A comparative study of computer-aided classification systems for focal hepatic lesions from B-mode ultrasound

Abstract

A comparative study of three computer-aided classification (CAC) systems for characterization of focal hepatic lesions (FHLs), such as cyst, hemangioma (HEM), hepatocellular carcinoma (HCC) and metastatic carcinoma (MET), along with normal (NOR) liver tissue is carried out in the present work. In order to develop efficient CAC systems a comprehensive and representative dataset consisting of B-mode ultrasound images with (1) typical and atypical cases of cyst, HEM and MET lesions, (2) small and large HCC lesions and (3) NOR liver cases have been used for designing K-nearest neighbour (KNN), probabilistic neural network (PNN) and a back propagation neural network (BPNN) classifiers. For differential diagnosis between atypical FHLs, expert radiologists often visualize the textural characteristics of regions inside and outside the lesion. Accordingly in the present work, texture features and texture ratio features are computed from regions inside and outside the lesions. A feature set consisting of 208 texture features (i.e. 104 texture features and 104 texture ratio features) is subjected to principal component analysis (PCA) for dimensionality reduction; it is observed that maximum accuracy of 87.7% is obtained for a PCA-BPNN-based CAC system in comparison to 86.1% and 85% as obtained by PCA-PNN and PCA-KNN-based CAC systems. The sensitivity of the proposed PCA-BPNN based CAC system for NOR, Cyst, HEM, HCC and MET cases is 82.5%, 96%, 93.3%, 90% and 82.2%, respectively. The sensitivity values with respect to typical, atypical, small HCC and large HCC cases are 85.9%, 88.1%, 100% and 87%, respectively. Keeping in view the comprehensive and representative dataset used for designing the classifier, the results obtained by the proposed PCA-BPNN-based CAC system are quite encouraging and indicate its usefulness to assist experienced radiologists for interpretation and diagnosis of FHLs.     

Evaluation of Texture for Classification of Abdominal Aortic Aneurysm After Endovascular Repair

The use of the endovascular prostheses in abdominal aortic aneurysm has proven to be an effective technique to reduce the pressure and rupture risk of aneurysm. Nevertheless, in a long-term perspective, complications such as leaks inside the aneurysm sac (endoleaks) could appear causing a pressure elevation and increasing the danger of rupture consequently. At present, computed tomographic angiography (CTA) is the most common examination for medical surveillance. However, endoleak complications cannot always be detected by visual inspection on CTA scans. The investigation on new techniques to detect endoleaks and analyse their effects on treatment evolution is of great importance for endovascular aneurysm repair (EVAR) technique. The purpose of this work was to evaluate the capability of texture features obtained from the aneurysmatic thrombus CT images to discriminate different types of evolutions caused by endoleaks. The regions of interest (ROIs) from patients with different post-EVAR evolution were extracted by experienced radiologists. Three techniques were applied to each ROI to obtain texture parameters, namely the grey level co-occurrence matrix (GLCM), the grey level run length matrix (GLRLM) and the grey level difference method (GLDM). The results showed that GLCM, GLRLM and GLDM features presented a good discrimination ability to differentiate between favourable or unfavourable evolutions. GLCM was the most efficient in terms of classification accuracy (93.41% ± 0.024) followed by GLRLM (90.17% ± 0.077) and finally by GLDM (81.98% ± 0.045). According to the results, we can consider texture analysis as complementary information to classified abdominal aneurysm evolution after EVAR.     

Wavelet-packet-based texture analysis for differentiation between benign and malignant liver tumours in ultrasound images

The purpose of this study was to apply a novel method of multiscale echo texture analysis for distinguishing benign (hemangiomas) from malignant (hepatocellular carcinomas (HCCs) and metastases) focal liver lesions in B-mode ultrasound images. In this method, regions of interest (ROIs) extracted from within the lesions were decomposed into subimages by wavelet packets. Multiscale texture features that quantify homogeneity of the echogenicity were calculated from these subimages and were combined by an artificial neural network (ANN). A subset of the multiscale features was selected that yielded the highest performance in the classification of lesions measured by the area under the receiver operating characteristic curve (Az). In an analysis of 193 ROIs consisting of 50 hemangiomas, 87 hepatocellular carcinomas and 56 metastases, the multiscale features yielded a high A: value of 0.92 in distinguishing benign from malignant lesions, 0.93 in distinguishing hemangiomas from HCCs and 0.94 in distinguishing hemangiomas from metastases. Our new multiscale texture analysis method can effectively differentiate malignant from benign lesions, and thus has the potential to increase the accuracy of diagnosis of focal liver lesions in ultrasound images.     

Differential diagnosis of CT focal liver lesions using texture features, feature selection and ensemble driven classifiers

OBJECTIVES: The aim of the present study is to define an optimally performing computer-aided diagnosis (CAD) architecture for the classification of liver tissue from non-enhanced computed tomography (CT) images into normal liver (C1), hepatic cyst (C2), hemangioma (C3), and hepatocellular carcinoma (C4). To this end, various CAD architectures, based on texture features and ensembles of classifiers (ECs), are comparatively assessed. MATERIALS AND METHODS: Number of regions of interests (ROIs) corresponding to C1-C4 have been defined by experienced radiologists in non-enhanced liver CT images. For each ROI, five distinct sets of texture features were extracted using first order statistics, spatial gray level dependence matrix, gray level difference method, Laws' texture energy measures, and fractal dimension measurements. Two different ECs were constructed and compared. The first one consists of five multilayer perceptron neural networks (NNs), each using as input one of the computed texture feature sets or its reduced version after genetic algorithm-based feature selection. The second EC comprised five different primary classifiers, namely one multilayer perceptron NN, one probabilistic NN, and three k-nearest neighbor classifiers, each fed with the combination of the five texture feature sets or their reduced versions. The final decision of each EC was extracted by using appropriate voting schemes, while bootstrap re-sampling was utilized in order to estimate the generalization ability of the CAD architectures based on the available relatively small-sized data set. RESULTS: The best mean classification accuracy (84.96%) is achieved by the second EC using a fused feature set, and the weighted voting scheme. The fused feature set was obtained after appropriate feature selection applied to specific subsets of the original feature set. CONCLUSIONS: The comparative assessment of the various CAD architectures shows that combining three types of classifiers with a voting scheme, fed with identical feature sets obtained after appropriate feature selection and fusion, may result in an accurate system able to assist differential diagnosis of focal liver lesions from non-enhanced CT images.     

Feature Selection in Computer-Aided Breast Cancer Diagnosis via Dynamic Contrast-Enhanced Magnetic Resonance Images

The accuracy of computer-aided diagnosis (CAD) for early detection and classification of breast cancer in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is dependent upon the features used by the CAD classifier. Here, we show that fast orthogonal search (FOS), which provides a more efficient iterative manner of computing stepwise regression feature selection, can select features with predictive value from a set of kinetic and texture candidate features computed from dynamic contrast-enhanced magnetic resonance images. FOS can in minutes search candidate feature sets of millions of terms, which may include cross-products of features up to second-, third- or fourth-order. This method is tested on a set of 83 DCE-MRI images, of which 20 are for cancerous and 63 for benign cases, using a leave-one-out trial. The features selected by FOS were used in a FOS predictor and nearest-neighbour predictor and had an area under the receiver operating curve (AUC) of 0.889 and 0.791 respectively. The FOS predictor AUC is significantly improved over the signal enhancement ratio predictor with an AUC of 0.706 (p = 0.0035 for the difference in the AUCs). Moreover, using FOS-selected features in a support vector machine increased the AUC over that resulting when the features were manually selected.     

Support Vector Machine Model for Diagnosing Pneumoconiosis Based on Wavelet Texture Features of Digital Chest Radiographs

This study aims to explore the classification ability of decision trees (DTs) and support vector machines (SVMs) to discriminate between the digital chest radiographs (DRs) of pneumoconiosis patients and control subjects. Twenty-eight wavelet-based energy texture features were calculated at the lung fields on DRs of 85 healthy controls and 40 patients with stage I and stage II pneumoconiosis. DTs with algorithm C5.0 and SVMs with four different kernels were trained by samples with two combinations of the texture features to classify a DR as of a healthy subject or of a patient with pneumoconiosis. All of the models were developed with fivefold cross-validation, and the final performances of each model were compared by the area under receiver operating characteristic (ROC) curve. For both SVM (with a radial basis function kernel) and DT (with algorithm C5.0), areas under ROC curves (AUCs) were 0.94 ± 0.02 and 0.86 ± 0.04 (P = 0.02) when using the full feature set and 0.95 ± 0.02 and 0.88 ± 0.04 (P = 0.05) when using the selected feature set, respectively. When built on the selected texture features, the SVM with a polynomial kernel showed a higher diagnostic performance with an AUC value of 0.97 ± 0.02 than SVMs with a linear kernel, a radial basis function kernel and a sigmoid kernel with AUC values of 0.96 ± 0.02 (P = 0.37), 0.95 ± 0.02 (P = 0.24), and 0.90 ± 0.03 (P = 0.01), respectively. The SVM model with a polynomial kernel built on the selected feature set showed the highest diagnostic performance among all tested models when using either all the wavelet texture features or the selected ones. The model has a good potential in diagnosing pneumoconiosis based on digital chest radiographs.     

Computer-aided diagnosis in mammography: classification of mass and normal tissue by texture analysis

Computer-aided diagnosis schemes are being developed to assist radiologists in mammographic interpretation. In this study, we investigated whether texture features could be used to distinguish between mass and non-mass regions in clinical mammograms. Forty-five regions of interest (ROIs) containing true masses with various degrees of visibility and 135 ROIs containing normal breast parenchyma were extracted manually from digitized mammograms as case samples. Spatial-grey-level-dependence (SGLD) matrices of each ROI were calculated and eight texture features were calculated from the SGLD matrices. The correlation and class-distance properties of extracted texture features were analysed. Selected texture features were input into a modified decision-tree classification scheme. The performance of the classifier was evaluated for different feature combinations and orders of features on the tree. A classification accuracy of about 89% sensitivity and 76% specificity was obtained for ordered features, sum average, correlation, and energy, during the training procedure. With a leave-one-out method, the test result was about 76% sensitivity and 64% specificity. The results of this preliminary study demonstrate the feasibility of using texture information for classification of mass and normal breast tissue, which will be likely to be useful for classifying true and false detections in computer-aided diagnosis programmes.     

Thyroid Nodule Recognition Based on Feature Selection and Pixel Classification Methods

Statistical approach is a valuable way to describe texture primitives. The aim of this study is to design and implement a classifier framework to automatically identify the thyroid nodules from ultrasound images. Using rigorous mathematical foundations, this article focuses on developing a discriminative texture analysis method based on texture variations corresponding to four biological areas (normal thyroid, thyroid nodule, subcutaneous tissues, and trachea). Our research follows three steps: automatic extraction of the most discriminative first-order statistical texture features, building a classifier that automatically optimizes and selects the valuable features, and correlating significant texture parameters with the four biological areas of interest based on pixel classification and location characteristics. Twenty ultrasound images of normal thyroid and 20 that present thyroid nodules were used. The analysis involves both the whole thyroid ultrasound images and the region of interests (ROIs). The proposed system and the classification results are validated using the receiver operating characteristics which give a better overall view of the classification performance of methods. It is found that the proposed approach is capable of identifying thyroid nodules with a correct classification rate of 83 % when whole image is analyzed and with a percent of 91 % when the ROIs are analyzed.     

Characterization of Primary and Secondary Malignant Liver Lesions from B-Mode Ultrasound

Characterization of hepatocellular carcinomas (HCCs) and metastatic carcinomas (METs) from B-mode ultrasound presents a daunting challenge for radiologists due to their highly overlapping appearances. The differential diagnosis between HCCs and METs is often carried out by observing the texture of regions inside the lesion and the texture of background liver on which the lesion has evolved. The present study investigates the contribution made by texture patterns of regions inside and outside of the lesions for binary classification between HCC and MET lesions. The study is performed on 51 real ultrasound liver images with 54 malignant lesions, i.e., 27 images with 27 solitary HCCs (13 small HCCs and 14 large HCCs) and 24 images with 27 MET lesions (12 typical cases and 15 atypical cases). A total of 120 within-lesion regions of interest and 54 surrounding lesion regions of interest are cropped from 54 lesions. Subsequently, 112 texture features (56 texture features and 56 texture ratio features) are computed by statistical, spectral, and spatial filtering based texture features extraction methods. A two-step methodology is used for feature set optimization, i.e., feature pruning by removal of nondiscriminatory features followed by feature selection by genetic algorithm–support vector machine (SVM) approach. The SVM classifier is designed based on optimum features. The proposed computer-aided diagnostic system achieved the overall classification accuracy of 91.6 % with sensitivity of 90 % and 93.3 % for HCCs and METs, respectively. The promising results obtained by the proposed system indicate its usefulness to assist radiologists in diagnosing liver malignancies.     

Computer-Aided Diagnosis for Phase-Contrast X-ray Computed Tomography: Quantitative Characterization of Human Patellar Cartilage with High-Dimensional Geometric Features

Phase-contrast computed tomography (PCI-CT) has shown tremendous potential as an imaging modality for visualizing human cartilage with high spatial resolution. Previous studies have demonstrated the ability of PCI-CT to visualize (1) structural details of the human patellar cartilage matrix and (2) changes to chondrocyte organization induced by osteoarthritis. This study investigates the use of high-dimensional geometric features in characterizing such chondrocyte patterns in the presence or absence of osteoarthritic damage. Geometrical features derived from the scaling index method (SIM) and statistical features derived from gray-level co-occurrence matrices were extracted from 842 regions of interest (ROI) annotated on PCI-CT images of ex vivo human patellar cartilage specimens. These features were subsequently used in a machine learning task with support vector regression to classify ROIs as healthy or osteoarthritic; classification performance was evaluated using the area under the receiver-operating characteristic curve (AUC). SIM-derived geometrical features exhibited the best classification performance (AUC, 0.95 ± 0.06) and were most robust to changes in ROI size. These results suggest that such geometrical features can provide a detailed characterization of the chondrocyte organization in the cartilage matrix in an automated and non-subjective manner, while also enabling classification of cartilage as healthy or osteoarthritic with high accuracy. Such features could potentially serve as imaging markers for evaluating osteoarthritis progression and its response to different therapeutic intervention strategies.     

Test–Retest Reproducibility Analysis of Lung CT Image Features

Quantitative size, shape, and texture features derived from computed tomographic (CT) images may be useful as predictive, prognostic, or response biomarkers in non-small cell lung cancer (NSCLC). However, to be useful, such features must be reproducible, non-redundant, and have a large dynamic range. We developed a set of quantitative three-dimensional (3D) features to describe segmented tumors and evaluated their reproducibility to select features with high potential to have prognostic utility. Thirty-two patients with NSCLC were subjected to unenhanced thoracic CT scans acquired within 15 min of each other under an approved protocol. Primary lung cancer lesions were segmented using semi-automatic 3D region growing algorithms. Following segmentation, 219 quantitative 3D features were extracted from each lesion, corresponding to size, shape, and texture, including features in transformed spaces (laws, wavelets). The most informative features were selected using the concordance correlation coefficient across test–retest, the biological range and a feature independence measure. There were 66 (30.14 %) features with concordance correlation coefficient ≥ 0.90 across test–retest and acceptable dynamic range. Of these, 42 features were non-redundant after grouping features with R²_Bet ≥ 0.95. These reproducible features were found to be predictive of radiological prognosis. The area under the curve (AUC) was 91 % for a size-based feature and 92 % for the texture features (runlength, laws). We tested the ability of image features to predict a radiological prognostic score on an independent NSCLC (39 adenocarcinoma) samples, the AUC for texture features (runlength emphasis, energy) was 0.84 while the conventional size-based features (volume, longest diameter) was 0.80. Test–retest and correlation analyses have identified non-redundant CT image features with both high intra-patient reproducibility and inter-patient biological range. Thus making the case that quantitative image features are informative and prognostic biomarkers for NSCLC.

Electronic supplementary material

The online version of this article (doi:10.1007/s10278-014-9716-x) contains supplementary material, which is available to authorized users.     

Extracting Fuzzy Classification Rules from Texture Segmented HRCT Lung Images

Automatic tools for detection and identification of lung and lesion from high-resolution CT (HRCT) are becoming increasingly important both for diagnosis and for delivering high-precision radiation therapy. However, development of robust and interpretable classifiers still presents a challenge especially in case of non-small cell lung carcinoma (NSCLC) patients. In this paper, we have attempted to devise such a classifier by extracting fuzzy rules from texture segmented regions from HRCT images of NSCLC patients. A fuzzy inference system (FIS) has been constructed starting from a feature extraction procedure applied on overlapping regions from the same organs and deriving simple if–then rules so that more linguistically interpretable decisions can be implemented. The proposed method has been tested on 138 regions extracted from CT scan images acquired from patients with lung cancer. Assuming two classes of tissues C1 (healthy tissues) and C2 (lesion) as negative and positive, respectively; preliminary results report an AUC = 0.98 for lesions and AUC = 0.93 for healthy tissue, with an optimal operating condition related to sensitivity = 0.96, and specificity = 0.98 for lesions and sensitivity 0.99, and specificity = 0.94 for healthy tissue. Finally, the following results have been obtained: false-negative rate (FNR) = 6 % (C1), FNR = 2 % (C2), false-positive rate (FPR) = 4 % (C1), FPR = 3 % (C2), true-positive rate (TPR) = 94 %, (C1) and TPR = 98 % (C2).     

Breast Ultrasound Image Classification Based on Multiple-Instance Learning

Breast ultrasound (BUS) image segmentation is a very difficult task due to poor image quality and speckle noise. In this paper, local features extracted from roughly segmented regions of interest (ROIs) are used to describe breast tumors. The roughly segmented ROI is viewed as a bag. And subregions of the ROI are considered as the instances of the bag. Multiple-instance learning (MIL) method is more suitable for classifying breast tumors using BUS images. However, due to the complexity of BUS images, traditional MIL method is not applicable. In this paper, a novel MIL method is proposed for solving such task. First, a self-organizing map is used to map the instance space to the concept space. Then, we use the distribution of the instances of each bag in the concept space to construct the bag feature vector. Finally, a support vector machine is employed for classifying the tumors. The experimental results show that the proposed method can achieve better performance: the accuracy is 0.9107 and the area under receiver operator characteristic curve is 0.96 (p < 0.005).     

A novel approach for detection and classification of mammographic microcalcifications using wavelet analysis and extreme learning machine

The objective of this paper is to reveal the effectiveness of wavelet based tissue texture analysis for microcalcification detection in digitized mammograms using Extreme Learning Machine (ELM). Microcalcifications are tiny deposits of calcium in the breast tissue which are potential indicators for early detection of breast cancer. The dense nature of the breast tissue and the poor contrast of the mammogram image prohibit the effectiveness in identifying microcalcifications. Hence, a new approach to discriminate the microcalcifications from the normal tissue is done using wavelet features and is compared with different feature vectors extracted using Gray Level Spatial Dependence Matrix (GLSDM) and Gabor filter based techniques. A total of 120 Region of Interests (ROIs) extracted from 55 mammogram images of mini-Mias database, including normal and microcalcification images are used in the current research. The network is trained with the above mentioned features and the results denote that ELM produces relatively better classification accuracy (94%) with a significant reduction in training time than the other artificial neural networks like Bayesnet classifier, Naivebayes classifier, and Support Vector Machine. ELM also avoids problems like local minima, improper learning rate, and over fitting.     

Computerized Classification of Pneumoconiosis on Digital Chest Radiography Artificial Neural Network with Three Stages

It is difficult for radiologists to classify pneumoconiosis from category 0 to category 3 on chest radiographs. Therefore, we have developed a computer-aided diagnosis (CAD) system based on a three-stage artificial neural network (ANN) method for classification based on four texture features. The image database consists of 36 chest radiographs classified as category 0 to category 3. Regions of interest (ROIs) with a matrix size of 32 × 32 were selected from chest radiographs. We obtained a gray-level histogram, histogram of gray-level difference, gray-level run-length matrix (GLRLM) feature image, and gray-level co-occurrence matrix (GLCOM) feature image in each ROI. For ROI-based classification, the first ANN was trained with each texture feature. Next, the second ANN was trained with output patterns obtained from the first ANN. Finally, we obtained a case-based classification for distinguishing among four categories with the third ANN method. We determined the performance of the third ANN by receiver operating characteristic (ROC) analysis. The areas under the ROC curve (AUC) of the highest category (severe pneumoconiosis) case and the lowest category (early pneumoconiosis) case were 0.89 ± 0.09 and 0.84 ± 0.12, respectively. The three-stage ANN with four texture features showed the highest performance for classification among the four categories. Our CAD system would be useful for assisting radiologists in classification of pneumoconiosis from category 0 to category 3.     

肝包虫病和肝囊肿CT图像的分型研究

探讨计算机辅助诊断技术在肝包虫病和肝囊肿CT图像分型中的应用。方法 对单囊型肝包虫病和单发性肝囊肿CT图像感兴趣区域,分别使用传统的预处理方法和图像融合方法,提取原始ROI、预处理后的和融合后的ROI图像Haar小波、DB2小波、Tamura、Gabor滤波器和灰度-梯度共生矩阵特征,通过支持向量机和BP神经网络分类模型分类,比较三种方法的分类准确率,并对各分类模型进行参数评估。结果 从原始ROI图像直接提取的Haar小波、DB2小波、Tamura和GGCM特征的最佳分类准确率均达到了95%以上;融合后的ROI图像五种特征的分类准确率都较高,在90%以上。结论 本研究所使用的方法应用于肝包虫病和肝囊肿CT图像的分型中具有一定的分类优势,为影像学诊断提供依据。     

Prostate cancer multi-feature analysis using trans-rectal ultrasound images

This note focuses on extracting and analysing prostate texture features from trans-rectal ultrasound (TRUS) images for tissue characterization. One of the principal contributions of this investigation is the use of the information of the images' frequency domain features and spatial domain features to attain a more accurate diagnosis. Each image is divided into regions of interest (ROIs) by the Gabor multi-resolution analysis, a crucial stage, in which segmentation is achieved according to the frequency response of the image pixels. The pixels with a similar response to the same filter are grouped to form one ROI. Next, from each ROI two different statistical feature sets are constructed; the first set includes four grey level dependence matrix (GLDM) features and the second set consists of five grey level difference vector (GLDV) features. These constructed feature sets are then ranked by the mutual information feature selection (MIFS) algorithm. Here, the features that provide the maximum mutual information of each feature and class (cancerous and non-cancerous) and the minimum mutual information of the selected features are chosen, yielding a reduced feature subset. The two constructed feature sets, GLDM and GLDV, as well as the reduced feature subset, are examined in terms of three different classifiers: the condensed k-nearest neighbour (CNN), the decision tree (DT) and the support vector machine (SVM). The accuracy classification results range from 87.5% to 93.75%, where the performance of the SVM and that of the DT are significantly better than the performance of the CNN.     

基于Local Jet变换空间纹理特征的肺结节分类方法

为了在纹理特征下改善肺结节良、恶性的模式识别,提出一种基于local jet变换空间的纹理特征提取方法。首先利用二维高斯函数的前三阶偏微分函数将结节原图像变换到local jet纹理图像空间,然后利用纹理描述子在该空间提取特征参数。以灰度值的前四阶矩和基于灰度共生矩阵的特征参数作为纹理描述子,分别提取结节原图像和变换后纹理图像的特征参数,以BP神经网络作为分类器,对同一纹理描述子下的2个不同图像空间的经核主成分分析优化后的特征参数集进行结节良、恶性分类。以157个肺结节(51个良性,106个恶性)作为实验数据进行对比实验,结果显示:两种纹理描述子基于local jet变换空间提取的特征参数分别获得82.69%和86.54%的分类正确率,较原图像空间提高6%～8%,同时AUC值提高约10%。实验结果表明,基于local jet变换空间提取的纹理特征可以有效地改善肺结节良、恶性的模式识别。     

A wavelet-based optimal texture feature set for classification of brain tumours

In this work, the classification of brain tumours in magnetic resonance images is studied by using optimal texture features. These features are used to classify three sets of brain images—normal brain, benign tumour and malignant tumour. A wavelet-based texture feature set is derived from the region of interest. Each selected brain region of interest is characterized with both its energy and texture features extracted from the selected high frequency subband. An artificial neural network classifier is employed to evaluate the performance of these features. Feature selection is performed by a genetic algorithm. Principal component analysis and classical sequential methods are compared against the genetic approach in terms of the best recognition rate achieved and the optimal number of features. A classification performance of 98% is achieved in a genetic algorithm with only four of the available 29 features. Principal component analysis and classical sequential methods require a larger feature set to attain the similar classification accuracy of 98%. The optimal texture features such as range of angular second moment, range of sum variance, range of information measure of correlation II and energy selected by the genetic algorithm provide best classification performance with lower computational effort.