Search

Purpose: To develop a new algorithm to measure the similarity between the query lung mass and reference lung mass data set for content-based medical image retrieval (CBMIR). Methods: A lung mass data set including 746 mass regions of interest (ROIs) was assembled. Among them, 375 ROIs depicted malignant lesions and 371 depicted benign lesions. Each mass ROI is represented by a vector of 26 texture features. A kernel function was employed to map the original data in input space to a feature space. In this space, a semisupervised distance metric was learned, which used differential scatter discriminant criterion to represent the semantic relevance, and the regularization term to represent the visual similarity. The learned distance metric can measure the similarity of the query mass and reference mass data set. The clustering accuracy is used to configure the parameters. The retrieval accuracy and classification accuracy are used as the performance assessment index. Results: After configuring the parameters, a mean clustering accuracy of 0.87 can be achieved. For retrieval accuracy, our algorithm achieves better performance than other state-of-the-art retrieval algorithms when applying a leave-one-out validation method to the testing data set. For classification accuracy, the area under the ROC curve of our algorithm can be achieved as 0.941 ± 0.006. The running times of 346 query images with the proposed algorithm are 5.399 and 6.0 s, respectively. Conclusions: The study results demonstrated the proposed algorithm outperforms the compared algorithms, when taking the semantic relevant and visual similarity into account in kernel space. The algorithm can be used in a CBMIR system for a query mass to retrieve similarity masses, which can help doctors make better decisions.

Purpose: Mammographic percent density (PD%) is known to be a strong risk factor for breast cancer. Recent studies also suggest that parenchymal texture features, which are more granular descriptors of the parenchymal pattern, can provide additional information about breast cancer risk. To date, most studies have measured mammographic texture within selected regions of interest (ROIs) in the breast, which cannot adequately capture the complexity of the parenchymal pattern throughout the whole breast. To better characterize patterns of the parenchymal tissue, the authors have developed a fully automated software pipeline based on a novel lattice-based strategy to extract a range of parenchymal texture features from the entire breast region. Methods: Digital mammograms from 106 cases with 318 age-matched controls were retrospectively analyzed. The lattice-based approach is based on a regular grid virtually overlaid on each mammographic image. Texture features are computed from the intersection (i.e., lattice) points of the grid lines within the breast, using a local window centered at each lattice point. Using this strategy, a range of statistical (gray-level histogram, co-occurrence, and run-length) and structural (edge-enhancing, local binary pattern, and fractal dimension) features are extracted. To cover the entire breast, the size of the local window for feature extraction is set equal to the lattice grid spacing and optimized experimentally by evaluating different windows sizes. The association between their lattice-based texture features and breast cancer was evaluated using logistic regression with leave-one-out cross validation and further compared to that of breast PD% and commonly used single-ROI texture features extracted from the retroareolar or the central breast region. Classification performance was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC). DeLong’s test was used to compare the different ROCs in terms of AUC performance. Results: The average univariate performance of the lattice-based features is higher when extracted from smaller than larger window sizes. While not every individual texture feature is superior to breast PD% (AUC: 0.59, STD: 0.03), their combination in multivariate analysis has significantly better performance (AUC: 0.85, STD: 0.02, p 0.05). Conclusions: The proposed lattice-based strategy for mammographic texture analysis enables to characterize the parenchymal pattern over the entire breast. As such, these features provide richer information compared to currently used descriptors and may ultimately improve breast cancer risk assessment. Larger studies are warranted to validate these findings and also compare to standard demographic and reproductive risk factors.

Purpose: To investigate the effects of dose level and reconstruction method on density and texture based features computed from CT lung nodules. Methods: This study had two major components. In the first component, a uniform water phantom was scanned at three dose levels and images were reconstructed using four conventional filtered backprojection (FBP) and four iterative reconstruction (IR) methods for a total of 24 different combinations of acquisition and reconstruction conditions. In the second component, raw projection (sinogram) data were obtained for 33 lung nodules from patients scanned as a part of their clinical practice, where low dose acquisitions were simulated by adding noise to sinograms acquired at clinical dose levels (a total of four dose levels) and reconstructed using one FBP kernel and two IR kernels for a total of 12 conditions. For the water phantom, spherical regions of interest (ROIs) were created at multiple locations within the water phantom on one reference image obtained at a reference condition. For the lung nodule cases, the ROI of each nodule was contoured semiautomatically (with manual editing) from images obtained at a reference condition. All ROIs were applied to their corresponding images reconstructed at different conditions. For 17 of the nodule cases, repeat contours were performed to assess repeatability. Histogram (eight features) and gray level co-occurrence matrix (GLCM) based texture features (34 features) were computed for all ROIs. For the lung nodule cases, the reference condition was selected to be 100% of clinical dose with FBP reconstruction using the B45f kernel; feature values calculated from other conditions were compared to this reference condition. A measure was introduced, which the authors refer to as Q, to assess the stability of features across different conditions, which is defined as the ratio of reproducibility (across conditions) to repeatability (across repeat contours) of each feature. Results: The water phantom results demonstrated substantial variability among feature values calculated across conditions, with the exception of histogram mean. Features calculated from lung nodules demonstrated similar results with histogram mean as the most robust feature (Q ≤ 1), having a mean and standard deviation Q of 0.37 and 0.22, respectively. Surprisingly, histogram standard deviation and variance features were also quite robust. Some GLCM features were also quite robust across conditions, namely, diff. variance, sum variance, sum average, variance, and mean. Except for histogram mean, all features have a Q of larger than one in at least one of the 3% dose level conditions. Conclusions: As expected, the histogram mean is the most robust feature in their study. The effects of acquisition and reconstruction conditions on GLCM features vary widely, though trending toward features involving summation of product between intensities and probabilities being more robust, barring a few exceptions. Overall, care should be taken into account for variation in density and texture features if a variety of dose and reconstruction conditions are used for the quantification of lung nodules in CT, otherwise changes in quantification results may be more reflective of changes due to acquisition and reconstruction conditions than in the nodule itself.

Purpose: Precisely and efficiently locating a tumor with less manual intervention in ultrasound-guided high-intensity focused ultrasound (HIFU) therapy is one of the keys to guaranteeing the therapeutic result and improving the efficiency of the treatment. The segmentation of ultrasound images has always been difficult due to the influences of speckle, acoustic shadows, and signal attenuation as well as the variety of tumor appearance. The quality of HIFU guidance images is even poorer than that of conventional diagnostic ultrasound images because the ultrasonic probe used for HIFU guidance usually obtains images without making contact with the patient’s body. Therefore, the segmentation becomes more difficult. To solve the segmentation problem of ultrasound guidance image in the treatment planning procedure for HIFU therapy, a novel region-based segmentation method for uterine fibroids in HIFU guidance images is proposed. Methods: Tumor partitioning in HIFU guidance image without manual intervention is achieved by a region-based split-and-merge framework. A new iterative multiple region growing algorithm is proposed to first split the image into homogenous regions (superpixels). The features extracted within these homogenous regions will be more stable than those extracted within the conventional neighborhood of a pixel. The split regions are then merged by a superpixel-based adaptive spectral clustering algorithm. To ensure the superpixels that belong to the same tumor can be clustered together in the merging process, a particular construction strategy for the similarity matrix is adopted for the spectral clustering, and the similarity matrix is constructed by taking advantage of a combination of specifically selected first-order and second-order texture features computed from the gray levels and the gray level co-occurrence matrixes, respectively. The tumor region is picked out automatically from the background regions by an algorithm according to a priori information about the tumor position, shape, and size. Additionally, an appropriate cluster number for spectral clustering can be determined by the same algorithm, thus the automatic segmentation of the tumor region is achieved. Results: To evaluate the performance of the proposed method, 50 uterine fibroid ultrasound images from different patients receiving HIFU therapy were segmented, and the obtained tumor contours were compared with those delineated by an experienced radiologist. For area-based evaluation results, the mean values of the true positive ratio, the false positive ratio, and the similarity were 94.42%, 4.71%, and 90.21%, respectively, and the corresponding standard deviations were 2.54%, 3.12%, and 3.50%, respectively. For distance-based evaluation results, the mean values of the normalized Hausdorff distance and the normalized mean absolute distance were 4.93% and 0.90%, respectively, and the corresponding standard deviations were 2.22% and 0.34%, respectively. The running time of the segmentation process was 12.9 s for a 318 × 333 (pixels) image. Conclusions: Experiments show that the proposed method can segment the tumor region accurately and efficiently with less manual intervention, which provides for the possibility of automatic segmentation and real-time guidance in HIFU therapy.

Purpose: Increasing evidence suggests radiomics features extracted from computed tomography (CT) images may be useful in prognostic models for patients with nonsmall cell lung cancer (NSCLC). This study was designed to determine whether such features can be reproducibly obtained from cone-beam CT (CBCT) images taken using medical Linac onboard-imaging systems in order to track them through treatment. Methods: Test-retest CBCT images of ten patients previously enrolled in a clinical trial were retrospectively obtained and used to determine the concordance correlation coefficient (CCC) for 68 different texture features. The volume dependence of each feature was also measured using the Spearman rank correlation coefficient. Features with a high reproducibility (CCC > 0.9) that were not due to volume dependence in the patient test-retest set were further examined for their sensitivity to differences in imaging protocol, level of scatter, and amount of motion by using two phantoms. The first phantom was a texture phantom composed of rectangular cartridges to represent different textures. Features were measured from two cartridges, shredded rubber and dense cork, in this study. The texture phantom was scanned with 19 different CBCT imagers to establish the features’ interscanner variability. The effect of scatter on these features was studied by surrounding the same texture phantom with scattering material (rice and solid water). The effect of respiratory motion on these features was studied using a dynamic-motion thoracic phantom and a specially designed tumor texture insert of the shredded rubber material. The differences between scans acquired with different Linacs and protocols, varying amounts of scatter, and with different levels of motion were compared to the mean intrapatient difference from the test-retest image set. Results: Of the original 68 features, 37 had a CCC >0.9 that was not due to volume dependence. When the Linac manufacturer and imaging protocol were kept consistent, 4–13 of these 37 features passed our criteria for reproducibility more than 50% of the time, depending on the manufacturer-protocol combination. Almost all of the features changed substantially when scatter material was added around the phantom. For the dense cork, 23 features passed in the thoracic scans and 11 features passed in the head scans when the differences between one and two layers of scatter were compared. Using the same test for the shredded rubber, five features passed the thoracic scans and eight features passed the head scans. Motion substantially impacted the reproducibility of the features. With 4 mm of motion, 12 features from the entire volume and 14 features from the center slice measurements were reproducible. With 6–8 mm of motion, three features (Laplacian of Gaussian filtered kurtosis, gray-level nonuniformity, and entropy), from the entire volume and seven features (coarseness, high gray-level run emphasis, gray-level nonuniformity, sum-average, information measure correlation, scaled mean, and entropy) from the center-slice measurements were considered reproducible. Conclusions: Some radiomics features are robust to the noise and poor image quality of CBCT images when the imaging protocol is consistent, relative changes in the features are used, and patients are limited to those with less than 1 cm of motion.

Purpose: Breast tissue composition is considered to be associated with breast cancer risk. This study aimed to develop a computer-aided classification (CAC) system to automatically classify echotexture patterns as heterogeneous or homogeneous using automated breast ultrasound (ABUS) images. Methods: A CAC system was proposed that can recognize breast echotexture patterns in ABUS images. For each case, the echotexture pattern was assessed by two expert radiologists and classified as heterogeneous or homogeneous. After neutrosophic image transformation and fuzzy c-mean clusterings, the lower and upper boundaries of the fibroglandular tissues were defined. Then, the number of hypoechoic regions and histogram features were extracted from the fibroglandular tissues, and the support vector machine model with the leave-one-out cross-validation method was utilized as the classifier. The authors’ database included a total of 208 ABUS images of the breasts of 104 females. Results: The accuracies of the proposed system for the classification of heterogeneous and homogeneous echotexture patterns were 93.48% (43/46) and 92.59% (150/162), respectively, with an overall Az (area under the receiver operating characteristic curve) of 0.9786. The agreement between the radiologists and the proposed system was almost perfect, with a kappa value of 0.814. Conclusions: The use of ABUS and the proposed method can provide quantitative information on the echotexture patterns of the breast and can be used to evaluate whether breast echotexture patterns are associated with breast cancer risk in the future.

Purpose: Triple-negative breast cancer (TNBC), an aggressive subtype, is frequently misclassified as fibroadenoma due to benign morphologic features on breast ultrasound (US). This study aims to develop a computer-aided diagnosis (CAD) system based on texture features for distinguishing between TNBC and benign fibroadenomas in US images. Methods: US images of 169 pathology-proven tumors (mean size, 1.65 cm; range, 0.7–3.0 cm) composed of 84 benign fibroadenomas and 85 TNBC tumors are used in this study. After a tumor is segmented out using the level-set method, morphological, conventional texture, and multiresolution gray-scale invariant texture feature sets are computed using a best-fitting ellipse, gray-level co-occurrence matrices, and the ranklet transform, respectively. The linear support vector machine with leave-one-out cross-validation schema is used as a classifier, and the diagnostic performance is assessed with receiver operating characteristic curve analysis. Results: The Az values of the morphology, conventional texture, and multiresolution gray-scale invariant texture feature sets are 0.8470 [95% confidence intervals (CIs), 0.7826–0.8973], 0.8542 (95% CI, 0.7911–0.9030), and 0.9695 (95% CI, 0.9376–0.9865), respectively. The Az of the CAD system based on the combined feature sets is 0.9702 (95% CI, 0.9334–0.9882). Conclusions: The CAD system based on texture features extracted via the ranklet transform may be useful for improving the ability to discriminate between TNBC and benign fibroadenomas.

Purpose: To quantitatively evaluate local carotid arterial statin effects in 3D USimages using multiclassifier image texture analysis tools. Methods: Texture analysis tools were used to evaluate the effect of 80 mg atorvastatin administered daily to patients with carotid stenosis compared to those treated with placebo. Using three-dimensional carotid ultrasoundimages, 270 texture features from seven texture techniques were extracted from manually segmented carotid arteries based on the intima-media boundary [vessel wall (VW)]. Individual texture features were compared to the previously determined changes in VW volume (VWV) using the distance between classes, the Wilcoxon rank sum test, and accuracy of the classifiers. Texture features that resulted in maximal classification accuracy from each texture technique were selected using Pudil’s sequential floating forward selection (SFFS) as a method of ranking each technique. Finally, SFFS-selected texture features from all texture techniques were used in combination with 24 classifier fusion techniques to improve classification accuracy. Results: Using the measurement of change in VWV, the distance between classes (DBC), Wilcoxon rank sum (WRS) p -value, and median accuracy measures (ACC) were 0.3798, 0.076, and 54.50%, respectively. Texture features improved the detection of statin-related changes using DBC to 0.5199, using WRS to 0.002, and ACC to 63.87%, respectively. The texture techniques that most differentiated between atorvastatin and placebo classes were Fourier power spectrum and Laws texture energy measures. The average classification accuracy between atorvastatin and placebo classes was improved from 57.22 ± 12.11 % using VWV to 97.87 ± 3.93 % using specific texture features. Furthermore, the use of specific texture features resulted in the average area under the receiver-operator characteristic curve (AUC) a value of 0.9988 ± 0.0069 compared to 0.617 ± 0.15 using carotid VWV. Conclusions: Based on DBC, WRS, ACC, and AUC texture features derived from 3D carotid ultrasound were observed to be more sensitive in detecting statin-related changes in carotid atherosclerosis than VWV suggesting that texture classifiers can be used to detect changes in carotid atherosclerosis after therapy.

Purpose: This work is to provide a direct, quantitative comparison of image features measured by film and full-field digital mammography (FFDM). The purpose is to investigate whether there is any systematic difference between film and FFDM in terms of quantitative image features and their influence on the performance of a computer-aided diagnosis (CAD) system. Methods: The authors make use of a set of matched film-FFDM image pairs acquired from cadaver breast specimens with simulated microcalcifications consisting of bone and teeth fragments using both a GE digital mammography system and a screen-film system. To quantify the image features, the authors consider a set of 12 textural features of lesion regions and six image features of individual microcalcifications (MCs). The authors first conduct a direct comparison on these quantitative features extracted from film and FFDM images. The authors then study the performance of a CAD classifier for discriminating between MCs and false positives (FPs) when the classifier is trained on images of different types (film, FFDM, or both). Results: For all the features considered, the quantitative results show a high degree of correlation between features extracted from film and FFDM, with the correlation coefficients ranging from 0.7326 to 0.9602 for the different features. Based on a Fisher sign rank test, there was no significant difference observed between the features extracted from film and those from FFDM. For both MC detection and discrimination of FPs from MCs, FFDM had a slight but statistically significant advantage in performance; however, when the classifiers were trained on different types of images (acquired with FFDM or SFM) for discriminating MCs from FPs, there was little difference. Conclusions: The results indicate good agreement between film and FFDM in quantitative image features. While FFDM images provide better detection performance in MCs, FFDM and film images may be interchangeable for the purposes of training CAD algorithms, and a single CAD algorithm may be applied to either type of images.

Purpose: Fatty liver disease (FLD) is an increasing prevalent disease that can be reversed if detected early. Ultrasound is the safest and ubiquitous method for identifying FLD. Since expert sonographers are required to accurately interpret the liverultrasoundimages, lack of the same will result in interobserver variability. For more objective interpretation, high accuracy, and quick second opinions, computer aided diagnostic(CAD) techniques may be exploited. The purpose of this work is to develop one such CAD technique for accurate classification of normal livers and abnormal livers affected by FLD. Methods: In this paper, the authors present a CAD technique (called Symtosis) that uses a novel combination of significant features based on the texture, wavelet transform, and higher order spectra of the liverultrasoundimages in various supervised learning-based classifiers in order to determine parameters that classify normal and FLD-affected abnormal livers. Results: On evaluating the proposed technique on a database of 58 abnormal and 42 normal liverultrasoundimages, the authors were able to achieve a high classification accuracy of 93.3% using the decision tree classifier. Conclusions: This high accuracy added to the completely automated classification procedure makes the authors' proposed technique highly suitable for clinical deployment and usage.

Purpose : Optical coherence tomography(OCT) is a catheter-based imaging method that employs near-infrared light to produce high-resolution cross-sectional intravascular images. The authors propose a segmentation technique for automatic lumen area extraction and stent strut detection in intravascular OCTimages for the purpose of quantitative analysis of neointimal hyperplasia (NIH). Methods : A clinical dataset of frequency-domain OCT scans of the human femoral artery was analyzed. First, a segmentation method based on the Markov random field (MRF) model was employed for lumen area identification. Second, textural and edge information derived from local intensity distribution and continuous wavelet transform (CWT) analysis were integrated to extract the inner luminal contour. Finally, the stent strut positions were detected via the introduction of each strut wavelet response across scales into a feature extraction and classification scheme in order to optimize the strut position detection. Results : The inner lumen contour and the position of stent strut were extracted with very high accuracy. Compared with manual segmentation by an expert vascular physician the automatic segmentation had an average overlap value of 0.937 ± 0.045 for all OCTimages included in the study. The strut detection accuracy had an area under the curve (AUC) value of 0.95, together with sensitivity and specificity average values of 0.91 and 0.96, respectively. Conclusions : A robust automatic segmentation technique integrating textural and edge information for vessel lumen border extraction and strut detection in intravascular OCTimages was designed and presented. The proposed algorithm may be employed for automated quantitative morphological analysis of in-stent neointimal hyperplasia.

Purpose: The purpose of this study is to develop a dissimilarity measure for the classification of trabecular bone (TB) texture in knee radiographs. Problems associated with the traditional extraction and selection of texture features and with the invariance to imaging conditions such as image size, anisotropy, noise, blur, exposure, magnification, and projection angle were addressed. Methods: In the method developed, called a signature dissimilarity measure (SDM), a sum of earth mover's distances calculated for roughness and orientation signatures is used to quantify dissimilarities between textures. Scale-space theory was used to ensure scale and rotation invariance. The effects of image size, anisotropy, noise, and blur on the SDM developed were studied using computer generated fractal texture images. The invariance of the measure to image exposure, magnification, and projection angle was studied using x-ray images of human tibia head. For the studies, Mann-Whitney tests with significance level of 0.01 were used. A comparison study between the performances of a SDM based classification system and other two systems in the classification of Brodatz textures and the detection of knee osteoarthritis (OA) were conducted. The other systems are based on weighted neighbor distance using compound hierarchy of algorithms representing morphology (WND-CHARM) and local binarymore » patterns (LBP). Results: Results obtained indicate that the SDM developed is invariant to image exposure (2.5-30 mA s), magnification (x1.00-x1.35), noise associated with film graininess and quantum mottle ( 64x64 pixels). However, the measure is sensitive to changes in projection angle (>5 deg.), image anisotropy (>30 deg.), and blur generated by a regular film screen. For the classification of Brodatz textures, the SDM based system produced comparable results to the LBP system. For the detection of knee OA, the SDM based system achieved 78.8% classification accuracy and outperformed the WND-CHARM system (64.2%). Conclusions: The SDM is well suited for the classification of TB texture images in knee OA detection and may be useful for the texture classification of medical images in general.« less

Purpose: A system is presented for automated estimation of progression of interstitial lung disease in serial thoracic CT scans. Methods: The system compares corresponding 2D axial sections from baseline and follow-up scans and concludes whether this pair of sections represents regression, progression, or unchanged disease status. The correspondence between serial CT scans is achieved by intrapatient volumetric image registration. The system classification function is trained with two different feature sets. Features in the first set represent the intensity distribution of a difference image between the baseline and follow-up CT sections. Features in the second set represent dissimilarities computed between the baseline and follow-up images filtered with a bank of general purpose texture filters. Results: In an experiment on 74 scan pairs, the system classification accuracies were 76.1% and 79.5% for the two feature sets, respectively, while the accuracies of two observer radiologist were 78.5% and 82%, respectively. The agreements of the system with the reference standard, measured by weighted kappa statistics, were 0.611 and 0.683 for the two feature sets, respectively. Conclusions: The system employing the second feature set showed good agreement with the reference standard, and its accuracy approached that of two radiologists.

The authors propose a spatiotemporal enhancement pattern (STEP) for comprehensive characterization of breast tumors in contrast-enhanced MR images. By viewing serial contrast-enhanced MR images as a single spatiotemporal image, they formulate the STEP as a combination of (1) dynamic enhancement and architectural features of a tumor, and (2) the spatial variations of pixelwise temporal enhancements. Although the latter has been widely used by radiologists for diagnostic purposes, it has rarely been employed for computer-aided diagnosis. This article presents two major contributions. First, the STEP features are introduced to capture temporal enhancement and its spatial variations. This is essentially carried out through the Fourier transformation and pharmacokinetic modeling of various temporal enhancement features, followed by the calculation of moment invariants and Gabor texture features. Second, for effectively extracting the STEP features from tumors, we develop a graph-cut based segmentation algorithm that aims at refining coarse manual segmentations of tumors. The STEP features are assessed through their diagnostic performance for differentiating between benign and malignant tumors using a linear classifier (along with a simple ranking-based feature selection) in a leave-one-out cross-validation setting. The experimental results for the proposed features exhibit superior performance, when compared to the existing approaches, with the area under the ROC curve approaching 0.97.

We have developed a false positive (FP) reduction method based on analysis of bilateral mammograms for computerized mass detection systems. The mass candidates on each view were first detected by our unilateral computer-aided detection (CAD) system. For each detected object, a regional registration technique was used to define a region of interest (ROI) that is “symmetrical” to the object location on the contralateral mammogram. Texture features derived from the spatial gray level dependence matrices and morphological features were extracted from the ROI containing the detected object on a mammogram and its corresponding ROI on the contralateral mammogram. Bilateral features were then generated from corresponding pairs of unilateral features for each object. Two linear discriminant analysis (LDA) classifiers were trained from the unilateral and the bilateral feature spaces, respectively. Finally, the scores from the unilateral LDA classifier and the bilateral LDA asymmetry classifier were fused with a third LDA whose output score was used to distinguish true mass from FPs. A data set of 341 cases of bilateral two-view mammograms was used in this study, of which 276 cases with 552 bilateral pairs contained 110 malignant and 166 benign biopsy-proven masses and 65 cases with 130 bilateral pairs were normal. The mass data set was divided into two subsets for twofold cross-validation training and testing. The normal data set was used for estimation of FP rates. It was found that our bilateral CAD system achieved a case-based sensitivity of 70%, 80%, and 85% at average FP rates of 0.35, 0.75, and 0.95 FPs/image, respectively, on the test data sets with malignant masses. In comparison to the average FP rates for the unilateral CAD system of 0.58, 1.33, and 1.63, respectively, at the corresponding sensitivities, the FP rates were reduced by 40%, 44%, and 42% with the bilateral symmetry information. The improvement was statistically significance (p

In this study, our purpose was to improve the performance of our mass detection system by using a new dual system approach which combines a computer-added detection (CAD) system optimized with "average" masses with another CAD system optimized with "subtle" masses. The two single CAD systems have similar image processing steps, which include prescreening, object segmentation, morphological and texture feature extraction, and false positive (FP) reduction by rule-based and linear discriminant analysis (LDA) classifiers. A feed-forward backpropagation artificial neural network was trained to merge the scores from the LDA classifiers in the two single CAD systems and differentiate true masses from normal tissue. For an unknown test mammogram, the two single CAD systems are applied to the image in parallel to detect suspicious objects. A total of three data sets were used for training and testing the systems. The first data set of 230 current mammograms, referred to as the average mass set, was collected from 115 patients. We also collected 264 mammograms, referred to as the subtle mass set, which were one to two years prior to the current exam from these patients. Both the average and the subtle mass sets were partitioned into two independent data sets in a cross validation training and testing scheme. A third data set containing 65 cases with 260 normal mammograms was used to estimate the FP marker rates during testing. When the single CAD system trained on the average mass set was applied to the test set with average masses, the FP marker rates were 2.2, 1.8, and 1.5 per image at the case-based sensitivities of 90%, 85%, and 80%, respectively. With the dual CAD system, the FP marker rates were reduced to 1.2, 0.9, and 0.7 per image, respectively, at the same case-based sensitivities. Statistically significant (p < 0.05) improvements on the free response receiver operating characteristic curves were observed when the dual system and the single system were compared using the test sets with either average masses or subtle masses.

In this paper, we present a computer-aided detection (CAD) method to extract and use internal features to reduce false positive (FP) rate generated by surface-based measures on the inner colon wall in computed tomographic (CT) colonography. Firstly, a new shape description global curvature, which can provide an overall shape description of the colon wall, is introduced to improve the detection of suspicious patches on the colon wall whose geometrical features are similar to that of the colonic polyps. By a ray-driven edge finder, the volume of each detected patch is extracted as a fitted ellipsoid model. Within the ellipsoid model, CT image density distribution is analyzed. Three types of (geometrical, morphological and textural) internal features are extracted and applied to eliminate the FPs from the detected patches. The presented CAD method was tested by a total of 153 patient datasets in which 45 patients were found with 61 polyps of sizes 4–30 mm by optical colonoscopy. For a 100% detection sensitivity (on polyps), the presented CAD method had an average FPs of 2.68 per patient dataset and eliminated 93.1% of FPs generated by the surface-based measures. The presented CAD method was also evaluated by different polyp sizes. For polyp sizes of 10–30 mm, the method achieved mean number of FPs per dataset of 2.0 with 100% sensitivity. For polyp sizes of 4–10 mm, the method achieved 3.44 FP per dataset with 100% sensitivity.

Tissue structures, represented by textures in radiographs, can be quantified using texture analysis methods. Different texture analysis methods have been used to discriminate between different aspects of various diseases in primarily x rays of chest, bone, and breasts. However, most of these methods have not specifically been developed for use on radiographs. Certain characteristics of the radiographic process, e.g., noise and blurring, influence the visible texture. In order for a texture analysis method to be able to discriminate between different underlying textures, it should not be too sensitive for such processes as image noise and blur. In this study, we investigated the sensitivity of four different texture analysis methods for image noise and blur. First, a baseline measurement was performed of the discriminative performance of the Spatial Gray-Level Dependence method, the Fourier Power Spectrum, the Fractal Dimension, and the Morphological Gradient Method on images, which were not affected by radiographic noise and blur. Two types of images were used: fractal and Brodatz. Whereas the Brodatz images represent very different textures, the differences between the fractal images are more gradual. We assume that the behavior of the different texture analysis methods on the fractal images is representative for their performance on radiologic textures. On these types of images we simulated the effect of four different noise levels and the effect of two different modulation transfer functions, corresponding with different screenfilm combinations. The influence on the discriminative performance of the four texture analysis methods was evaluated. The influence of noise on the discriminative performance is, as expected, dependent on the image type used; the discrimination of more gradually different images, such as the fractal images, is already lowered for relatively low noise levels. In contrast, when the images are more different, only high noise levels decrease the discriminative performance. The discriminative power of the Morphological Gradient Method is least affected by image blur. We conclude that the discriminative performance of the Morphological Gradient Method is superior to that of other methods in circumstances which mimic the conditions prevailing in radiographs.

A computer-aided diagnosis (CAD) system is presented to automatically distinguish normal from abnormal tissue in high-resolution CT chest scans acquired during daily clinical practice. From high-resolution computed tomography scans of 116 patients, 657 regions of interest are extracted that are to be classified as displaying either normal or abnormal lung tissue. A principled texture analysis approach is used, extracting features to describe local image structure by means of a multi-scale filter bank. The use of various classifiers and feature subsets is compared and results are evaluated with ROC analysis. Performance of the system is shown to approach that of two expert radiologists in diagnosing the local regions of interest, with an area under the ROC curve of 0.862 for the CAD scheme versus 0.877 and 0.893 for the radiologists.

The purpose of this study was to investigate an information theoretic approach to feature selection for computer-aided diagnosis (CAD). The approach is based on the mutual information (MI) concept. MI measures the general dependence of random variables without making any assumptions about the nature of their underlying relationships. Consequently, MI can potentially offer some advantages over feature selection techniques that focus only on the linear relationships of variables. This study was based on a database of statistical texture features extracted from perfusion lung scans. The ultimate goal was to select the optimal subset of features for the computer-aided diagnosis of acute pulmonary embolism (PE). Initially, the study addressed issues regarding the approximation of MI in a limited dataset as it is often the case in CAD applications. The MI selected features were compared to those features selected using stepwise linear discriminant analysis and genetic algorithms for the same PE database. Linear and nonlinear decision models were implemented to merge the selected features into a final diagnosis. Results showed that the MI is an effective feature selection criterion for nonlinear CAD models overcoming some of the well-known limitations and computational complexities of other popular feature selection techniques in the field.

A new classification scheme was developed to classify mammographic masses as malignant and benign by using interval change information. The masses on both the current and the prior mammograms were automatically segmented using an active contour method. From each mass, 20 run length statistics (RLS) texture features, 3 speculation features, and 12 morphological features were extracted. Additionally, 20 difference RLS features were obtained by subtracting the prior RLS features from the corresponding current RLS features. The feature space consisted of the current RLS features, the difference RLS features, the current and prior speculation features, and the current and prior mass sizes. Stepwise feature selection and linear discriminant analysis classification were used to select and merge the most useful features. A leave-one-case-out resampling scheme was used to train and test the classifier using 140 temporal image pairs (85 malignant, 55 benign) obtained from 57 biopsy-proven masses (33 malignant, 24 benign) in 56 patients. An average of 10 features were selected from the 56 training subsets: 4 difference RLS features, 4 RLS features, and 1 speculation feature from the current image, and 1 speculation feature from the prior, were most often chosen. The classifier achieved an average training A z of 0.92 and a test A z of 0.88. For comparison, a classifier was trained and tested using features extracted from the 120 current single images. This classifier achieved an average training A z of 0.90 and a test A z of 0.82. The information on the prior image significantly (p=0.015) improved the accuracy for classification of the masses.

Purpose: Selecting optimal features from a large image feature pool remains a major challenge in developing computer-aided detection (CAD) schemes of medical images. The objective of this study is to investigate a new approach to significantly improve efficacy of image feature selection and classifier optimization in developing a CAD scheme of mammographic masses. Methods: An image dataset including 1600 regions of interest (ROIs) in which 800 are positive (depicting malignant masses) and 800 are negative (depicting CAD-generated false positive regions) was used in this study. After segmentation of each suspicious lesion by a multilayer topographic region growth algorithm, 271 features were computed in different feature categories including shape, texture, contrast, isodensity, spiculation, local topological features, as well as the features related to the presence and location of fat and calcifications. Besides computing features from the original images, the authors also computed new texture features from the dilated lesion segments. In order to select optimal features from this initial feature pool and build a highly performing classifier, the authors examined and compared four feature selection methods to optimize an artificial neural network (ANN) based classifier, namely: (1) Phased Searching with NEAT in a Time-Scaled Framework, (2) A sequential floating forward selection (SFFS) method, (3) A genetic algorithm (GA), and (4) A sequential forward selection (SFS) method. Performances of the four approaches were assessed using a tenfold cross validation method. Results: Among these four methods, SFFS has highest efficacy, which takes 3%–5% of computational time as compared to GA approach, and yields the highest performance level with the area under a receiver operating characteristic curve (AUC) = 0.864 ± 0.034. The results also demonstrated that except using GA, including the new texture features computed from the dilated mass segments improved the AUC results of the ANNs optimized using other three feature selection methods. In addition, among 271 features, the shape, local morphological features, fat and calcification based features were the most frequently selected features to build ANNs. Conclusions: Although conventional GA is a powerful tool in optimizing classifiers used in CAD schemes of medical images, it is very computationally intensive. This study demonstrated that using a new SFFS based approach enabled to significantly improve efficacy of image feature selection for developing CAD schemes.

We are developing new computer vision techniques for characterization of breast masses on mammograms. We had previously developed a characterization method based on texture features. The goal of the present work was to improve our characterization method by making use of morphological features. Toward this goal, we have developed a fully automated, three-stage segmentation method that includes clustering, active contour, and spiculation detection stages. After segmentation, morphological features describing the shape of the mass were extracted. Texture features were also extracted from a band of pixels surrounding the mass. Stepwise feature selection and linear discriminant analysis were employed in the morphological, texture, and combined feature spaces for classifier design. The classification accuracy was evaluated using the area Az under the receiver operating characteristic curve. A data set containing 249 films from 102 patients was used. When the leave-one-case-out method was applied to partition the data set into trainers and testers, the average test Az for the task of classifying the mass on a single mammographic view was 0.83 +/- 0.02, 0.84 +/- 0.02, and 0.87 +/- 0.02 in the morphological, texture, and combined feature spaces, respectively. The improvement obtained by supplementing texture features with morphological features in classification was statistically significant (p = 0.04). For classifying a mass as malignant or benign, we combined the leave-one-case-out discriminant scores from different views of a mass to obtain a summary score. In this task, the test Az value using the combined feature space was 0.91 +/- 0.02. Our results indicate that combining texture features with morphological features extracted from automatically segmented mass boundaries will be an effective approach for computer-aided characterization of mammographic masses.

Purpose: A novel approach to a computer aided diagnosis system for the Parkinson's disease is proposed. This tool is intended as a supporting tool for physicians, based on fully automated methods that lead to the classification of{sup 123}I-ioflupane SPECT images. Methods: {sup 123}I-ioflupane images from three different databases are used to train the system. The images are intensity and spatially normalized, then subimages are extracted and a 3D gray-level co-occurrence matrix is computed over these subimages, allowing the characterization of the texture using Haralick texture features. Finally, different discrimination estimation methods are used to select a feature vector that can be used to train and test the classifier. Results: Using the leave-one-out cross-validation technique over these three databases, the system achieves results up to a 97.4% of accuracy, and 99.1% of sensitivity, with positive likelihood ratios over 27. Conclusions: The system presents a robust feature extraction method that helps physicians in the diagnosis task by providing objective, operator-independent textural information about{sup 123}I-ioflupane images, commonly used in the diagnosis of the Parkinson's disease. Textural features computation has been optimized by using a subimage selection algorithm, and the discrimination estimation methods used here makes the system feature-independent, allowing us to extend itmore » to other databases and diseases.« less

Purpose: The treatment of ventral hernias (VH) has been a challenging problem for medical care. Repair of these hernias is fraught with failure; recurrence rates ranging from 24% to 43% have been reported, even with the use of biocompatible mesh. Currently, computed tomography (CT) is used to guide intervention through expert, but qualitative, clinical judgments, notably, quantitative metrics based on image-processing are not used. The authors propose that image segmentation methods to capture the three-dimensional structure of the abdominal wall and its abnormalities will provide a foundation on which to measure geometric properties of hernias and surrounding tissues and, therefore, to optimize intervention. Methods: In this study with 20 clinically acquired CT scans on postoperative patients, the authors demonstrated a novel approach to geometric classification of the abdominal. The authors’ approach uses a texture analysis based on Gabor filters to extract feature vectors and follows a fuzzy c-means clustering method to estimate voxelwise probability memberships for eight clusters. The memberships estimated from the texture analysis are helpful to identify anatomical structures with inhomogeneous intensities. The membership was used to guide the level set evolution, as well as to derive an initial start close to the abdominal wall. Results: Segmentation results on abdominal walls were both quantitatively and qualitatively validated with surface errors based on manually labeled ground truth. Using texture, mean surface errors for the outer surface of the abdominal wall were less than 2 mm, with 91% of the outer surface less than 5 mm away from the manual tracings; errors were significantly greater (2–5 mm) for methods that did not use the texture. Conclusions: The authors’ approach establishes a baseline for characterizing the abdominal wall for improving VH care. Inherent texture patterns in CT scans are helpful to the tissue classification, and texture analysis can improve the level set segmentation around the abdominal region.

We are developing computerized feature extraction and classification methods to analyze malignant and benign microcalcifications on digitized mammograms. Morphological features that described the size, contrast, and shape of microcalcifications and their variations within a cluster were designed to characterize microcalcifications segmented from the mammographic background. Texture features were derived from the spatial gray-level dependence (SGLD) matrices constructed at multiple distances and directions from tissue regions containing microcalcifications. A genetic algorithm (GA) based feature selection technique was used to select the best feature subset from the multi-dimensional feature spaces. The GA-based method was compared to the commonly used feature selection method based on the stepwise linear discriminant analysis (LDA) procedure. Linear discriminant classifiers using the selected features as input predictor variables were formulated for the classification task. The discriminant scores output from the classifiers were analyzed by receiver operating characteristic (ROC) methodology and the classification accuracy was quantified by the area, A z , under the ROC curve. We analyzed a data set of 145 mammographic microcalcification clusters in this study. It was found that the feature subsets selected by the GA-based method are comparable to or slightly better than those selected by the stepwise LDA method. The texture features (A z =0.84) were more effective than morphological features (A z =0.79) in distinguishing malignant and benign microcalcifications. The highest classification accuracy (A z =0.89) was obtained in the combined texture and morphological feature space. The improvement was statistically significant in comparison to classification in either the morphological (p=0.002) or the texture (p=0.04) feature space alone. The classifier using the best feature subset from the combined feature space and an appropriate decision threshold could correctly identify 35% of the benign clusters without missing a malignant cluster. When the average discriminant score from all views of the same cluster was used for classification, the A z value increased to 0.93 and the classifier could identify 50% of the benign clusters at 100% sensitivity for malignancy. Alternatively, if the minimum discriminant score from all views of the same cluster was used, the A z value would be 0.90 and a specificity of 32% would be obtained at 100% sensitivity. The results of this study indicate the potential of using combined morphological and texture features for computer-aided classification of microcalcifications.

A new rubber band straightening transform (RBST) is introduced for characterization of mammographic masses as malignant or benign. The RBST transforms a band of pixels surrounding a segmented mass onto the Cartesian plane (the RBST image). The border of a mammographic mass appears approximately as a horizontal line, and possible spiculations resemble vertical lines in the RBST image. In this study, the effectiveness of a set of directional texture features extracted from the RBST images was compared to the effectiveness of the same features extracted from the images before the RBST. A database of 168 mammograms containing biopsy-proven malignant and benign breast masses was digitized at a pixel size of 100 μm×100 μm . Regions of interest (ROIs) containing the biopsied mass were extracted from each mammogram by an experienced radiologist. A clustering algorithm was employed for automated segmentation of each ROI into a mass object and background tissue. Texture features extracted from spatial gray-level dependence matrices and run-length statistics matrices were evaluated for three different regions and representations: (i) the entire ROI; (ii) a band of pixels surrounding the segmented mass object in the ROI; and (iii) the RBST image. Linear discriminant analysis was used for classification, and receiver operating characteristic (ROC) analysis was used to evaluate the classification accuracy. Using the ROC curves as the performance measure, features extracted from the RBST images were found to be significantly more effective than those extracted from the original images. Features extracted from the RBST images yielded an area (A z ) of 0.94 under the ROC curve for classification of mammographic masses as malignant and benign.

Purpose: The aim of this study was to compare three demons registration-based methods to identify spatially matched regions in serial computed tomography (CT) scans for use in texture analysis. Methods: Two thoracic CT scans containing no lung abnormalities and acquired during serial examinations separated by at least one week were retrospectively collected from 27 patients. Over 1000 regions of interest (ROIs) were randomly placed in the lungs of each baseline scan. Anatomically matched ROIs in the corresponding follow-up scan were placed by mapping the baseline scan ROI center pixel to (1) the original follow-up scan, (2) the follow-up scan resampled to match the baseline scan voxel size, and (3) the follow-up scan aligned to the baseline scan through affine registration. Mappings used the vector field obtained through demons deformable registration of each follow-up scan variant to the baseline scan. 140 texture features distributed among five feature classes were calculated in all ROIs. Feature value differences between paired ROIs were evaluated using Bland-Altman 95% limits of agreement. For each feature, (1) the mean feature value change and (2) the difference between the upper and lower limits of agreement were normalized to the mean feature value to obtain, respectively, the normalized bias and normalized range of agreement (nRoA). Nonparametric tests were used to evaluate differences in normalized bias and nRoA across the three methods. Results: Because patient CT scans contained no pathology, minimal changes in feature values were expected (i.e., low nRoA and normalized bias). Seventy-five features with very large feature value variability (nRoA ≥ 100%) were excluded from further analysis. Across the remaining 65 features, significant differences in normalized bias were observed among the three methods. The lowest normalized bias (median: 0.06%) was achieved when feature values were calculated on original follow-up scans. The affine registration method achieved the lowest nRoA, though nRoA was not significantly increased using original follow-up scans. Features with low nRoA values also had low normalized bias, though the converse was not necessarily true. Using nRoA as a metric, a set of 20 features having both low nRoA and normalized bias were identified. Conclusions: Three methods to facilitate texture analysis of serial CT scans using demons registration for ROI placement were evaluated. The bias in feature value change between matched ROIs was minimized when feature values were calculated on original baseline and follow-up scans. A set of features that had both low bias and variability (nRoA) in feature value change using this method were identified. This texture analysis approach could facilitate future measurement of pathologic changes between CT scans without necessitating calculation of feature values on deformed scans.

I Enhancement Fundametal Enhancement Techniques Adaptive Image Filtering Enhancement by Multiscale Nonlinear Operators Medical Image Enhancement with Hybrid Filters. II Segmentation Overview and Fundamentals of Medical Image Segmentation Image Segmentation by Fuzzy Clustering: Methods and Issues Segmentation with Neural Networks Deformable Models Shape Constraints in Deformable Models Gradient Vector Flow Deformable Models Fully Automated Hybrid Segmentation of the Brain Volumetric Segmentation Partial Volume Segmentation with Voxel Histograms. III Quantification Two-dimensional Shape and Texture Quantification Texture Analysis in Three Dimensions as a Cue to Medical Diagnosis Computational Neuroanatomy Using Shape Transformation Morphometry of Arterial Trees and Microvascular Networks Image-Based Computational Biomechanics of the Musculoskeletal System Three-Dimensional Bone Angle Quantification Database Selection and Feature Extraction for Neural Networks Quantitative Image Analysis for Estimation of Breast Cancer Risk Classification of Breast Lesions in Mammograms Quantitative Analysis of Cardiac Function Image Processing and Analysis in Tagged Cardiac MRI Image Interpolation and Resampling. IV Registration The Physical Basis of Spatial Distortions in Magnetic Resonance Images Physical and Biological Bases of Spatial Distortiobns in Positron Emission Tomography Images Biological Underpinnings of Anatomic Consistency and Variability in the Human Brain Spatial Transformation Models Validation of Registration Accuracy Landmark-based Registration using Features Identified Through Differential Geometry Registration of Contours using Chamfer Matching Within-Modality Registration using Intensity-Based Cost Functions Across-Modality Registration using Intensity-Based Cost Functions Talairach Space as a Tool for Intersubject Standardization in the Brain Warping Strategies for Intersubject Registration Optimizing the Resampling of Registered Images Clinical Applications of Image Registration Registration for Image-Guided Surgery Image Registration and the Construction of Multidimensional Brain Atlases. V Visualization Visualization Pathways in Biomedicine Three-Dimensional Visualization in Medicine and Biology Volume Visualization in Medicine Fast Isosurface Extraction Methods for Large Image Data Sets Morphometric Methods for Virtual Endoscopy. VI Compression Storage and Communication Fundamentals and Standards of Compression and Communication Medical Image Archive and Retrieval Image Standardization in PACS Quality Evaluation for Compressed Medical Images: Fundamentals Quality Evaluation for Compressed Medical Images: Diagnostic Accuracy Quality Evaluation for Compressed Medical Images: Statistical Issues Three-Dimensional Image Compression with Wavelet Transforms Medical Image Processing and Analysis Software. Index