surgical specimen for malignant lesions and pathology from a surgical specimen or fine needle aspiration cytology, core or Tru-Cut needle biopsy for benign lesions. In total, 91 samples of 85 patients were analyzed. Furthermore, 48 textural metrics, 15 morphological and 81 dynamic parameters were extracted by manually segmenting regions of interest. Statistical analyses including univariate and multivariate approaches were performed: non-parametric Wilcoxon–Mann–Whitney test; receiver operating characteristic (ROC), linear classifier (LDA), decision tree (DT), k-nearest neighbors (KNN), and support vector machine (SVM) were utilized. A balancing approach and feature selection methods were used. Results: The univariate analysis showed low accuracy and area under the curve (AUC) for all considered features. Instead, in the multivariate textural analysis, the best performance (accuracy (ACC) = 0.78; AUC = 0.78) was reached with all 48 metrics and an LDA trained with balanced data. The best performance (ACC = 0.75; AUC = 0.80) using morphological features was reached with an SVM trained with 10-fold cross-variation (CV) and balanced data (with adaptive synthetic (ADASYN) function) and a subset of five robust morphological features (circularity, rectangularity, sphericity, gleaning and surface). The best performance (ACC = 0.82; AUC = 0.83) using dynamic features was reached with a trained SVM and balanced data (with ADASYN function). Conclusion: Multivariate analyses using pattern recognition approaches, including all morphological, textural and dynamic features, optimized by adaptive synthetic sampling and feature selection operations obtained the best results and showed the best performance in the discrimination of benign and malignant lesions.

scholarly journals Hybrid Harmony Search–Artificial Intelligence Models in Credit Scoring

Entropy ◽  
2020 ◽  
Vol 22 (9) ◽  
pp. 989
Author(s):  
Rui Ying Goh ◽  
Lai Soon Lee ◽  
Hsin-Vonn Seow ◽  
Kathiresan Gopal
Keyword(s):  
Artificial Intelligence ◽  
Feature Selection ◽  
Credit Scoring ◽  
Harmony Search ◽  
Model Performance ◽  
Hybrid Models ◽  
Performance Model ◽  
Computational Time ◽  
Support Vector ◽  
Artificial Intelligence Models

Credit scoring is an important tool used by financial institutions to correctly identify defaulters and non-defaulters. Support Vector Machines (SVM) and Random Forest (RF) are the Artificial Intelligence techniques that have been attracting interest due to their flexibility to account for various data patterns. Both are black-box models which are sensitive to hyperparameter settings. Feature selection can be performed on SVM to enable explanation with the reduced features, whereas feature importance computed by RF can be used for model explanation. The benefits of accuracy and interpretation allow for significant improvement in the area of credit risk and credit scoring. This paper proposes the use of Harmony Search (HS), to form a hybrid HS-SVM to perform feature selection and hyperparameter tuning simultaneously, and a hybrid HS-RF to tune the hyperparameters. A Modified HS (MHS) is also proposed with the main objective to achieve comparable results as the standard HS with a shorter computational time. MHS consists of four main modifications in the standard HS: (i) Elitism selection during memory consideration instead of random selection, (ii) dynamic exploration and exploitation operators in place of the original static operators, (iii) a self-adjusted bandwidth operator, and (iv) inclusion of additional termination criteria to reach faster convergence. Along with parallel computing, MHS effectively reduces the computational time of the proposed hybrid models. The proposed hybrid models are compared with standard statistical models across three different datasets commonly used in credit scoring studies. The computational results show that MHS-RF is most robust in terms of model performance, model explainability and computational time.

scholarly journals Preliminary Report on Computed Tomography Radiomics Features as Biomarkers to Immunotherapy Selection in Lung Adenocarcinoma Patients

Cancers ◽  
2021 ◽  
Vol 13 (16) ◽  
pp. 3992
Author(s):  
Vincenza Granata ◽  
Roberta Fusco ◽  
Matilde Costa ◽  
Carmine Picone ◽  
Diletta Cozzi ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Lung Adenocarcinoma ◽  
Center Of Mass ◽  
Univariate Analysis ◽  
Hounsfield Unit ◽  
Progression Free Survival ◽  
Higher Order ◽  
Support Vector ◽  
Control Group ◽  
Textural Features

Purpose: To assess the efficacy of radiomics features obtained by computed tomography (CT) examination as biomarkers in order to select patients with lung adenocarcinoma who would benefit from immunotherapy. Methods: Seventy-four patients (median age 63 years, range 42–86 years) with histologically confirmed lung cancer who underwent immunotherapy as first- or second-line therapy and who had baseline CT studies were enrolled in this approved retrospective study. As a control group, we selected 50 patients (median age 66 years, range 36–86 years) from 2005 to 2013 with histologically confirmed lung adenocarcinoma who underwent chemotherapy alone or in combination with targeted therapy. A total of 573 radiomic metrics were extracted: 14 features based on Hounsfield unit values specific for lung CT images; 66 first-order profile features based on intensity values; 43 second-order profile features based on lesion shape; 393 third-order profile features; and 57 features with higher-order profiles. Univariate and multivariate statistical analysis with pattern recognition approaches and the least absolute shrinkage and selection operator (LASSO) method were used to assess the capability of extracted radiomics features to predict overall survival (OS) and progression free survival (PFS) time. Results: A total of 38 patients (median age 61; range 41–78 years) with confirmed lung adenocarcinoma and subjected to immunotherapy satisfied inclusion criteria, and 50 patients in a control group were included in the analysis The shift in the center of mass of the lesion due to image intensity was significant both to predict OS in patients subjected to immunotherapy and to predict PFS in patients subjected to immunotherapy and in patients in the control group. With univariate analysis, low diagnostic accuracy was reached to stratify patients based on OS and PFS time. Regarding multivariate analysis, considering the robust (two morphological features, three textural features and three higher-order statistical metrics) application of the LASSO approach and all patients, a support vector machine reached the best results for stratifying patients based on OS (area under curve (AUC) of 0.89 and accuracy of 81.6%). Alternatively, considering the robust predictors (six textural features and one higher-order statistical metric) and application of the LASSO approach including all patients, a decision tree reached the best results for stratifying patients based on PFS time (AUC of 0.96 and accuracy of 94.7%). Conclusions: Specific radiomic features could be used to select patients with lung adenocarcinoma who would benefit from immunotherapy because a subset of imaging radiomic features useful to predict OS or PFS time were different between the control group and the immunotherapy group.

scholarly journals Radiomics Analysis on Contrast-Enhanced Spectral Mammography Images for Breast Cancer Diagnosis: A Pilot Study

Entropy ◽  
2019 ◽  
Vol 21 (11) ◽  
pp. 1110 ◽  
Author(s):  
Liliana Losurdo ◽  
Annarita Fanizzi ◽  
Teresa Maria A. Basile ◽  
Roberto Bellotti ◽  
Ubaldo Bottigli ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Pilot Study ◽  
Support Vector ◽  
Svm Classifier ◽  
Textural Features ◽  
Breast Lesions ◽  
Haar Transform ◽  
Statistical Measures ◽  
Contrast Enhanced ◽  
The Impact

Contrast-enhanced spectral mammography is one of the latest diagnostic tool for breast care; therefore, the literature is poor in radiomics image analysis useful to drive the development of automatic diagnostic support systems. In this work, we propose a preliminary exploratory analysis to evaluate the impact of different sets of textural features in the discrimination of benign and malignant breast lesions. The analysis is performed on 55 ROIs extracted from 51 patients referred to Istituto Tumori “Giovanni Paolo II” of Bari (Italy) from the breast cancer screening phase between March 2017 and June 2018. We extracted feature sets by calculating statistical measures on original ROIs, gradiented images, Haar decompositions of the same original ROIs, and on gray-level co-occurrence matrices of the each sub-ROI obtained by Haar transform. First, we evaluated the overall impact of each feature set on the diagnosis through a principal component analysis by training a support vector machine classifier. Then, in order to identify a sub-set for each set of features with higher diagnostic power, we developed a feature importance analysis by means of wrapper and embedded methods. Finally, we trained an SVM classifier on each sub-set of previously selected features to compare their classification performances with respect to those of the overall set. We found a sub-set of significant features extracted from the original ROIs with a diagnostic accuracy greater than 80 % . The features extracted from each sub-ROI decomposed by two levels of Haar transform were predictive only when they were all used without any selection, reaching the best mean accuracy of about 80 % . Moreover, most of the significant features calculated by HAAR decompositions and their GLCMs were extracted from recombined CESM images. Our pilot study suggested that textural features could provide complementary information about the characterization of breast lesions. In particular, we found a sub-set of significant features extracted from the original ROIs, gradiented ROI images, and GLCMs calculated from each sub-ROI previously decomposed by the Haar transform.

ARTIFICIAL INTELLIGENT SYSTEM FOR SKIN DISEASES CLASSIFICATION

2020 ◽  
Vol 32 (05) ◽  
pp. 2050036
Author(s):  
Kamel. K. Mohammed ◽  
Heba M. Afify ◽  
Aboul Ella Hassanien
Keyword(s):  
Feature Selection ◽  
Skin Lesion ◽  
Skin Disease ◽  
Skin Diseases ◽  
Treatment Plan ◽  
Support Vector ◽  
Svm Classifier ◽  
Textural Features ◽  
Artificial Intelligent ◽  
Intelligent Technique

In this paper, an artificial intelligent technique is proposed for skin disease detection and classification. The suggested method comprises four stages, including segmentation, extraction of textural features, and classification. The stretch-based enhanced algorithm has been adapted for image enhancement. Then the method of an active contour is used for segmentation to determine the skin lesion in tissue. Textural features are obtained from the segmented skin lesion. As several numbers of the features can affect the classification precision, ideal feature selection is made to exclude features that are less informative and unnecessary. The feature selection is adjusted with a regularized random forest. Finally, the classification algorithms by support vector machine and a back-propagation neural network (BPNN) are implemented. The dataset consists of 400 dermoscopic images in total divided into 200 benign and 200 malignant skin diseases extracted from the dermoscopic images PH2 database. The result of detecting and classifying the dermoscopic images on these images yielded an accuracy of 99.7%, a sensitivity of 99.4%, and a specificity of 100% by BPNN. The experiential results confirmed that the BPNN classifier is best rather than an SVM classifier for skin disease images. This proposed model will be advanced to support the skin image processing techniques that provided a more accurate diagnosis and rapid treatment plan.

Analysis of Sentiment of Moving a National Capital with Feature Selection Naive Bayes Algorithm and Support Vector Machine

2020 ◽  
Vol 4 (3) ◽  
pp. 504-512
Author(s):  
Faried Zamachsari ◽  
Gabriel Vangeran Saragih ◽  
Susafa'ati ◽  
Windu Gata
Keyword(s):  
Social Media ◽  
Support Vector Machine ◽  
Feature Selection ◽  
Public Opinion ◽  
Naive Bayes ◽  
Naïve Bayes ◽  
Capital City ◽  
Support Vector ◽  
National Capital ◽  
Bayes Algorithm

The decision to move Indonesia's capital city to East Kalimantan received mixed responses on social media. When the poverty rate is still high and the country's finances are difficult to be a factor in disapproval of the relocation of the national capital. Twitter as one of the popular social media, is used by the public to express these opinions. How is the tendency of community responses related to the move of the National Capital and how to do public opinion sentiment analysis related to the move of the National Capital with Feature Selection Naive Bayes Algorithm and Support Vector Machine to get the highest accuracy value is the goal in this study. Sentiment analysis data will take from public opinion using Indonesian from Twitter social media tweets in a crawling manner. Search words used are #IbuKotaBaru and #PindahIbuKota. The stages of the research consisted of collecting data through social media Twitter, polarity, preprocessing consisting of the process of transform case, cleansing, tokenizing, filtering and stemming. The use of feature selection to increase the accuracy value will then enter the ratio that has been determined to be used by data testing and training. The next step is the comparison between the Support Vector Machine and Naive Bayes methods to determine which method is more accurate. In the data period above it was found 24.26% positive sentiment 75.74% negative sentiment related to the move of a new capital city. Accuracy results using Rapid Miner software, the best accuracy value of Naive Bayes with Feature Selection is at a ratio of 9:1 with an accuracy of 88.24% while the best accuracy results Support Vector Machine with Feature Selection is at a ratio of 5:5 with an accuracy of 78.77%.

Identification of Chronic Hypersensitivity Pneumonitis Biomarkers with Machine Learning and Differential Co-expression Analysis

2020 ◽  
Vol 20 ◽  
Author(s):  
Hongwei Zhang ◽  
Steven Wang ◽  
Tao Huang
Keyword(s):  
Feature Selection ◽  
Expression Analysis ◽  
Hypersensitivity Pneumonitis ◽  
Enrichment Analysis ◽  
Functional Enrichment ◽  
Great Promise ◽  
Support Vector ◽  
Svm Classifier ◽  
Clinical Tool ◽  
Chronic Hypersensitivity Pneumonitis

Aims: We would like to identify the biomarkers for chronic hypersensitivity pneumonitis (CHP) and facilitate the precise gene therapy of CHP. Background: Chronic hypersensitivity pneumonitis (CHP) is an interstitial lung disease caused by hypersensitive reactions to inhaled antigens. Clinically, the tasks of differentiating between CHP and other interstitial lungs diseases, especially idiopathic pulmonary fibrosis (IPF), were challenging. Objective: In this study, we analyzed the public available gene expression profile of 82 CHP patients, 103 IPF patients, and 103 control samples to identify the CHP biomarkers. Method: The CHP biomarkers were selected with advanced feature selection methods: Monte Carlo Feature Selection (MCFS) and Incremental Feature Selection (IFS). A Support Vector Machine (SVM) classifier was built. Then, we analyzed these CHP biomarkers through functional enrichment analysis and differential co-expression analysis. Result: There were 674 identified CHP biomarkers. The co-expression network of these biomarkers in CHP included more negative regulations and the network structure of CHP was quite different from the network of IPF and control. Conclusion: The SVM classifier may serve as an important clinical tool to address the challenging task of differentiating between CHP and IPF. Many of the biomarker genes on the differential co-expression network showed great promise in revealing the underlying mechanisms of CHP.

An Improved Intelligent Approach to Enhance the Sentiment Classifier for Knowledge Discovery Using Machine Learning

2020 ◽  
Vol 10 (4) ◽  
pp. 582-593
Author(s):  
Midde Venkateswarlu Naik ◽  
D. Vasumathi ◽  
A.P. Siva Kumar
Keyword(s):  
Support Vector Machine ◽  
Feature Selection ◽  
Global Warming ◽  
Particle Swarm Optimization ◽  
Sentiment Analysis ◽  
Optimization Technique ◽  
Particle Swarm ◽  
Sentiment Classification ◽  
Support Vector ◽  
Swarm Optimization

Aims: The proposed research work is on an evolutionary enhanced method for sentiment or emotion classification on unstructured review text in the big data field. The sentiment analysis plays a vital role for current generation of people for extracting valid decision points about any aspect such as movie ratings, education institute or politics ratings, etc. The proposed hybrid approach combined the optimal feature selection using Particle Swarm Optimization (PSO) and sentiment classification through Support Vector Machine (SVM). The current approach performance is evaluated with statistical measures, such as precision, recall, sensitivity, specificity, and was compared with the existing approaches. The earlier authors have achieved an accuracy of sentiment classifier in the English text up to 94% as of now. In the proposed scheme, an average accuracy of sentiment classifier on distinguishing datasets outperformed as 99% by tuning various parameters of SVM, such as constant c value and kernel gamma value in association with PSO optimization technique. The proposed method utilized three datasets, such as airline sentiment data, weather, and global warming datasets, that are publically available. The current experiment produced results that are trained and tested based on 10- Fold Cross-Validations (FCV) and confusion matrix for predicting sentiment classifier accuracy. Background: The sentiment analysis plays a vital role for current generation people for extracting valid decisions about any aspect such as movie rating, education institute or even politics ratings, etc. Sentiment Analysis (SA) or opinion mining has become fascinated scientifically as a research domain for the present environment. The key area is sentiment classification on semi-structured or unstructured data in distinguish languages, which has become a major research aspect. User-Generated Content [UGC] from distinguishing sources has been hiked significantly with rapid growth in a web environment. The huge user-generated data over social media provides substantial value for discovering hidden knowledge or correlations, patterns, and trends or sentiment extraction about any specific entity. SA is a computational analysis to determine the actual opinion of an entity which is expressed in terms of text. SA is also called as computation of emotional polarity expressed over social media as natural text in miscellaneous languages. Usually, the automatic superlative sentiment classifier model depends on feature selection and classification algorithms. Methods: The proposed work used Support vector machine as classification technique and particle swarm optimization technique as feature selection purpose. In this methodology, we tune various permutations and combination parameters in order to obtain expected desired results with kernel and without kernel technique for sentiment classification on three datasets, including airline, global warming, weather sentiment datasets, that are freely hosted for research practices. Results: In the proposed scheme, The proposed method has outperformed with 99.2% of average accuracy to classify the sentiment on different datasets, among other machine learning techniques. The attained high accuracy in classifying sentiment or opinion about review text proves superior effectiveness over existing sentiment classifiers. The current experiment produced results that are trained and tested based on 10- Fold Cross-Validations (FCV) and confusion matrix for predicting sentiment classifier accuracy. Conclusion: The objective of the research issue sentiment classifier accuracy has been hiked with the help of Kernel-based Support Vector Machine (SVM) based on parameter optimization. The optimal feature selection to classify sentiment or opinion towards review documents has been determined with the help of a particle swarm optimization approach. The proposed method utilized three datasets to simulate the results, such as airline sentiment data, weather sentiment data, and global warming data that are freely available datasets.

scholarly journals Shared and unique metabolic features of the malignant and benign thyroid lesions determined with use of 1H HR MAS NMR spectroscopy

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Agnieszka Skorupa ◽  
Mateusz Ciszek ◽  
Ewa Chmielik ◽  
Łukasz Boguszewicz ◽  
Małgorzata Oczko-Wojciechowska ◽  
...  
Keyword(s):  
Nmr Spectra ◽  
Univariate Analysis ◽  
Mas Nmr ◽  
Metabolic Reprogramming ◽  
Benign Lesions ◽  
Chronic Thyroiditis ◽  
Malignant Lesions ◽  
Metabolic Heterogeneity ◽  
Benign Thyroid ◽  
Thyroid Lesions

AbstractThe purpose of this work was to investigate the distinct and common metabolic features of the malignant and benign thyroid lesions in reference to the non-transformed tissue from the contralateral gland (chronic thyroiditis and colloid goiter). 1H HR MAS NMR spectra of 38 malignant lesions, 32 benign lesions and 112 samples from the non-tumoral tissue (32 from chronic thyroiditis and 80 samples from colloid goiter) were subjected both to multivariate and univariate analysis. The increased succinate, glutamine, glutathione, serine/cysteine, ascorbate, lactate, taurine, threonine, glycine, phosphocholine/glycerophosphocholine and decreased lipids were found in both lesion types in comparison to either colloid goiter or chronic thyroiditis. The elevated glutamate and choline, and reduced citrate and glucose were additionally evident in these lesions in reference to goiter, while the increased myo-inositol—in comparison to thyroiditis. The malignant lesions were characterized by the higher alanine and lysine levels than colloid goiter and thyroiditis, while scyllo-inositol was uniquely increased in the benign lesions (not in cancer) in comparison to both non-tumoral tissue types. Moreover, the benign lesions presented with the unique increase of choline in reference to thyroiditis (not observed in the cancerous tissue). The metabolic heterogeneity of the non-tumoral tissue should be considered in the analysis of metabolic reprogramming in the thyroid lesions.

Sign in / Sign up

Export Citation Format

Share Document