A computer-aided diagnostic system to characterize CT focal liver lesions: design and optimization of a neural network classifier

Content-based image retrieval of multiphase CT images for focal liver lesion characterization

PURPOSE

Characterization of focal liver lesions with various imaging modalities can be very challenging in the clinical practice and is experience-dependent. The authors' aim is to develop an automatic method to facilitate the characterization of focal liver lesions (FLLs) using multiphase computed tomography (CT) images by radiologists.

METHODS

A multiphase-image retrieval system is proposed to retrieve a preconstructed database of FLLs with confirmed diagnoses, which can assist radiologists' decision-making in FLL characterization. It first localizes the FLL on multiphase CT scans using a hybrid generative-discriminative FLL detection method and a nonrigid B-spline registration method. Then, it extracts the multiphase density and texture features to numerically represent the FLL. Next, it compares the query FLL with the model FLLs in the database in terms of the feature and measures their similarities using the L1-norm based similarity scores. The model FLLs are ranked by similarities and the top results are finally provided to the users for their evidence studies.

RESULTS

The system was tested on a database of 69 four-phase contrast-enhanced CT scans, consisting of six classes of liver lesions, and evaluated in terms of the precision-recall curve and the Bull's Eye Percentage Score (BEP). It obtained a BEP score of 78%. Compared with any single-phase based representation, the multiphase-based representation increased the BEP scores of the system, from 63%-65% to 78%. In a pilot study, two radiologists performed characterization of FLLs without and with the knowledge of the top five retrieved results. The results were evaluated in terms of the diagnostic accuracy, the receiver operating characteristic (ROC) curve and the mean diagnostic confidence. One radiologist's accuracy improved from 75% to 92%, the area under ROC curves (AUC) from 0.85 to 0.95 (p = 0.081), and the mean diagnostic confidence from 4.6 to 7.3 (p = 0.039). The second radiologist's accuracy did not change, at 75%, with AUC increasing from 0.72 to 0.75 (p = 0.709), and the mean confidence from 4.5 to 4.9 (p = 0.607).

CONCLUSIONS

Multiphase CT images can be used in content-based image retrieval for FLL's categorization and result in good performance in comparison with single-phase CT images. The proposed method has the potential to improve the radiologists' diagnostic accuracy and confidence by providing visually similar lesions with confirmed diagnoses for their interpretation of clinical studies.

A retrieval-based computer-aided diagnosis system for the characterization of liver lesions in CT scans

RATIONALE AND OBJECTIVES

To evaluate a computer-aided diagnosis (CADx) system for the characterization of liver lesions in computed tomography (CT) scans. The stand-alone predictive performance of the CADx system was assessed and compared to that of three radiologists who were provided with the same amount of image information to which the CADx system had access.

MATERIALS AND METHODS

The CADx system operates as an image search engine exploiting texture analysis of liver lesion image data for the lesion in question and lesions from a database. A region of interest drawn around an indeterminate liver lesion is used as input query. The CADx system retrieves lesions of similar histology (benign/malignant), density (hypodense/hyperdense), or type (cyst/hemangioma/metastasis). The system's performance was evaluated with leave-one-patient-out receiver operating characteristic area under the curve on 685 CT scans from 372 patients that contained 2325 liver lesions (193 <1 cm(3)). Sensitivity, specificity, and positive and negative predictive values were evaluated separately for subcentimeter lesions. Results were compared to those of three radiologists who rated 83 liver lesions (20 hemangiomas, 20 metastases, 20 cysts, 20 hepatocellular carcinomas, and 3 focal nodular hyperplasias) displaying only the liver.

RESULTS

The CADx system's leave-one-patient-out receiver operating characteristic area under the curve was 97.1% for density, 91.4% for histology, and 95.5% for lesion type. For subcentimeter lesions, input of additional semantic information improved the system's performance. The CADx system has been proved to significantly outperform radiologists in discriminating lesion histology and type, provided the radiologists have no access to information other than the image. The radiologists were most reliable in diagnosing hemangioma given the limited image data.

CONCLUSIONS

The CADx system under study discriminated reliably between various liver lesions, even outperforming radiologists when accessing the same image information and demonstrated promising performance in classifying subcentimeter lesions in particular.

Content-based retrieval of focal liver lesions using bag-of-visual-words representations of single- and multiphase contrast-enhanced CT images

This paper is aimed at developing and evaluating a content-based retrieval method for contrast-enhanced liver computed tomographic (CT) images using bag-of-visual-words (BoW) representations of single and multiple phases. The BoW histograms are extracted using the raw intensity as local patch descriptor for each enhance phase by densely sampling the image patches within the liver lesion regions. The distance metric learning algorithms are employed to obtain the semantic similarity on the Hellinger kernel feature map of the BoW histograms. The different visual vocabularies for BoW and learned distance metrics are evaluated in a contrast-enhanced CT image dataset comprised of 189 patients with three types of focal liver lesions, including 87 hepatomas, 62 cysts, and 60 hemangiomas. For each single enhance phase, the mean of average precision (mAP) of BoW representations for retrieval can reach above 90 % which is significantly higher than that of intensity histogram and Gabor filters. Furthermore, the combined BoW representations of the three enhance phases can improve mAP to 94.5 %. These preliminary results demonstrate that the BoW representation is effective and feasible for retrieval of liver lesions in contrast-enhanced CT images.

Quantifying tumour heterogeneity with CT

Heterogeneity is a key feature of malignancy associated with adverse tumour biology. Quantifying heterogeneity could provide a useful non-invasive imaging biomarker. Heterogeneity on computed tomography (CT) can be quantified using texture analysis which extracts spatial information from CT images (unenhanced, contrast-enhanced and derived images such as CT perfusion) that may not be perceptible to the naked eye. The main components of texture analysis can be categorized into image transformation and quantification. Image transformation filters the conventional image into its basic components (spatial, frequency, etc.) to produce derived subimages. Texture quantification techniques include structural-, model- (fractal dimensions), statistical- and frequency-based methods. The underlying tumour biology that CT texture analysis may reflect includes (but is not limited to) tumour hypoxia and angiogenesis. Emerging studies show that CT texture analysis has the potential to be a useful adjunct in clinical oncologic imaging, providing important information about tumour characterization, prognosis and treatment prediction and response.

Medical decision-making system of ultrasound carotid artery intima-media thickness using neural networks

The objective of this work is to develop and implement a medical decision-making system for an automated diagnosis and classification of ultrasound carotid artery images. The proposed method categorizes the subjects into normal, cerebrovascular, and cardiovascular diseases. Two contours are extracted for each and every preprocessed ultrasound carotid artery image. Two types of contour extraction techniques and multilayer back propagation network (MBPN) system have been developed for classifying carotid artery categories. The results obtained show that MBPN system provides higher classification efficiency, with minimum training and testing time. The outputs of decision support system are validated with medical expert to measure the actual efficiency. MBPN system with contour extraction algorithms and preprocessing scheme helps in developing medical decision-making system for ultrasound carotid artery images. It can be used as secondary observer in clinical decision making.

Diagnostic Support Systems and Computational Intelligence

Recent advances in computer science provide the intelligent computation tools needed to design and develop Diagnostic Support Systems (DSSs) that promise to increase the efficiency of physicians during their clinical practice. This chapter provides a brief overview of the use of computational intelligence methods in the design and development of DSSs aimed at the differential diagnosis of hepatic lesions from Computed Tomography (CT) images. Furthermore, examples of DSSs developed by our research team for supporting the diagnosis of focal liver lesions from non-enhanced CT images are presented.

The recent progress in quantitative medical image analysis for computer aided diagnosis systems

Computer-aided diagnosis (CAD) has become one of the major research subjects in medical imaging and diagnostic radiology. Many different CAD schemes are being developed for use in the detection and/or characterization of various lesions found through various types of medical imaging. These imaging technologies employ conventional projection radiography, computed tomography, magnetic resonance imaging, ultrasonography, etc. In order to achieve a high performance level for a computerized diagnosis, it is important to employ effective image analysis techniques in the major steps of a CAD scheme. The main objective of this review is to attempt to introduce the diverse methods used for quantitative image analysis, and to provide a guide for clinicians.

Automated medical image segmentation techniques

Accurate segmentation of medical images is a key step in contouring during radiotherapy planning. Computed topography (CT) and Magnetic resonance (MR) imaging are the most widely used radiographic techniques in diagnosis, clinical studies and treatment planning. This review provides details of automated segmentation methods, specifically discussed in the context of CT and MR images. The motive is to discuss the problems encountered in segmentation of CT and MR images, and the relative merits and limitations of methods currently available for segmentation of medical images.

Dynamic contrast-enhanced texture analysis of the liver: initial assessment in colorectal cancer

PURPOSE

To undertake an initial assessment of the potential utility of dynamic contrast-enhanced texture analysis (DCE-TA) of the liver in patients with colorectal cancer.

MATERIALS AND METHODS

TA comprised measurement of mean gray-level intensity, entropy, and uniformity with and without selective-scale filtration using a band-pass filter to highlight different spatial frequencies reflecting fine, medium, and coarse textures. An initial phantom study assessed the sensitivity of each texture qualifier to computed tomography (CT) acquisition parameters. Texture was analyzed in DCE-CT series from 27 colorectal cancer patients having apparently normal hepatic morphology (node-negative: n = 8, node-positive: n = 19). Averaged changes in hepatic texture induced by contrast material were assessed qualitatively and quantitatively by using kinetic modeling to calculate hepatic perfusion indices following fine, medium, and coarse image filtration.

RESULTS

All texture qualifiers were less sensitive to changes in CT acquisition parameters than measurement of CT attenuation. Temporal changes in hepatic texture were qualitatively different from changes in enhancement. Statistically significant differences between node-negative and node-positive patients were observed for at least 1 time period for measurements of hepatic enhancement and for all texture parameters. The differences were most statistically significant and occurred over the greatest number of time periods for fine texture quantified as mean gray-level intensity (5 time periods, minimum P value: 0.006) followed by fine texture quantified as entropy (4 time points, minimum P value: 0.006). There was no difference in hepatic perfusion indices for the 2 groups.

CONCLUSIONS

DCE-TA is a potentially useful adjunct to DCE-CT warranting further investigation.

Content-based image retrieval in radiology: current status and future directions

Diagnostic radiology requires accurate interpretation of complex signals in medical images. Content-based image retrieval (CBIR) techniques could be valuable to radiologists in assessing medical images by identifying similar images in large archives that could assist with decision support. Many advances have occurred in CBIR, and a variety of systems have appeared in nonmedical domains; however, permeation of these methods into radiology has been limited. Our goal in this review is to survey CBIR methods and systems from the perspective of application to radiology and to identify approaches developed in nonmedical applications that could be translated to radiology. Radiology images pose specific challenges compared with images in the consumer domain; they contain varied, rich, and often subtle features that need to be recognized in assessing image similarity. Radiology images also provide rich opportunities for CBIR: rich metadata about image semantics are provided by radiologists, and this information is not yet being used to its fullest advantage in CBIR systems. By integrating pixel-based and metadata-based image feature analysis, substantial advances of CBIR in medicine could ensue, with CBIR systems becoming an important tool in radiology practice.

A computer-aided diagnostic system to discriminate SPIO-enhanced magnetic resonance hepatocellular carcinoma by a neural network classifier

In this paper, a computer-aided diagnostic (CAD) system for the classification of rat liver lesions from MR imaging is presented. The proposed system consists of two modules: the feature extraction and the classification modules. 40 rats are used for hepatocellular carcinoma (HCC) induction with Diethylnitrosamine via drinking water. After Resovist is administrated by tail vein the animals are scanned by a 1.5-T MR scanner with T2-weighted FRFSE sequence. SPIO-enhanced images of 106 nodules (RNs(:) 24, HCCs: 82) are acquired, and 161 regions of interest (ROIs) are taken from the MR images .Six parameters of texture characteristics including Angular Second Moment, Contrast, Correlation, Inverse Difference Moment, Entropy, and Variance of 161 ROIs are calculated and assessed by gray-level co-occurrence matrices, then fed into a BP neural network (NN) classifier to classify the liver tissue into two classes: cirrhosis and HCC. Difference of each texture parameter between cirrhosis and HCC group is significant. The accuracy of classification of HCC nodules from cirrhosis is 91.67%. It indicates the ANN classifier based on texture is effective for classifying HCC nodules from cirrhosis on rat SPIO-enhanced imaging.

Comparison and evaluation of methods for liver segmentation from CT datasets

This paper presents a comparison study between 10 automatic and six interactive methods for liver segmentation from contrast-enhanced CT images. It is based on results from the "MICCAI 2007 Grand Challenge" workshop, where 16 teams evaluated their algorithms on a common database. A collection of 20 clinical images with reference segmentations was provided to train and tune algorithms in advance. Participants were also allowed to use additional proprietary training data for that purpose. All teams then had to apply their methods to 10 test datasets and submit the obtained results. Employed algorithms include statistical shape models, atlas registration, level-sets, graph-cuts and rule-based systems. All results were compared to reference segmentations five error measures that highlight different aspects of segmentation accuracy. All measures were combined according to a specific scoring system relating the obtained values to human expert variability. In general, interactive methods reached higher average scores than automatic approaches and featured a better consistency of segmentation quality. However, the best automatic methods (mainly based on statistical shape models with some additional free deformation) could compete well on the majority of test images. The study provides an insight in performance of different segmentation approaches under real-world conditions and highlights achievements and limitations of current image analysis techniques.

Computer-aided image analysis of focal hepatic lesions in ultrasonography: preliminary results

PURPOSE

To develop a computer-aided image analysis (CAIA) algorithm for analyzing US features of focal hepatic lesions and to correlate the feature values of CAIA with radiologists' grading.

MATERIALS AND METHODS

Two abdominal radiologists, blinded to the final diagnosis, independently evaluated sonographic images of 51 focal hepatic lesions in 47 patients: hemangiomas (n = 19), hepatic simple cysts or cystic lesions (n = 14), hepatocellular carcinoma (n = 11), metastases (n = 6), and focal fat deposition (n = 1). All images were graded using a 3- to 5-point scale, in terms of border (roundness, sharpness, and the presence of peripheral rim), texture (echogenicity, homogeneity, and internal artifact), posterior enhancement, and lesion conspicuity. Using a CAIA, texture and morphological parameters representing radiologists' subjective evaluations were extracted. Correlations between the radiologists and the CAIA for assessing parameters in corresponding categories were computed by means of weighted kappa statistics and Spearman correlation test.

RESULTS

A good agreement was achieved between CAIA and radiologists for grading echogenicity (weighted kappa = 0.675) and the presence of hyper- or hypoechoic rim (weighted kappa = 0.743). Several CAIA-derived features representing homogeneity of the lesions showed good correlations (correlation coefficient (gamma) = 0.603 approximately 0.641) with radiologists' grading (P < 0.05). For internal artifact (gamma = 0.469-0.490) and posterior enhancement (gamma = -0.516) of the cyst and lesion conspicuity (gamma = 0.410), a fair correlation between CAIA and radiologists was obtained (P < 0.05). However, parameters representing lesions' border such as sharpness (gamma = 0.252-0.299) and roundness (gamma = -0.134-0.163) showed no significant correlation (P > 0.05).

CONCLUSION

As a preliminary step in US computer-aided diagnosis for focal hepatic lesions, a CAIA algorithm was constructed with a good agreement and correlation with human observers in most US features. In addition, these features should be weighted highly when a computer-aided diagnosis for characterizing focal liver lesions on US is designed and developed.

Colorectal cancer: texture analysis of portal phase hepatic CT images as a potential marker of survival

PURPOSE

To assess the utility of texture analysis of liver computed tomographic (CT) images by determining the effect of acquisition parameters on texture and by comparing the abilities of texture analysis and hepatic perfusion CT to help predict survival for patients with colorectal cancer.

MATERIALS AND METHODS

The study comprised a phantom test and a clinical evaluation of 48 patients with colorectal cancer who had consented to retrospective analysis of hepatic perfusion CT data acquired during a research study approved by the institutional review board. Both components involved texture analysis to quantify the relative contribution of CT features between 2 and 12 pixels wide to overall image brightness and uniformity. The effect of acquisition factors on texture was assessed on CT images of a cylindric phantom filled with water obtained by using tube currents between 100 and 250 mAs and voltages between 80 and 140 kVp. Texture on apparently normal portal phase CT images of the liver and hepatic perfusion parameters were related to patient survival by using Kaplan-Meier survival analysis.

RESULTS

A texture parameter that compared the uniformity of distribution of CT image features 10 and 12 pixels wide exhibited the least variability with CT acquisition parameters (maximum coefficient of variation, 2.6%) and was the best predictor of patient survival (P < .005). There was no significant association between survival and hepatic perfusion parameters.

CONCLUSION

The study provides preliminary evidence that analysis of liver texture on portal phase CT images is potentially a superior predictor of survival for patients with colorectal cancer than CT perfusion imaging.

SUPPLEMENTAL MATERIAL

http://radiology.rsnajnls.org/cgi/content/full/2502071879/DC1.

Computer-aided diagnosis for the classification of focal liver lesions by use of contrast-enhanced ultrasonography

The authors developed a computer-aided diagnostic (CAD) scheme for classifying focal liver lesions (FLLs) as liver metastasis, hemangioma, and three histologic differentiation types of hepatocellular carcinoma (HCC), by use of microflow imaging (MFI) of contrast-enhanced ultrasonography. One hundred and three FLLs obtained from 97 cases used in this study consisted of 26 metastases (15 hyper- and 11 hypovascularity types), 16 hemangiomas (five hyper- and 11 hypovascularity types) and 61 HCCs: 24 well differentiated (w-HCC), 28 moderately differentiated (m-HCC), and nine poorly differentiated (p-HCC). Pathologies of all cases were determined based on biopsy or surgical specimens. Locations and contours of FLLs on contrast-enhanced images were determined manually by an experienced physician. MFI was obtained with contrast-enhanced low-mechanical-index (MI) pulse subtraction imaging at a fixed plane which included a distinctive cross section of the FLL. In MFI, the inflow high signals in the plane, which were due to the vascular patterns and the contrast agent, were accumulated following flash scanning with a high-MI ultrasound exposure. In the initial step of our computerized scheme, a series of the MFI images was extracted from the original cine clip (AVI format). We applied a smoothing filter and time-sequential running average techniques in order to reduce signal noise on the single MFI image and cyclic noise on the sequential MFI images, respectively. A kidney, vessels, and a liver parenchyma region were segmented automatically by use of the last image of a series of MFI images. The authors estimated time-intensity curves for an FLL by use of a series of the temporally averaged MFI images in order to determine temporal features such as estimated replenishment times at early and delayed phases, flow rates, and peak times. In addition, they extracted morphologic and gray-level image features which were determined based on the physicians' knowledge of the diagnosis of the FLL, such as the size of lesion, vascular patterns, and the presence of hypoechoic regions. They employed a cascade of six independent artificial neural networks (ANNs) by use of extracted temporal and image features for classifying five types of liver diseases. A total of 16 temporal and image features, which were selected from 43 initially extracted features, were used for six different ANNs for making decisions at each decision in the cascade. The ANNs were trained and tested with a leave-one-lesion-out test method. The classification accuracies for the 103 FLLs were 88.5% for metastasis, 93.8% for hemangioma, and 86.9% for all HCCs. In addition, the classification accuracies for histologic differentiation types of HCCs were 79.2% for w-HCC, 50.0% for m-HCC, and 77.8% for p-HCC. The CAD scheme for classifying FLLs by use of the MFI on contrast-enhanced ultrasonography has the potential to improve the diagnostic accuracy in the histologic diagnosis of HCCs and the other liver diseases.

Segmentation and classification of medical images using texture-primitive features: Application of BAM-type artificial neural network

The objective of developing this software is to achieve auto-segmentation and tissue characterization. Therefore, the present algorithm has been designed and developed for analysis of medical images based on hybridization of syntactic and statistical approaches, using artificial neural network (ANN). This algorithm performs segmentation and classification as is done in human vision system, which recognizes objects; perceives depth; identifies different textures, curved surfaces, or a surface inclination by texture information and brightness. The analysis of medical image is directly based on four steps: 1) image filtering, 2) segmentation, 3) feature extraction, and 4) analysis of extracted features by pattern recognition system or classifier. In this paper, an attempt has been made to present an approach for soft tissue characterization utilizing texture-primitive features with ANN as segmentation and classifier tool. The present approach directly combines second, third, and fourth steps into one algorithm. This is a semisupervised approach in which supervision is involved only at the level of defining texture-primitive cell; afterwards, algorithm itself scans the whole image and performs the segmentation and classification in unsupervised mode. The algorithm was first tested on Markov textures, and the success rate achieved in classification was 100%; further, the algorithm was able to give results on the test images impregnated with distorted Markov texture cell. In addition to this, the output also indicated the level of distortion in distorted Markov texture cell as compared to standard Markov texture cell. Finally, algorithm was applied to selected medical images for segmentation and classification. Results were in agreement with those with manual segmentation and were clinically correlated.

Evaluation of texture features in hepatic tissue characterization from non-enhanced CT images

Aim of this paper is to evaluate the diagnostic contribution of various types of texture features in discrimination of hepatic tissue in abdominal non-enhanced Computed Tomography (CT) images. Regions of Interest (ROIs) corresponding to the classes: normal liver, cyst, hemangioma, and hepatocellular carcinoma were drawn by an experienced radiologist. For each ROI, five distinct sets of texture features are extracted using First Order Statistics (FOS), Spatial Gray Level Dependence Matrix (SGLDM), Gray Level Difference Method (GLDM), Laws' Texture Energy Measures (TEM), and Fractal Dimension Measurements (FDM). In order to evaluate the ability of the texture features to discriminate the various types of hepatic tissue, each set of texture features, or its reduced version after genetic algorithm based feature selection, was fed to a feed-forward Neural Network (NN) classifier. For each NN, the area under Receiver Operating Characteristic (ROC) curves (Az) was calculated for all one-vs-all discriminations of hepatic tissue. Additionally, the total Az for the multi-class discrimination task was estimated. The results show that features derived from FOS perform better than other texture features (total Az: 0.802+/-0.083) in the discrimination of hepatic tissue.

Computer-aided identification of ovarian cancer in confocal microendoscope images

The confocal microendoscope is an instrument for imaging the surface of the human ovary. Images taken with this instrument from normal and diseased tissue show significant differences in cellular distribution. A real-time computer-aided system to facilitate the identification of ovarian cancer is introduced. The cellular-level structure present in ex vivo confocal microendoscope images is modeled as texture. Features are extracted based on first-order statistics, spatial gray-level-dependence matrices, and spatial-frequency content. Selection of the features is performed using stepwise discriminant analysis, forward sequential search, a nonparametric method, principal component analysis, and a heuristic technique that combines the results of these other methods. The selected features are used for classification, and the performance of various machine classifiers is compared by analyzing areas under their receiver operating characteristic curves. The machine classifiers studied included linear discriminant analysis, quadratic discriminant analysis, and the k-nearest-neighbor algorithm. The results suggest it is possible to automatically identify pathology based on texture features extracted from confocal microendoscope images and that the machine performance is superior to that of a human observer.

Texture analysis in non-contrast enhanced CT: impact of malignancy on texture in apparently disease-free areas of the liver

OBJECTIVES

To determine whether texture analysis of non-contrast enhanced computed tomography (CT) images in apparently disease-free areas of the liver is altered by the presence of extra- and intra-hepatic malignancy in colorectal cancer patients.

MATERIALS AND METHODS

Hepatic attenuation and texture were assessed from non-contrast enhanced CT in three groups of colorectal cancer patients: (A) 15 controls with no malignancy; (B) nine patients with extra-hepatic malignancy but no liver involvement; (C) eight patients with hepatic metastases. Regions of interest were manually constructed only over apparently normal areas of liver tissue excluding major blood vessels and areas of intra-hepatic fat, which may otherwise alter CT texture irrespective of the presence of malignancy. Texture was analysed on unfiltered images and following band-pass image filtration to highlight image features at different spatial frequencies (fine: 2 pixels/1.68 mm in width, medium: 6 pixels/5.04 mm and coarse: 12 pixels/10.08 mm). The relative contributions made to the image by features at two different spatial frequencies were expressed as filter ratios (fine/medium, fine/coarse and medium/coarse). Texture was quantified as mean grey-level intensity, entropy and uniformity.

RESULTS

Texture was not altered on unfiltered images whereas relative texture analysis following image filtration identified differences in fine to medium texture ratios in apparently disease-free areas of the liver in patients with hepatic metastases as compared to patients with no tumour (entropy, p=0.0257) and patients with extra-hepatic disease (uniformity, p=0.0143).

CONCLUSIONS

Relative texture analysis of unenhanced hepatic CT can reveal changes in apparently disease-free areas of the liver that have previously required more complex perfusion measurements for detection.

Hepatic enhancement in colorectal cancer: texture analysis correlates with hepatic hemodynamics and patient survival

RATIONALE AND OBJECTIVES

Perfusion imaging of the liver has attracted interest as a potential means for earlier detection of hepatic metastases, but the techniques are complex and do not form part of routine imaging protocols. This study assesses whether the hemodynamic status of the liver of patients with colorectal cancer but apparently normal hepatic morphology is reflected by texture features within a single portal-phase contrast enhanced computed tomography (CT) image and correlates texture with overall survival.

MATERIALS AND METHODS

Portal-phase CT images from 27 patients with colorectal cancer but no apparent hepatic metastases were processed using a band-pass filter that highlighted image features at different spatial frequencies. A range of parameters reflecting liver texture on filtered images were correlated against CT hepatic perfusion index (HPI) and patient survival.

RESULTS

After image filtration, entropy values from hepatic regions were inversely correlated with HPI (r=-0.503978, P=.007355), and directly correlated with survival (r=0.489642, P=.009533). An entropy value below 2.0 identified four patients who died within 36 months of their CT scan with sensitivity 100% and specificity 65% (P<.03).

CONCLUSION

The hemodynamic status of the liver is reflected by subtle changes in coarse texture on portal phase images that can be revealed by texture analysis. Texture analysis has the potential to identify colorectal cancer patients with an apparently normal portal phase hepatic CT but reduced survival.

In search of biologic correlates for liver texture on portal-phase CT

RATIONALE AND OBJECTIVES

The acceptance of computer-assisted diagnosis (CAD) in clinical practice has been constrained by the scarcity of identifiable biologic correlates for CAD-based image parameters. This study aims to identify biologic correlates for computed tomography (CT) liver texture in a series of patients with colorectal cancer.

MATERIALS AND METHODS

In 28 patients with colorectal cancer, total hepatic perfusion (THP), hepatic arterial perfusion, and hepatic portal perfusion (HPP) were measured using perfusion CT. Hepatic glucose use was also determined from positron emission tomography (PET) and expressed as standardized uptake value (SUV). A hepatic phosphorylation fraction index (HPFI) was determined from both SUV and THP. These physiologic parameters were correlated with CAD parameters namely hepatic densitometry, selective-scale, and relative-scale texture features in apparently normal areas of portal-phase hepatic CT.

RESULTS

For patients without liver metastases, a relative-scale texture parameter correlated inversely with SUV (r = -0.587, P = .007) and, positively with THP (r = 0.512, P = .021) and HPP (r = 0.451, P = .046). However, this relative texture parameter correlated most significantly with HPFI (r = -0.590, P = .006). For patients with liver metastases, although not significant an opposite trend was observed between these physiologic parameters and relative texture features (THP: r < -0.4, HPFI: r > 0.35).

CONCLUSION

Total hepatic blood flow and glucose metabolism are two distinct but related biologic correlates for liver texture on portal phase CT, providing a rationale for the use of hepatic texture analysis as a indicator for patients with colorectal cancer.

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.

Classification of Liver Disease from CT Images Using a Support Vector Machine

We propose a classifier based on the support vector machine (SVM) for automatic classification in liver disease. The SVM, stemming from statistical learning theory, involves state-of-the-art machine learning. The classifier is a part of computer-aided diagnosis (CADx), which assists radiologists in accurately diagnosing liver disease. We formulate discriminating between cysts, hepatoma, cavernous hemangioma, and normal tissue as a supervised learning problem, and apply SVM to classifying the diseases using gray level and co-occurrence matrix features and region-based shape descriptors, calculated from regions of interest (ROIs), as input. Significant features of ROI enable us to simplify SVM input space and to feed the SVM representative information. By simplifying and clarifying the diagnosis process, we separate the classification of liver disease into hierarchical multiclass classification. We use the receiver operating characteristic (ROC) curve to evaluate diagnosis performance, demonstrating the classifier’s good performance.

Computer-aided diagnosis of carotid atherosclerosis based on ultrasound image statistics, laws' texture and neural networks

Quantitative characterisation of carotid atherosclerosis and classification into symptomatic or asymptomatic is crucial in planning optimal treatment of atheromatous plaque. The computer-aided diagnosis (CAD) system described in this paper can analyse ultrasound (US) images of carotid artery and classify them into symptomatic or asymptomatic based on their echogenicity characteristics. The CAD system consists of three modules: a) the feature extraction module, where first-order statistical (FOS) features and Laws' texture energy can be estimated, b) the dimensionality reduction module, where the number of features can be reduced using analysis of variance (ANOVA), and c) the classifier module consisting of a neural network (NN) trained by a novel hybrid method based on genetic algorithms (GAs) along with the back propagation algorithm. The hybrid method is able to select the most robust features, to adjust automatically the NN architecture and to optimise the classification performance. The performance is measured by the accuracy, sensitivity, specificity and the area under the receiver-operating characteristic (ROC) curve. The CAD design and development is based on images from 54 symptomatic and 54 asymptomatic plaques. This study demonstrates the ability of a CAD system based on US image analysis and a hybrid trained NN to identify atheromatous plaques at high risk of stroke.

A computer-aided temporal and dynamic subtraction technique of the liver for detection of small hepatocellular carcinomas on abdominal CT images

It is often difficult for radiologists to identify small hepatocellular carcinomas (HCCs) due to insufficient contrast enhancement. Therefore, we have developed a new computer-aided temporal and dynamic subtraction technique to enhance small HCCs, after automatically selecting images set at the same anatomical position from the present (non-enhanced and arterial-phase CT images) and previous images. The present study was performed with CT images from 14 subjects. First, we used template-matching based on similarities in liver shape between the present (non-enhanced and arterial-phase CT images) and previous arterial-phase CT images at the same position. Temporal subtraction images were then obtained by subtraction of the previous image from the present image taken at the same position of the liver. Dynamic subtraction images were also obtained by subtraction of non-enhanced CT images from arterial-phase CT images taken at the same position of the liver. Twenty-one of 22 nodules (95.5%) with contrast enhancement were visualized in temporal and dynamic subtraction images. Compared with present arterial-phase CT images, increases of 150% and 140% in nodule-to-liver contrast were observed on dynamic and temporal subtraction images, respectively. These subtraction images may be useful as reference images in the detection of small moderately differentiated HCCs.

Diagnosis of hepatic tumors with texture analysis in nonenhanced computed tomography images

RATIONALE AND OBJECTIVES

Computed tomography (CT) after iodinated contrast agent injection is highly accurate for diagnosis of hepatic tumors. However, iodinating may have problems of renal toxicity and allergic reaction. We aimed to evaluate the potential role of the computer-aided diagnosis (CAD) with texture analysis in the differential of hepatic tumors on nonenhanced CT.

MATERIALS AND METHODS

This study evaluated 164 liver lesions (80 malignant tumors and 84 hemangiomas). The suspicious tumor region in the digitized CT image was manually selected and extracted as a circular subimage. Proposed preprocessing adjustments for subimages were used to equalize the information needed for a differential diagnosis. The autocovariance texture features of subimage were extracted and a support vector machine classifier identified the tumor as benign or malignant.

RESULTS

The accuracy of the proposed diagnosis system for classifying malignancies is 81.7%, the sensitivity is 75.0%, the specificity is 88.1%, the positive predictive value is 85.7%, and the negative predictive value is 78.7%.

CONCLUSIONS

This system differentiates benign from malignant hepatic tumors with relative high accuracy and is therefore clinically useful to reduce patients needed for iodinated contrast agent injection in CT examination. Because the support vector machine is trainable, it could be further optimized if a larger set of tumor images is to be supplied.

The use of artificial neural networks in decision support in cancer: a systematic review

Artificial neural networks have featured in a wide range of medical journals, often with promising results. This paper reports on a systematic review that was conducted to assess the benefit of artificial neural networks (ANNs) as decision making tools in the field of cancer. The number of clinical trials (CTs) and randomised controlled trials (RCTs) involving the use of ANNs in diagnosis and prognosis increased from 1 to 38 in the last decade. However, out of 396 studies involving the use of ANNs in cancer, only 27 were either CTs or RCTs. Out of these trials, 21 showed an increase in benefit to healthcare provision and 6 did not. None of these studies however showed a decrease in benefit. This paper reviews the clinical fields where neural network methods figure most prominently, the main algorithms featured, methodologies for model selection and the need for rigorous evaluation of results.

GMDH-based feature ranking and selection for improved classification of medical data

Medical applications are often characterized by a large number of disease markers and a relatively small number of data records. We demonstrate that complete feature ranking followed by selection can lead to appreciable reductions in data dimensionality, with significant improvements in the implementation and performance of classifiers for medical diagnosis. We describe a novel approach for ranking all features according to their predictive quality using properties unique to learning algorithms based on the group method of data handling (GMDH). An abductive network training algorithm is repeatedly used to select groups of optimum predictors from the feature set at gradually increasing levels of model complexity specified by the user. Groups selected earlier are better predictors. The process is then repeated to rank features within individual groups. The resulting full feature ranking can be used to determine the optimum feature subset by starting at the top of the list and progressively including more features until the classification error rate on an out-of-sample evaluation set starts to increase due to overfitting. The approach is demonstrated on two medical diagnosis datasets (breast cancer and heart disease) and comparisons are made with other feature ranking and selection methods. Receiver operating characteristics (ROC) analysis is used to compare classifier performance. At default model complexity, dimensionality reduction of 22 and 54% could be achieved for the breast cancer and heart disease data, respectively, leading to improvements in the overall classification performance. For both datasets, considerable dimensionality reduction introduced no significant reduction in the area under the ROC curve. GMDH-based feature selection results have also proved effective with neural network classifiers.

Computer-assisted diagnosis of focal liver lesions on CT images evaluation of the Perceptron algorithm

RATIONALE AND OBJECTIVE

The purpose of the study was to investigate a modified version of a so-called Perceptron algorithm in detecting focal liver lesions on CT scans.

MATERIALS AND METHODS

The modified Perceptron algorithm is based on simulated annealing with a logarithmic cooling schedule and was implemented on a standard workstation. The algorithm was trained with 400 normal and 400 pathologic CT scans of the liver. An additional 100 normal and 100 pathologic scans were then used to test the detection of pathology by the algorithm. The total of 1000 scans used in the study were selected from the portal venous phase of upper abdominal CT examinations performed in patients with normal findings or hypovascularized liver lesions. The pathologic scans contained 1 to 4 focal liver lesions. For the preliminary version of the algorithm used in this study, it was necessary to define regions of interest that were converted to a matrix of 119 x 119.

RESULTS

Training of the algorithm with 400 examples each of normal and abnormal findings took about 75 hours. Subsequently, the testing took several seconds for processing each scan. The diagnostic accuracy in discriminating scans with and without focal liver lesions achieved for the 200 test scans was approximately 99%. The error rate for pathologic and normal scans was comparable to results reported in the literature, which, however, were obtained for much smaller test sets.

CONCLUSION

The modified Perceptron algorithm has an accuracy of close to 99% in detecting pathology on CT scans of the liver showing either normal findings or hypovascularized focal liver lesions.

Phase- and size-adjusted CT cut-off for differentiating neoplastic lesions from normal colon in contrast-enhanced CT colonography

A computed tomography (CT) cut-off for differentiating neoplastic lesions (polyps/carcinoma) from normal colon in contrast-enhanced CT colonography (CTC) relating to the contrast phase and lesion size is determined. CT values of 64 colonic lesions (27 polyps <10 mm, 13 polyps > or =10 mm, 24 carcinomas) were determined by region-of-interest (ROI) measurements in 38 patients who underwent contrast-enhanced CTC. In addition, the height (H) of the colonic lesions was measured in CT. CT values were also measured in the aorta (A), superior mesenteric vein (V) and colonic wall. The contrast phase was defined by xA + (1-x)V using x as a weighting factor for describing the different contrast phases ranging from the pure arterial phase (x=1) over the intermediate phases (x=0.9-0.1) to the pure venous phase (x=0). The CT values of the lesions were correlated with their height (H), the different phases (xA+(1-x)V) and the ratio [xA+(I-x)V]/H. The CT cut-off was linearly adjusted to the imaged contrast phase and height of the lesion by the line y=m[xA+(1-x)V]H+y(0). The slope m was determined by linear regression in the correlation (lesion approximately [xA+(i-x)V]/H) and the Y-intercept y(0) by the minimal shift of the line needed to maximize the accuracy of separating the colonic wall from the lesions. The CT value of the lesions correlated best with the intermediate phase: 0.4A + 0.6V (r=0.8 for polyps > or =10 mm, r=0.6 for carcinomas, r=0.4 for polyps <10 mm). The accuracy in the differentiation between lesions and normal colonic wall increased with the height implemented as divisor, reached 91% and was obtained by the dynamic cut-off described by the formula: cut-off (A,V,H)=1.1 [0.4A+0.6V]/H+69.8. The CT value of colonic polyps or carcinomas can be increased extrinsically by scanning in the phase in which 0.4A + 0.6V reaches its maximum. Differentiating lesions from normal colon based on CT values is possible in contrast-enhanced CTC and improves when the cut-off is adjusted (normalized) to the contrast phase and lesion size.