Analysis of spiculation in the computerized classification of mammographic masses

Mammography radiomics features at diagnosis and progression-free survival among patients with breast cancer

BACKGROUND

The associations between mammographic radiomics and breast cancer clinical endpoints are unclear. We aimed to identify mammographic radiomics features associated with breast cancer prognosis.

METHODS

Nested from a large breast cancer cohort in our institution, we conducted an extreme case-control study consisting of 207 cases with any invasive disease-free survival (iDFS) endpoint <5 years and 207 molecular subtype-matched controls with >5-year iDFS. A total of 632 radiomics features in craniocaudal (CC) and mediolateral oblique (MLO) views were extracted from pre-treatment mammography. Logistic regression was used to identify iDFS-associated features with multiple testing corrections (Benjamini-Hochberg method). In a subsample with RNA-seq data (n = 96), gene set enrichment analysis was employed to identify pathways associated with lead features.

RESULTS

We identified 15 iDFS-associated features from CC-view yet none from MLO-view. S(1,-1)SumAverg and WavEnLL_s-6 were the lead ones and associated with favourable (OR 0.64, 95% CI 0.42-0.87, P = 0.01) and poor iDFS (OR 1.53, 95% CI 1.31-1.76, P = 0.01), respectively. Both features were associated with eight pathways (primarily involving cell cycle regulation) in tumour but not adjacent normal tissues.

CONCLUSION

Our findings suggest mammographic radiomics features are associated with breast cancer iDFS, potentially through pathways involving cell cycle regulation.

Computer-aided diagnosis of masses in breast computed tomography imaging: deep learning model with combined handcrafted and convolutional radiomic features

Purpose: A computer-aided diagnosis (CADx) system for breast masses is proposed, which incorporates both handcrafted and convolutional radiomic features embedded into a single deep learning model. Approach: The model combines handcrafted and convolutional radiomic signatures into a multi-view architecture, which retrieves three-dimensional (3D) image information by simultaneously processing multiple two-dimensional mass patches extracted along different planes through the 3D mass volume. Each patch is processed by a stream composed of two concatenated parallel branches: a multi-layer perceptron fed with automatically extracted handcrafted radiomic features, and a convolutional neural network, for which discriminant features are learned from the input patches. All streams are then concatenated together into a final architecture, where all network weights are shared and the learning occurs simultaneously for each stream and branch. The CADx system was developed and tested for diagnosis of breast masses ( N = 284 ) using image datasets acquired with independent dedicated breast computed tomography systems from two different institutions. The diagnostic classification performance of the CADx system was compared against other machine and deep learning architectures adopting handcrafted and convolutional approaches, and three board-certified breast radiologists. Results: On a test set of 82 masses (45 benign, 37 malignant), the proposed CADx system performed better than all other model architectures evaluated, with an increase in the area under the receiver operating characteristics curve (AUC) of 0.05 ± 0.02 , and achieving a final AUC of 0.947, outperforming the three radiologists ( AUC = 0.814 - 0.902 ). Conclusions: In conclusion, the system demonstrated its potential usefulness in breast cancer diagnosis by improving mass malignancy assessment.

Comparison of segmentation-free and segmentation-dependent computer-aided diagnosis of breast masses on a public mammography dataset

PURPOSE

To compare machine learning methods for classifying mass lesions on mammography images that use predefined image features computed over lesion segmentations to those that leverage segmentation-free representation learning on a standard, public evaluation dataset.

METHODS

We apply several classification algorithms to the public Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM), in which each image contains a mass lesion. Segmentation-free representation learning techniques for classifying lesions as benign or malignant include both a Bag-of-Visual-Words (BoVW) method and a Convolutional Neural Network (CNN). We compare classification performance of these techniques to that obtained using two different segmentation-dependent approaches from the literature that rely on specific combinations of end classifiers (e.g. linear discriminant analysis, neural networks) and predefined features computed over the lesion segmentation (e.g. spiculation measure, morphological characteristics, intensity metrics).

RESULTS

We report area under the receiver operating characteristic curve (AZ) values for malignancy classification on CBIS-DDSM for each technique. We find average AZ values of 0.73 for a segmentation-free BoVW method, 0.86 for a segmentation-free CNN method, 0.75 for a segmentation-dependent linear discriminant analysis of Rubber-Band Straightening Transform features, and 0.58 for a hybrid rule-based neural network classification using a small number of hand-designed features.

CONCLUSIONS

We find that malignancy classification performance on the CBIS-DDSM dataset using segmentation-free BoVW features is comparable to that of the best segmentation-dependent methods we study, but also observe that a common segmentation-free CNN model substantially and significantly outperforms each of these (p < 0.05). These results reinforce recent findings suggesting that representation learning techniques such as BoVW and CNNs are advantageous for mammogram analysis because they do not require lesion segmentation, the quality and specific characteristics of which can vary substantially across datasets. We further observe that segmentation-dependent methods achieve performance levels on CBIS-DDSM inferior to those achieved on the original evaluation datasets reported in the literature. Each of these findings reinforces the need for standardization of datasets, segmentation techniques, and model implementations in performance assessments of automated classifiers for medical imaging.

Multi-marker quantitative radiomics for mass characterization in dedicated breast CT imaging

Purpose

To develop and evaluate the diagnostic performance of an algorithm for multi‐marker radiomic‐based classification of breast masses in dedicated breast computed tomography (bCT) images.

Methods

Over 1000 radiomic descriptors aimed at quantifying mass and border heterogeneity, morphology, and margin sharpness were developed and implemented. These included well‐established texture and shape feature descriptors, which were supplemented with additional approaches for contour irregularity quantification, spicule and lobe detection, characterization of degree of infiltration, and differences in peritumoral compartments. All descriptors were extracted from a training set of 202 bCT masses (133 benign and 69 malignant), and their individual diagnostic performance was investigated in terms of area under the receiver operating characteristics (ROC) curve (AUC) of single‐feature‐based linear discriminant analysis (LDA) classifiers. Subsequently, the most relevant descriptors were selected through a multiple‐step feature selection process (including stability analysis, statistical significance, evaluation of feature interaction, and dimensionality reduction), and used to develop a final LDA radiomic model for classification of benign and malignant masses, which was then tested on an independent test set of 82 cases (45 benign and 37 malignant).

Results

The majority of the individual radiomic descriptors showed, on the training set, an AUC value deriving from a linear decision boundary higher than 0.65, with the lower limit of the associated 95% confidence interval (C.I.) not overlapping with random chance (AUC = 0.5). The final LDA radiomic model resulted in a test set AUC of 0.90 (95% C.I. 0.80–0.96).

Conclusions

The proposed multi‐marker radiomic approach achieved high diagnostic accuracy in bCT mass classification, using a radiomic signature based on different feature types. While future studies with larger datasets are needed to further validate these results, quantitative radiomics applied to bCT shows potential to improve the breast cancer diagnosis pipeline.

Digital Mammography in Breast Cancer: Additive Value of Radiomics of Breast Parenchyma

Background Previous studies have suggested that breast parenchymal texture features may reflect the biologic risk factors associated with breast cancer development. Therefore, combining the characteristics of normal parenchyma from the contralateral breast with radiomic features of breast tumors may improve the accuracy of digital mammography in the diagnosis of breast cancer. Purpose To determine whether the addition of radiomic analysis of contralateral breast parenchyma to the characterization of breast lesions with digital mammography improves lesion classification over that with radiomic tumor features alone. Materials and Methods This HIPAA-compliant, retrospective study included 182 patients (age range, 25-90 years; mean age, 55.9 years ± 14.9) who underwent mammography between June 2002 and July 2009. There were 106 malignant and 76 benign lesions. Automatic lesion segmentation and radiomic analysis were performed for each breast lesion. Radiomic texture analysis was applied in the normal regions of interest in the contralateral breast parenchyma to assess the mammographic parenchymal patterns. The classification performance of both individual features and the output from a Bayesian artificial neural network classifier was evaluated with the leave-one-patient-out method by using the area under the receiver operating characteristic curve (AUC) as the figure of merit in the task of differentiating between malignant and benign lesions. Results The performance of the combined lesion and parenchyma classifier in the differentiation between malignant and benign mammographic lesions was better than that with the lesion features alone (AUC = 0.84 ± 0.03 vs 0.79 ± 0.03, respectively; P = .047). Overall, six radiomic features-spiculation, margin sharpness, size, circularity from the tumor feature set, and skewness and power law beta from the parenchymal feature set-were selected more than 50% of the time during the feature selection process on the combined feature set. Conclusion Combining quantitative radiomic data from tumors with contralateral parenchyma characterizations may improve diagnostic accuracy for breast cancer. © RSNA, 2019 Online supplemental material is available for this article. See also the editorial by Shaffer in this issue.

Overview on subjective similarity of images for content-based medical image retrieval

Computer-aided diagnosis systems for assisting the classification of various diseases have the potential to improve radiologists' diagnostic accuracy and efficiency, as reported in several studies. Conventional systems generally provide the probabilities of disease types in terms of numerical values, a method that may not be efficient for radiologists who are trained by reading a large number of images. Presentation of reference images similar to those of a new case being diagnosed can supplement the probability outputs based on computerized analysis as an intuitive guide, and it can assist radiologists in their diagnosis, reporting, and treatment planning. Many studies on content-based medical image retrievals have been reported on. For retrieval of perceptually similar and diagnostically relevant images, incorporation of perceptual similarity data by radiologists has been suggested. In this paper, studies on image retrieval methods are reviewed with a special focus on quantification, utilization, and the evaluation of subjective similarities between pairs of images.

A deep feature fusion methodology for breast cancer diagnosis demonstrated on three imaging modality datasets

BACKGROUND

Deep learning methods for radiomics/computer-aided diagnosis (CADx) are often prohibited by small datasets, long computation time, and the need for extensive image preprocessing.

AIMS

We aim to develop a breast CADx methodology that addresses the aforementioned issues by exploiting the efficiency of pre-trained convolutional neural networks (CNNs) and using pre-existing handcrafted CADx features.

MATERIALS & METHODS

We present a methodology that extracts and pools low- to mid-level features using a pretrained CNN and fuses them with handcrafted radiomic features computed using conventional CADx methods. Our methodology is tested on three different clinical imaging modalities (dynamic contrast enhanced-MRI [690 cases], full-field digital mammography [245 cases], and ultrasound [1125 cases]).

RESULTS

From ROC analysis, our fusion-based method demonstrates, on all three imaging modalities, statistically significant improvements in terms of AUC as compared to previous breast cancer CADx methods in the task of distinguishing between malignant and benign lesions. (DCE-MRI [AUC = 0.89 (se = 0.01)], FFDM [AUC = 0.86 (se = 0.01)], and ultrasound [AUC = 0.90 (se = 0.01)]).

DISCUSSION/CONCLUSION

We proposed a novel breast CADx methodology that can be used to more effectively characterize breast lesions in comparison to existing methods. Furthermore, our proposed methodology is computationally efficient and circumvents the need for image preprocessing.

Globally supported radial basis function based collocation method for evolution of level set in mass segmentation using mammograms

Computer-aided detection systems play an important role for the detection of breast abnormalities using mammograms. Global segmentation of mass in mammograms is a complex process due to low contrast mammogram images, irregular shape of mass, speculated margins, and the presence of intensity variations of pixels. This work presents a new approach for mass detection in mammograms, which is based on the variational level set function. Mesh-free based radial basis function (RBF) collocation approach is employed for the evolution of level set function for segmentation of breast as well as suspicious mass region. The mesh-based finite difference method (FDM) is used in literature for evolution of level set function. This work also showcases a comparative study of mesh-free and mesh-based approaches. An anisotropic diffusion filter is employed for enhancement of mammograms. The performance of mass segmentation is analyzed by computing statistical measures. Binarized statistical image features (BSIF) and variants of local binary pattern (LBP) are computed from the segmented suspicious mass regions. These features are given as input to the supervised support vector machine (SVM) classifier to classify suspicious mass region as mass (abnormal) or non-mass (normal) region. Validation of the proposed algorithm is done on sample mammograms taken from publicly available Mini-mammographic image analysis society (MIAS) and Digital Database for Screening Mammography (DDSM) datasets. Combined BSIF features perform better as compared to LBP variants with the performance reported as 97.12% sensitivity, 92.43% specificity, and 98% AUC with 5.12 FP/I on DDSM dataset; and 95.12% sensitivity, 92.41% specificity, and 95% AUC with 4.01FP/I on MIAS dataset.

Development of intelligent systems based on Bayesian regularization network and neuro-fuzzy models for mass detection in mammograms: A comparative analysis

Female breast cancer is the second most common cancer in the world. Several efforts in artificial intelligence have been made to help improving the diagnostic accuracy at earlier stages. However, the identification of breast abnormalities, like masses, on mammographic images is not a trivial task, especially for dense breasts. In this paper we describe our novel mass detection process that includes three successive steps of enhancement, characterization and classification. The proposed enhancement system is based mainly on the analysis of the breast texture. First of all, a filtering step with morphological operators and soft thresholding is achieved. Then, we remove from the filtered breast region, all the details that may interfere with the eventual masses, including pectoral muscle and galactophorous tree. The pixels belonging to this tree will be interpolated and replaced by the average of the neighborhood. In the characterization process, measurement of the Gaussian density in the wavelet domain allows the segmentation of the masses. Finally, a comparative classification mechanism based on the Bayesian regularization back-propagation networks and ANFIS techniques is proposed. The tests were conducted on the MIAS database. The results showed the robustness of the proposed enhancement method.

A computational model to generate simulated three-dimensional breast masses

PURPOSE

To develop algorithms for creating realistic three-dimensional (3D) simulated breast masses and embedding them within actual clinical mammograms. The proposed techniques yield high-resolution simulated breast masses having randomized shapes, with user-defined mass type, size, location, and shape characteristics.

METHODS

The authors describe a method of producing 3D digital simulations of breast masses and a technique for embedding these simulated masses within actual digitized mammograms. Simulated 3D breast masses were generated by using a modified stochastic Gaussian random sphere model to generate a central tumor mass, and an iterative fractal branching algorithm to add complex spicule structures. The simulated masses were embedded within actual digitized mammograms. The authors evaluated the realism of the resulting hybrid phantoms by generating corresponding left- and right-breast image pairs, consisting of one breast image containing a real mass, and the opposite breast image of the same patient containing a similar simulated mass. The authors then used computer-aided diagnosis (CAD) methods and expert radiologist readers to determine whether significant differences can be observed between the real and hybrid images.

RESULTS

The authors found no statistically significant difference between the CAD features obtained from the real and simulated images of masses with either spiculated or nonspiculated margins. Likewise, the authors found that expert human readers performed very poorly in discriminating their hybrid images from real mammograms.

CONCLUSIONS

The authors' proposed method permits the realistic simulation of 3D breast masses having user-defined characteristics, enabling the creation of a large set of hybrid breast images containing a well-characterized mass, embedded within real breast background. The computational nature of the model makes it suitable for detectability studies, evaluation of computer aided diagnosis algorithms, and teaching purposes.

Stellate masses and histologic grades in breast cancer

Breast masses with a radiologic stellate pattern often transform into malignancies, but their tendency to be of low histologic grade yields a better survival rate compared with tumors with other patterns on mammography screening. This study was designed to investigate the correlation of histologic grade with stellate features extracted from the coronal plane of 3-D ultrasound images. A pre-processing method was proposed to facilitate the extraction of stellate features. Extracted features were statistically measured to derive a set of indices that quantitatively represent the stellate pattern. These indices then went through a selection procedure to build proper decision trees. The splitting rules of decision trees indicated that stellate tumors are associated with low grade. A set of indices from the low grade-associated rules has the potential to represent the stellate feature. Further investigation of the hypoechoic region of peripheral tissue is essential to establishment of a complete discriminating model for tumor grades.

Computerized determination scheme for histological classification of breast mass using objective features corresponding to clinicians' subjective impressions on ultrasonographic images

It is often difficult for clinicians to decide correctly on either biopsy or follow-up for breast lesions with masses on ultrasonographic images. The purpose of this study was to develop a computerized determination scheme for histological classification of breast mass by using objective features corresponding to clinicians' subjective impressions for image features on ultrasonographic images. Our database consisted of 363 breast ultrasonographic images obtained from 363 patients. It included 150 malignant (103 invasive and 47 noninvasive carcinomas) and 213 benign masses (87 cysts and 126 fibroadenomas). We divided our database into 65 images (28 malignant and 37 benign masses) for training set and 298 images (122 malignant and 176 benign masses) for test set. An observer study was first conducted to obtain clinicians' subjective impression for nine image features on mass. In the proposed method, location and area of the mass were determined by an experienced clinician. We defined some feature extraction methods for each of nine image features. For each image feature, we selected the feature extraction method with the highest correlation coefficient between the objective features and the average clinicians' subjective impressions. We employed multiple discriminant analysis with the nine objective features for determining histological classification of mass. The classification accuracies of the proposed method were 88.4 % (76/86) for invasive carcinomas, 80.6 % (29/36) for noninvasive carcinomas, 86.0 % (92/107) for fibroadenomas, and 84.1 % (58/69) for cysts, respectively. The proposed method would be useful in the differential diagnosis of breast masses on ultrasonographic images as diagnosis aid.

Optimization of breast mass classification using sequential forward floating selection (SFFS) and a support vector machine (SVM) model

PURPOSE

Improving radiologists' performance in classification between malignant and benign breast lesions is important to increase cancer detection sensitivity and reduce false-positive recalls. For this purpose, developing computer-aided diagnosis schemes has been attracting research interest in recent years. In this study, we investigated a new feature selection method for the task of breast mass classification.

METHODS

We initially computed 181 image features based on mass shape, spiculation, contrast, presence of fat or calcifications, texture, isodensity, and other morphological features. From this large image feature pool, we used a sequential forward floating selection (SFFS)-based feature selection method to select relevant features and analyzed their performance using a support vector machine (SVM) model trained for the classification task. On a database of 600 benign and 600 malignant mass regions of interest, we performed the study using a tenfold cross-validation method. Feature selection and optimization of the SVM parameters were conducted on the training subsets only.

RESULTS

The area under the receiver operating characteristic curve [Formula: see text] was obtained for the classification task. The results also showed that the most frequently selected features by the SFFS-based algorithm in tenfold iterations were those related to mass shape, isodensity, and presence of fat, which are consistent with the image features frequently used by radiologists in the clinical environment for mass classification. The study also indicated that accurately computing mass spiculation features from the projection mammograms was difficult, and failed to perform well for the mass classification task due to tissue overlap within the benign mass regions.

CONCLUSION

In conclusion, this comprehensive feature analysis study provided new and valuable information for optimizing computerized mass classification schemes that may have potential to be useful as a "second reader" in future clinical practice.

Breast image analysis for risk assessment, detection, diagnosis, and treatment of cancer

The role of breast image analysis in radiologists' interpretation tasks in cancer risk assessment, detection, diagnosis, and treatment continues to expand. Breast image analysis methods include segmentation, feature extraction techniques, classifier design, biomechanical modeling, image registration, motion correction, and rigorous methods of evaluation. We present a review of the current status of these task-based image analysis methods, which are being developed for the various image acquisition modalities of mammography, tomosynthesis, computed tomography, ultrasound, and magnetic resonance imaging. Depending on the task, image-based biomarkers from such quantitative image analysis may include morphological, textural, and kinetic characteristics and may depend on accurate modeling and registration of the breast images. We conclude with a discussion of future directions.

A similarity study of content-based image retrieval system for breast cancer using decision tree

PURPOSE

We are developing a decision tree content-based image retrieval (DTCBIR) CADx system to assist radiologists in characterization of breast masses on ultrasound images.

METHODS

Three DTCBIR configurations, including decision tree with boosting (DTb), decision tree with full leaf features (DTL), and decision tree with selected leaf features (DTLs) were compared. For DTb, features of a query mass were combined first into a merged feature score and then masses with similar scores were retrieved. For DTL and DTLs, similar masses were retrieved based on the Euclidean distance between feature vectors of the query and those of selected references. For each DTCBIR configuration, we investigated the use of full feature set and subset of features selected by the stepwise linear discriminant analysis (LDA) and simplex optimization method, resulting in six retrieval methods and selected five, DTb-lda, DTL-lda, DTb-full, DTL-full, and DTLs-full, for the observer study. Three MQSA radiologists rated similarities between the query mass and computer-retrieved three most similar masses using nine-point similarity scale (9 = very similar).

RESULTS

For DTb-lda, DTL-lda, DTb-full, DTL-full, and DTLs-full, average A(z) values were 0.90 ± 0.03, 0.85 ± 0.04, 0.87 ± 0.04, 0.79 ± 0.05, and 0.71 ± 0.06, respectively, and average similarity ratings were 5.00, 5.41, 4.96, 5.33, and 5.13, respectively.

CONCLUSIONS

The DTL-lda is a promising DTCBIR CADx configuration which had simple tree structure, good classification performance, and highest similarity rating.

Mass segmentation using a combined method for cancer detection

Background

Breast cancer is one of the leading causes of cancer death for women all over the world and mammography is thought of as one of the main tools for early detection of breast cancer. In order to detect the breast cancer, computer aided technology has been introduced. In computer aided cancer detection, the detection and segmentation of mass are very important. The shape of mass can be used as one of the factors to determine whether the mass is malignant or benign. However, many of the current methods are semi-automatic. In this paper, we investigate fully automatic segmentation method.

Results

In this paper, a new mass segmentation algorithm is proposed. In the proposed algorithm, a fully automatic marker-controlled watershed transform is proposed to segment the mass region roughly, and then a level set is used to refine the segmentation. For over-segmentation caused by watershed, we also investigated different noise reduction technologies. Images from DDSM were used in the experiments and the results show that the new algorithm can improve the accuracy of mass segmentation.

Conclusions

The new algorithm combines the advantages of both methods. The combination of the watershed based segmentation and level set method can improve the efficiency of the segmentation. Besides, the introduction of noise reduction technologies can reduce over-segmentation.

Multimodality computer-aided breast cancer diagnosis with FFDM and DCE-MRI

RATIONALE AND OBJECTIVES

To investigate a multimodality computer-aided diagnosis (CAD) scheme that combines image information from full-field digital mammography (FFDM) and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for computerized breast cancer classification.

MATERIALS AND METHODS

From a retrospective FFDM database with 432 lesions (255 malignant, 177 benign) and a retrospective DCE-MRI database including 476 lesions (347 malignant, 129 benign), we constructed a multimodality dataset of 213 lesions (168 malignant, 45 benign). Each lesion was present on both FFDM and DCE-MRI images and deemed to be a difficult case given the necessity of having both clinical imaging exams. Using a manually indicated lesion location (ie, a seed point on FFDM images or a region of interest on DCE-MRI images, the computer automatically segmented the mass lesions and extracted lesion features). A subset of features was selected using linear stepwise feature selection and merged by a Bayesian artificial neural network to yield an estimate of the probability of malignancy. Receiver operating characteristic (ROC) analysis was used to evaluate the performance of the selected features in distinguishing between malignant and benign lesions.

RESULTS

With leave-one-lesion-out cross-validation on the multimodality dataset, the mammography-only features yielded an area under the ROC curve (AUC) of 0.74 +/- 0.04, and the DCE-MRI-only features yielded an AUC of 0.78 +/- 0.04. The combination of these two modalities, which included a spiculation feature from mammography and two kinetic features from DCE-MRI, yielded an AUC of 0.87 +/- 0.03. The improvement of combining multimodality information was statistically significant as compared to the use of single modality information alone.

CONCLUSIONS

A CAD scheme that combines features extracted from FFDM and DCE-MRI images may be advantageous to single-modality CAD in the task of differentiating between malignant and benign lesions.

Image analysis in medical imaging: recent advances in selected examples

Medical imaging has developed into one of the most important fields within scientific imaging due to the rapid and continuing progress in computerised medical image visualisation and advances in analysis methods and computer-aided diagnosis. Several research applications are selected to illustrate the advances in image analysis algorithms and visualisation. Recent results, including previously unpublished data, are presented to illustrate the challenges and ongoing developments.

Exploring nonlinear feature space dimension reduction and data representation in breast Cadx with Laplacian eigenmaps and t-SNE

PURPOSE

In this preliminary study, recently developed unsupervised nonlinear dimension reduction (DR) and data representation techniques were applied to computer-extracted breast lesion feature spaces across three separate imaging modalities: Ultrasound (U.S.) with 1126 cases, dynamic contrast enhanced magnetic resonance imaging with 356 cases, and full-field digital mammography with 245 cases. Two methods for nonlinear DR were explored: Laplacian eigenmaps [M. Belkin and P. Niyogi, "Laplacian eigenmaps for dimensionality reduction and data representation," Neural Comput. 15, 1373-1396 (2003)] and t-distributed stochastic neighbor embedding (t-SNE) [L. van der Maaten and G. Hinton, "Visualizing data using t-SNE," J. Mach. Learn. Res. 9, 2579-2605 (2008)].

METHODS

These methods attempt to map originally high dimensional feature spaces to more human interpretable lower dimensional spaces while preserving both local and global information. The properties of these methods as applied to breast computer-aided diagnosis (CADx) were evaluated in the context of malignancy classification performance as well as in the visual inspection of the sparseness within the two-dimensional and three-dimensional mappings. Classification performance was estimated by using the reduced dimension mapped feature output as input into both linear and nonlinear classifiers: Markov chain Monte Carlo based Bayesian artificial neural network (MCMC-BANN) and linear discriminant analysis. The new techniques were compared to previously developed breast CADx methodologies, including automatic relevance determination and linear stepwise (LSW) feature selection, as well as a linear DR method based on principal component analysis. Using ROC analysis and 0.632+bootstrap validation, 95% empirical confidence intervals were computed for the each classifier's AUC performance.

RESULTS

In the large U.S. data set, sample high performance results include, AUC0.632+ = 0.88 with 95% empirical bootstrap interval [0.787;0.895] for 13 ARD selected features and AUC0.632+ = 0.87 with interval [0.817;0.906] for four LSW selected features compared to 4D t-SNE mapping (from the original 81D feature space) giving AUC0.632+ = 0.90 with interval [0.847;0.919], all using the MCMC-BANN.

CONCLUSIONS

Preliminary results appear to indicate capability for the new methods to match or exceed classification performance of current advanced breast lesion CADx algorithms. While not appropriate as a complete replacement of feature selection in CADx problems, DR techniques offer a complementary approach, which can aid elucidation of additional properties associated with the data. Specifically, the new techniques were shown to possess the added benefit of delivering sparse lower dimensional representations for visual interpretation, revealing intricate data structure of the feature space.

Determination of similarity measures for pairs of mass lesions on mammograms by use of BI-RADS lesion descriptors and image features

RATIONALE AND OBJECTIVES

To determine similarity measures for selection of pathology-known similar images that would be useful for radiologists as a reference guide in the diagnosis of new breast lesions on mammograms.

MATERIALS AND METHODS

The images were obtained from the Digital Database for Screening Mammography developed by the University of South Florida. For determination and evaluation of similarity measures, the "gold standard" of similarities for 300 pairs of masses was determined by 10 breast radiologists. For determining similarity measures that would agree with radiologists' similarity determination, an artificial neural network (ANN) was trained with the radiologists' subjective similarity ratings and the image features. The image features were determined subjectively using the Breast Imaging Reporting and Data System (BI-RADS) lesion descriptors and objectively by computerized image analysis. The similarity measures determined by the ANN were compared to the gold standard and evaluated in terms of the correlation coefficient.

RESULTS

The similarity measures determined using the BI-RADS descriptors only were not as useful as those determined by use of the image features only. When the BI-RADS margin ratings were combined with the image features, the correlation coefficient between the subjective ratings and the objective measures improved slightly (r = 0.76) compared to those based on the image features alone (r = 0.74).

CONCLUSIONS

The inclusion of the BI-RADS margin descriptors may be useful for determination of similarity measures, especially when it is difficult to obtain the manual outlines of the masses and if the BI-RADS descriptors were provided consistently by radiologists.

Correlative feature analysis on FFDM

Identifying the corresponding images of a lesion in different views is an essential step in improving the diagnostic ability of both radiologists and computer-aided diagnosis (CAD) systems. Because of the nonrigidity of the breasts and the 2D projective property of mammograms, this task is not trivial. In this pilot study, we present a computerized framework that differentiates between corresponding images of the same lesion in different views and noncorresponding images, i.e., images of different lesions. A dual-stage segmentation method, which employs an initial radial gradient index (RGI) based segmentation and an active contour model, is applied to extract mass lesions from the surrounding parenchyma. Then various lesion features are automatically extracted from each of the two views of each lesion to quantify the characteristics of density, size, texture and the neighborhood of the lesion, as well as its distance to the nipple. A two-step scheme is employed to estimate the probability that the two lesion images from different mammographic views are of the same physical lesion. In the first step, a correspondence metric for each pairwise feature is estimated by a Bayesian artificial neural network (BANN). Then, these pairwise correspondence metrics are combined using another BANN to yield an overall probability of correspondence. Receiver operating characteristic (ROC) analysis was used to evaluate the performance of the individual features and the selected feature subset in the task of distinguishing corresponding pairs from noncorresponding pairs. Using a FFDM database with 123 corresponding image pairs and 82 noncorresponding pairs, the distance feature yielded an area under the ROC curve (AUC) of 0.81 +/- 0.02 with leave-one-out (by physical lesion) evaluation, and the feature metric subset, which included distance, gradient texture, and ROI-based correlation, yielded an AUC of 0.87 +/- 0.02. The improvement by using multiple feature metrics was statistically significant compared to single feature performance.

Anniversary paper: History and status of CAD and quantitative image analysis: the role of Medical Physics and AAPM

The roles of physicists in medical imaging have expanded over the years, from the study of imaging systems (sources and detectors) and dose to the assessment of image quality and perception, the development of image processing techniques, and the development of image analysis methods to assist in detection and diagnosis. The latter is a natural extension of medical physicists' goals in developing imaging techniques to help physicians acquire diagnostic information and improve clinical decisions. Studies indicate that radiologists do not detect all abnormalities on images that are visible on retrospective review, and they do not always correctly characterize abnormalities that are found. Since the 1950s, the potential use of computers had been considered for analysis of radiographic abnormalities. In the mid-1980s, however, medical physicists and radiologists began major research efforts for computer-aided detection or computer-aided diagnosis (CAD), that is, using the computer output as an aid to radiologists-as opposed to a completely automatic computer interpretation-focusing initially on methods for the detection of lesions on chest radiographs and mammograms. Since then, extensive investigations of computerized image analysis for detection or diagnosis of abnormalities in a variety of 2D and 3D medical images have been conducted. The growth of CAD over the past 20 years has been tremendous-from the early days of time-consuming film digitization and CPU-intensive computations on a limited number of cases to its current status in which developed CAD approaches are evaluated rigorously on large clinically relevant databases. CAD research by medical physicists includes many aspects-collecting relevant normal and pathological cases; developing computer algorithms appropriate for the medical interpretation task including those for segmentation, feature extraction, and classifier design; developing methodology for assessing CAD performance; validating the algorithms using appropriate cases to measure performance and robustness; conducting observer studies with which to evaluate radiologists in the diagnostic task without and with the use of the computer aid; and ultimately assessing performance with a clinical trial. Medical physicists also have an important role in quantitative imaging, by validating the quantitative integrity of scanners and developing imaging techniques, and image analysis tools that extract quantitative data in a more accurate and automated fashion. As imaging systems become more complex and the need for better quantitative information from images grows, the future includes the combined research efforts from physicists working in CAD with those working on quantitative imaging systems to readily yield information on morphology, function, molecular structure, and more-from animal imaging research to clinical patient care. A historical review of CAD and a discussion of challenges for the future are presented here, along with the extension to quantitative image analysis.

Automated detection of breast mass spiculation levels and evaluation of scheme performance

RATIONALE AND OBJECTIVES

Although the spiculation levels of breast mass boundaries are a primary sign of malignancy for masses detected on mammography, developing an automated computerized method to detect spiculation levels and quantitatively evaluating the performance of such a method is a difficult task. The objectives of this study were to (1) develop and test a new method to improve mass segmentation and detect mass boundary spiculation levels and (2) assess the performance of this method using a relatively large imaging data set.

MATERIALS AND METHODS

The fully automated method developed for this study includes three image-processing steps. In the first step, the principle of maximum entropy is applied in the selected region of interest (ROI) after correcting the background trend to enhance the initial outlines of a mass. In the second step, an active-contour model is used to refine the initial outlines. In the third step, spiculated lines connected to the mass boundary are detected and identified using a special line detector. A quantitative spiculation index is computed to assess the degree of spiculation. To develop and evaluate this automated method, 211 ROIs depicting masses were extracted from a publicly available image database. Among these ROIs, 106 depicted circumscribed mass regions and 105 involved spiculated mass regions. The performance of the method was evaluated using receiver-operating characteristic (ROC) analysis.

RESULTS

The computed area under the ROC curve, when applying the method to the data set, was 0.701 +/- 0.027. By setting up a threshold at a spiculation index of 5.0, the method achieved an overall classification accuracy of 66.4%, with 54.3% sensitivity and 78.3% specificity.

CONCLUSIONS

In this study, a new computerized method with a number of unique characteristics was developed to detect spiculated mass regions, and a simple spiculation index was applied to quantify mass spiculation levels. Although this quantitative index can be used to distinguish between spiculated and circumscribed masses, the results also suggest that the automated detection of mass spiculation levels remains a technical challenge.

Evaluation of computer-aided diagnosis on a large clinical full-field digital mammographic dataset

RATIONALE AND OBJECTIVES

To convert and optimize our previously developed computerized analysis methods for use with images from full-field digital mammography (FFDM) for breast mass classification to aid in the diagnosis of breast cancer.

MATERIALS AND METHODS

An institutional review board approved protocol was obtained, with waiver of consent for retrospective use of mammograms and pathology data. Seven hundred thirty-nine FFDM images, which contained 287 biopsy-proven breast mass lesions, of which 148 lesions were malignant and 139 lesions were benign, were retrospectively collected. Lesion margins were delineated by an expert breast radiologist and were used as the truth for lesion-segmentation evaluation. Our computerized image analysis method consisted of several steps: 1) identified lesions were automatically extracted from the parenchymal background using computerized segmentation methods; 2) a set of image characteristics (mathematic descriptors) were automatically extracted from image data of the lesions and surrounding tissues; and 3) selected features were merged into an estimate of the probability of malignancy using a Bayesian artificial neural network classifier. Performance of the analyses was evaluated at various stages of the conversion using receiver-operating characteristic analysis.

RESULTS

An area under the curve value of 0.81 was obtained in the task of distinguishing between malignant and benign mass lesions in a round-robin by case evaluation on the entire FFDM dataset. We failed to show a statistically significant difference (P = .83) compared to results from our previous study in which the computerized classification was performed on digitized screen-film mammograms.

CONCLUSIONS

Our computerized analysis methods developed on digitized screen-film mammography can be converted for use with FFDM. Results show that the computerized analysis methods for the diagnosis of breast mass lesions on FFDM are promising, and can potentially be used to aid clinicians in the diagnostic interpretation of FFDM.

Prevalence scaling: applications to an intelligent workstation for the diagnosis of breast cancer

RATIONALE AND OBJECTIVES

Our goal was to investigate the effects of changes that the prevalence of cancer in a population have on the probability of malignancy (PM) output and an optimal combination of a true-positive fraction (TPF) and a false-positive fraction (FPF) of a mammographic and sonographic automatic classifier for the diagnosis of breast cancer.

MATERIALS AND METHODS

We investigate how a prevalence-scaling transformation that is used to change the prevalence inherent in the computer estimates of the PM affects the numerical and histographic output of a previously developed multimodality intelligent workstation. Using Bayes' rule and the binormal model, we study how changes in the prevalence of cancer in the diagnostic breast population affect our computer classifiers' optimal operating points, as defined by maximizing the expected utility.

RESULTS

Prevalence scaling affects the threshold at which a particular TPF and FPF pair is achieved. Tables giving the thresholds on the scaled PM estimates that result in particular pairs of TPF and FPF are presented. Histograms of PMs scaled to reflect clinically relevant prevalence values differ greatly from histograms of laboratory-designed PMs. The optimal pair (TPF, FPF) of our lower performing mammographic classifier is more sensitive to changes in clinical prevalence than that of our higher performing sonographic classifier.

CONCLUSIONS

Prevalence scaling can be used to change computer PM output to reflect clinically more appropriate prevalence. Relatively small changes in clinical prevalence can have large effects on the computer classifier's optimal operating point.

Eigendetection of masses considering false positive reduction and breast density information

The purpose of this article is to present a novel algorithm for the detection of masses in mammographic computer-aided diagnosis systems. Four key points provide the novelty of our approach: (1) the use of eigenanalysis for describing variation in mass shape and size; (2) a Bayesian detection methodology providing a mathematical sound framework, flexible enough to include additional information; (3) the use of a two-dimensional principal components analysis approach to facilitate false positive reduction; and (4) the incorporation of breast density information, a parameter correlated with the performance of most mass detection algorithms and which is not considered in existing approaches. To study the performance of the system two experiments were carried out. The first is related to the ability of the system to detect masses, and thus, free-response receiver operating characteristic analysis was used, showing that the method is able to give high accuracy at a high specificity (80% detection at 1.40 false positives per image). Second, the ability of the system to highlight the pixels belonging to a mass is studied using receiver operating characteristic analysis, resulting in A(z) = 0.89 +/- 0.04. In addition, the robustness of the approach is demonstrated in an experiment where we used the Digital Database for Screening Mammography database for training and the Mammographic Image Analysis Society database for testing the algorithm.

Characterization of mammographic masses based on level set segmentation with new image features and patient information

Computer-aided diagnosis (CAD) for characterization of mammographic masses as malignant or benign has the potential to assist radiologists in reducing the biopsy rate without increasing false negatives. The purpose of this study was to develop an automated method for mammographic mass segmentation and explore new image based features in combination with patient information in order to improve the performance of mass characterization. The authors' previous CAD system, which used the active contour segmentation, and morphological, textural, and spiculation features, has achieved promising results in mass characterization. The new CAD system is based on the level set method and includes two new types of image features related to the presence of microcalcifications with the mass and abruptness of the mass margin, and patient age. A linear discriminant analysis (LDA) classifier with stepwise feature selection was used to merge the extracted features into a classification score. The classification accuracy was evaluated using the area under the receiver operating characteristic curve. The authors' primary data set consisted of 427 biopsy-proven masses (200 malignant and 227 benign) in 909 regions of interest (ROIs) (451 malignant and 458 benign) from multiple mammographic views. Leave-one-case-out resampling was used for training and testing. The new CAD system based on the level set segmentation and the new mammographic feature space achieved a view-based Az value of 0.83 +/- 0.01. The improvement compared to the previous CAD system was statistically significant (p = 0.02). When patient age was included in the new CAD system, view-based and case-based Az values were 0.85 +/- 0.01 and 0.87 +/- 0.02, respectively. The study also demonstrated the consistency of the newly developed CAD system by evaluating the statistics of the weights of the LDA classifiers in leave-one-case-out classification. Finally, an independent test on the publicly available digital database for screening mammography with 132 benign and 197 malignant ROIs containing masses achieved a view-based Az value of 0.84 +/- 0.02.

Potential effect of different radiologist reporting methods on studies showing benefit of CAD

RATIONALE AND OBJECTIVES

To investigate the effect of different reporting methods and performance measures on the assessment of the benefit of computer-aided diagnosis (CAD) in characterizing malignant and benign breast lesions on mammography and sonography.

MATERIALS AND METHODS

In a previous study, 10 observers provided three types of reporting data (probability of malignancy [PM] estimates, Breast Imaging Reporting and Data System [BI-RADS] ratings, and biopsy decisions), both without and with CAD. The current study compares alternative performance measures computed from the three types of reporting data. The area under the receiver operating characteristic curve (AUC) was computed from both the PM estimates and the BI-RADS ratings, whereas sensitivity and specificity were computed from all three data types. Sensitivity and specificity values calculated from either the PM estimates or the BI-RADS ratings were determined by setting both constant and user-dependent thresholds. Student's t-tests were used to evaluate the statistical significance of the differences in the performance measures without and with CAD.

RESULTS

The average AUC values of the 10 observers calculated from either PM estimates or BI-RADS ratings demonstrated statistically significant improvements in performance with CAD, increasing from 0.87 to 0.92 or 0.93, respectively. However, the statistical significance of improvements in sensitivity or specificity depended on the type of reporting data used.

CONCLUSIONS

Use of different types of reporting data in the computation of sensitivity and specificity may result in different conclusions concerning the benefit of CAD. Meaningful determination of sensitivity and specificity from PM estimates require the use of user-dependent thresholds.

A dual-stage method for lesion segmentation on digital mammograms

Mass lesion segmentation on mammograms is a challenging task since mass lesions are usually embedded and hidden in varying densities of parenchymal tissue structures. In this article, we present a method for automatic delineation of lesion boundaries on digital mammograms. This method utilizes a geometric active contour model that minimizes an energy function based on the homogeneities inside and outside of the evolving contour. Prior to the application of the active contour model, a radial gradient index (RGI)-based segmentation method is applied to yield an initial contour closer to the lesion boundary location in a computationally efficient manner. Based on the initial segmentation, an automatic background estimation method is applied to identify the effective circumstance of the lesion, and a dynamic stopping criterion is implemented to terminate the contour evolution when it reaches the lesion boundary. By using a full-field digital mammography database with 739 images, we quantitatively compare the proposed algorithm with a conventional region-growing method and an RGI-based algorithm by use of the area overlap ratio between computer segmentation and manual segmentation by an expert radiologist. At an overlap threshold of 0.4, 85% of the images are correctly segmented with the proposed method, while only 69% and 73% of the images are correctly delineated by our previous developed region-growing and RGI methods, respectively. This resulting improvement in segmentation is statistically significant.

Characterization of mammographic masses using a gradient-based segmentation algorithm and a neural classifier

Computerized methods have recently shown a great potential in providing radiologists with a second opinion about the visual diagnosis of the malignancy of mammographic masses. The computer-aided diagnosis (CAD) system we developed for the mass characterization is mainly based on a segmentation algorithm and on the neural classification of several features computed on the segmented mass. Mass-segmentation plays a key role in most computerized systems. Our technique is a gradient-based one, showing the main characteristic that no free parameters have been evaluated on the data set used in this analysis, thus it can directly be applied to data sets acquired in different conditions without any ad hoc modification. A data set of 226 masses (109 malignant and 117 benign) has been used in this study. The segmentation algorithm works with a comparable efficiency both on malignant and benign masses. Sixteen features based on shape, size and intensity of the segmented masses are extracted and analyzed by a multi-layered perceptron neural network trained with the error back-propagation algorithm. The capability of the system in discriminating malignant from benign masses has been evaluated in terms of the receiver-operating characteristic (ROC) analysis. A feature selection procedure has been carried out on the basis of the feature discriminating power and of the linear correlations interplaying among them. The comparison of the areas under the ROC curves obtained by varying the number of features to be classified has shown that 12 selected features out of the 16 computed ones are powerful enough to achieve the best classifier performances. The radiologist assigned the segmented masses to three different categories: correctly-, acceptably- and non-acceptably-segmented masses. We initially estimated the area under ROC curve only on the first category of segmented masses (the 88.5% of the data set), then extending the classification to the second subclass (reaching the 97.8% of the data set) and finally to the whole data set, obtaining A(z)=0.805+/-0.030, 0.787+/-0.024 and 0.780+/-0.023, respectively.

Computer-aided diagnostic scheme for the detection of lung nodules on chest radiographs: localized search method based on anatomical classification

We developed an advanced computer-aided diagnostic (CAD) scheme for the detection of various types of lung nodules on chest radiographs intended for implementation in clinical situations. We used 924 digitized chest images (992 noncalcified nodules) which had a 500 x 500 matrix size with a 1024 gray scale. The images were divided randomly into two sets which were used for training and testing of the computerized scheme. In this scheme, the lung field was first segmented by use of a ribcage detection technique, and then a large search area (448 x 448 matrix size) within the chest image was automatically determined by taking into account the locations of a midline and a top edge of the segmented ribcage. In order to detect lung nodule candidates based on a localized search method, we divided the entire search area into 7 x 7 regions of interest (ROIs: 64 x 64 matrix size). In the next step, each ROI was classified anatomically into apical, peripheral, hilar, and diaphragm/heart regions by use of its image features. Identification of lung nodule candidates and extraction of image features were applied for each localized region (128 x 128 matrix size), each having its central part (64 x 64 matrix size) located at a position corresponding to a ROI that was classified anatomically in the previous step. Initial candidates were identified by use of the nodule-enhanced image obtained with the average radial-gradient filtering technique, in which the filter size was varied adaptively depending on the location and the anatomical classification of the ROI. We extracted 57 image features from the original and nodule-enhanced images based on geometric, gray-level, background structure, and edge-gradient features. In addition, 14 image features were obtained from the corresponding locations in the contralateral subtraction image. A total of 71 image features were employed for three sequential artificial neural networks (ANNs) in order to reduce the number of false-positive candidates. All parameters for ANNs, i.e., the number of iterations, slope of sigmoid functions, learning rate, and threshold values for removing the false positives, were determined automatically by use of a bootstrap technique with training cases. We employed four different combinations of training and test image data sets which was selected randomly from the 924 cases. By use of our localized search method based on anatomical classification, the average sensitivity was increased to 92.5% with 59.3 false positives per image at the level of initial detection for four different sets of test cases, whereas our previous technique achieved an 82.8% of sensitivity with 56.8 false positives per image. The computer performance in the final step obtained from four different data sets indicated that the average sensitivity in detecting lung nodules was 70.1% with 5.0 false positives per image for testing cases and 70.4% sensitivity with 4.2 false positives per image for training cases. The advanced CAD scheme involving the localized search method with anatomical classification provided improved detection of pulmonary nodules on chest radiographs for 924 lung nodule cases.

Classification of breast lesions with multimodality computer-aided diagnosis: observer study results on an independent clinical data set

PURPOSE

To evaluate a computer-aided diagnosis multimodality intelligent workstation as an aid to radiologists in the interpretation of mammograms and breast sonograms.

MATERIALS AND METHODS

An institutional review board approved the protocol for an observer study with signed consent, as well as the retrospective use of the mammograms, sonograms, and clinical data with waiver of consent. The HIPAA-compliant observer study was conducted with five breast radiologists and five breast imaging fellows, all of whom gave confidence ratings and patient management decisions, both without and with the computer aid, for 97 lesions that were unknown to both the observers and the computer. The performance of each observer without and with the computer aid was quantified by using four performance measures: area under the receiver operating characteristic curve (A(z)) value, partial A(z) value, sensitivity, and specificity. The statistical significance of the differences in the performance measures without and with the computer aid was determined by using a two-tailed t test for paired data.

RESULTS

Use of the computer aid resulted in an improvement of the average performance of the 10 observers, as measured by means of a statistically significant increase in A(z) value (0.87-0.92; P < .001), partial A(z) value (0.47-0.68; P < .001), and sensitivity (0.88-0.93; P = .005). A statistically significant difference was not found in the specificity without and with the computer aid (0.66-0.69; P = .20).

CONCLUSION

Use of multimodality intelligent workstations can improve the performance of radiologists in the task of differentiating malignant and benign lesions at mammography and sonography.

Automated detection of lung nodules in CT scans: false-positive reduction with the radial-gradient index

We present a number of approaches based on the radial gradient index (RGI) to achieve false-positive reduction in automated CT lung nodule detection. A database of 38 cases was used that contained a total of 82 lung nodules. For each CT section, a complementary image known as an "RGI map" was constructed to enhance regions of high circularity and thus improve the contrast between nodules and normal anatomy. Thresholds on three RGI parameters were varied to construct RGI filters that sensitively eliminated false-positive structures. In a consistency approach, RGI filtering eliminated 36% of the false-positive structures detected by the automated method without the loss of any true positives. Use of an RGI filter prior to a linear discriminant classifier yielded notable improvements in performance, with the false-positive rate at a sensitivity of 70% being reduced from 0.5 to 0.28 per section. Finally, the performance of the linear discriminant classifier was evaluated with RGI-based features. RGI-based features achieved a substantial improvement in overall performance, with a 94.8% reduction in the false-positive rate at a fixed sensitivity of 70%. These results demonstrate the potential role of RGI analysis in an automated lung nodule detection method.

Computerized mass detection for digital breast tomosynthesis directly from the projection images

Digital breast tomosynthesis (DBT) has recently emerged as a new and promising three-dimensional modality in breast imaging. In DBT, the breast volume is reconstructed from 11 projection images, taken at source angles equally spaced over an arc of 50 degrees. Reconstruction algorithms for this modality are not fully optimized yet. Because computerized lesion detection in the reconstructed breast volume will be affected by the reconstruction technique, we are developing a novel mass detection algorithm that operates instead on the set of raw projection images. Mass detection is done in three stages. First, lesion candidates are obtained for each projection image separately, using a mass detection algorithm that was initially developed for screen-film mammography. Second, the locations of a lesion candidate are backprojected into the breast volume. In this feature volume, voxel intensities are a combined measure of detection frequency (e.g., the number of projections in which a given lesion candidate was detected), and a measure of the angular range over which a given lesion was detected. Third, features are extracted after reprojecting the three-dimensional (3-D) locations of lesion candidates into projection images. Features are combined using linear discriminant analysis. The database used to test the algorithm consisted of 21 mass cases (13 malignant, 8 benign) and 15 cases without mass lesions. Based on this database, the algorithm yielded a sensitivity of 90% at 1.5 false positives per breast volume. Algorithm performance is positively biased because this dataset was used for development, training, and testing, and because the number of algorithm parameters was approximately the same as the number.of patient cases. Our results indicate that computerized mass detection in the sequence of projection images for DBT may be effective despite the higher noise level in those images.

A study on the computerized fractal analysis of architectural distortion in screening mammograms

Architectural distortion (AD) is a sign of malignancy often missed during mammographic interpretation. The purpose of this study was to explore the application of fractal analysis to the investigation of AD in screening mammograms. The study was performed using mammograms from the Digital Database for Screening Mammography (DDSM). The fractal dimension (FD) of mammographic regions of interest (ROIs) was calculated using the circular average power spectrum technique. Initially, the variability of the FD estimates depending on ROI location, mammographic view and breast side was studied on normal mammograms. Then, the estimated FD was evaluated using receiver operating characteristics (ROC) analysis to determine if it can discriminate ROIs depicting AD from those depicting normal breast parenchyma. The effect of several factors such as ROI size, image subsampling and breast density was studied in detail. Overall, the average FD of the normal ROIs was statistically significantly higher than that of the ROIs with AD. This result was consistent across all factors studied. For the studied set of implementation parameters, the best ROC performance achieved was 0.89 +/- 0.02. The generalizability of these conclusions across different digitizers was also demonstrated.

Use of border information in the classification of mammographic masses

We are developing a new method to characterize the margin of a mammographic mass lesion to improve the classification of benign and malignant masses. Towards this goal, we designed features that measure the degree of sharpness and microlobulation of mass margins. We calculated these features in a border region of the mass defined as a thin band along the mass contour. The importance of these features in the classification of benign and malignant masses was studied in relation to existing features used for mammographic mass detection. Features were divided into three groups, each representing a different mass segment: the interior region of a mass, the border and the outer area. The interior and the outer area of a mass were characterized using contrast and spiculation measures. Classification was done in two steps. First, features representing each of the three mass segments were merged into a neural network classifier resulting in a single regional classification score for each segment. Secondly, a classifier combined the three single scores into a final output to discriminate between benign and malignant lesions. We compared the classification performance of each regional classifier and the combined classifier on a data set of 1076 biopsy proved masses (590 malignant and 486 benign) from 481 women included in the Digital Database for Screening Mammography. Receiver operating characteristic (ROC) analysis was used to evaluate the accuracy of the classifiers. The area under the ROC curve (A(z)) was 0.69 for the interior mass segment, 0.76 for the border segment and 0.75 for the outer mass segment. The performance of the combined classifier was 0.81 for image-based and 0.83 for case-based evaluation. These results show that the combination of information from different mass segments is an effective approach for computer-aided characterization of mammographic masses. An advantage of this approach is that it allows the assessment of the contribution of regions rather than individual features. Results suggest that the border and the outer areas contained the most valuable information for discrimination between benign and malignant masses.

Investigation of psychophysical measure for evaluation of similar images for mammographic masses: preliminary results

We investigated a psychophysical similarity measure for selection of images similar to those of unknown masses on mammograms, which may assist radiologists in the distinction between benign and malignant masses. Sixty pairs of masses were selected from 1445 mass images prepared for this study, which were obtained from the Digital Database for Screening Mammography by the University of South Florida. Five radiologists provided subjective similarity ratings for these 60 pairs of masses based on the overall impression for diagnosis. Radiologists' subjective ratings were marked on a continuous rating scale and quantified between 0 and 1, which correspond to pairs not similar at all and pairs almost identical, respectively. By use of the subjective ratings as "gold standard," similarity measures based on the Euclidean distance between pairs in feature space and the psychophysical measure were determined. For determination of the psychophysical similarity measure, an artificial neural network (ANN) was employed to learn the relationship between radiologists' average subjective similarity ratings and computer-extracted image features. To evaluate the usefulness of the similarity measures, the agreement with the radiologists' subjective similarity ratings was assessed in terms of correlation coefficients between the average subjective ratings and the similarity measures. A commonly used similarity measure based on the Euclidean distance was moderately correlated (r=0.644) with the radiologists' average subjective ratings, whereas the psychophysical measure by use of the ANN was highly correlated (r=0.798). The preliminary result indicates that a psychophysical similarity measure would be useful in the selection of images similar to those of unknown masses on mammograms.

A study on two different CAD systems for mammography as an aid to radiological diagnosis in the search of microcalcification clusters

OBJECTIVE

The aim of the present study was to evaluate the efficacy of two different computer aided detection (CAD) systems for mammography in improving radiological diagnosis in the search of microcalcification clusters. The CAD systems used are: the SecondLooktrade mark (CADx Medical Systems, Canada) commercial system and the CALMA (computer assisted library in MAmmography) research CAD system. Three radiologists were asked to read mammographic images with and without the support of the CAD systems.

MATERIAL AND METHODS

Three radiologists with respectively 3, 5 and 7 years of practice in mammogram reading in an Italian public hospital analysed a dataset composed of 120 digitized mammograms of healthy subjects with no lesion (proven by a radiological follow up of at least 3 years) and 70 images of patients with malignant cluster of microcalcification (proven by histopathological examination) both with no CAD support as well as with the help of the SecondLooktrade mark system. After 3 months they were asked to observe the same digitized mammograms with the assistance of the CALMA system. The radiologists worked independently and were unaware of the final diagnosis. The values of the area A(z) under the ROC curve, diagnostic sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy were evaluated with and without the support of the CAD systems. The reading time and qualitative evaluations of each radiologist were also reported.

RESULTS

With the support of the two CAD systems an improvement in A(z) area was obtained ranging from 0.01 to 0.04. Sensitivity increased from +8.6 to +15.7% and specificity decreased from 0.8 to 4.2%.

CONCLUSION

In our study, not conditioned by the dataset, the CAD systems as second reader produced an increase in overall sensitivity of up to 15.7%, with a little decrease in specificity of up to 4.2%. Based on these results both CAD systems might be used in the current practise to improve the sensitivity values of conventional reading (radiologist alone). The results of this study show that no significant differences exist in term of A(z), sensitivity and specificity between CALMA and CADx.

Computerized analysis of mesothelioma on CT scans

Computed tomography (CT) has become an essential tool for monitoring the progression of mesothelioma tumor or a patient's response to treatment. Despite the accepted role of CT for the evaluation of mesothelioma, however, the current standard practice remains the manual measurement of tumor extent on CT scans. The process of manual measurement acquisition is tedious and subject to inconsistency. We have quantitatively assessed the variability of manual mesothelioma measurements both in baseline CT scans and in the context of tumor response classification across temporally sequential scans. To facilitate the acquisition of mesothelioma measurements, we have developed and evaluated computerized methods for the assessment of mesothelioma tumor thickness on CT scans; we anticipate that such computer-assisted approaches will make the radiologic assessment of mesothelioma more efficient, thus facilitating the establishment of improved clinical protocols.

Computerized analysis of images in the detection and diagnosis of breast cancer

Improvements in mammographic acquisition techniques have resulted in making the early signs of breast cancer more apparent on mammograms. However, the accuracy of the overall mammographic examination depends on both the quality of the mammographic images and the ability of the radiologist to interpret those images. While mammography is the best screening method for the early detection of breast cancer, radiologists do miss lesions on mammograms. Use of output, however, from a computerized analysis of an image by a radiologist may help him/her in the detection or diagnostic tasks, and potentially improve the overall interpretation of breast images and the subsequent patient care. Computer-aided detection and diagnosis (CAD) involves the application of computer technology to the process of medical image interpretation. CAD can be defined as a diagnosis made by a radiologist, who uses the output from a computerized analysis of medical images as a "second opinion" in detecting and diagnosing lesions, with the final diagnosis being made by the radiologist. The computer output must be at a sufficient performance level, and in addition, the output must be displayed in a user-friendly format for effective and efficient use by the radiologist. This chapter reviews CAD in breast cancer detection and diagnosis, including examples of image analyses, multi-modality approaches (i.e., special-view diagnostic mammography, ultrasound, and MRI), and means of communicating the computer output to the human.

Estimating three-class ideal observer decision variables for computerized detection and classification of mammographic mass lesions

We are using Bayesian artificial neural networks (BANNs) to classify mammographic masses in schemes for computer-aided diagnosis, and we are extending this methodology to a three-class classification task. We investigated whether a BANN can estimate ideal observer decision variables to distinguish malignant, benign, and false-positive computer detections. Five features were calculated for 63 malignant and 29 benign computer-detected mass lesions, and for 1049 false-positive computer detections, in 440 mammograms randomly divided into a training and testing set. A BANN was trained on the training set features and applied to the testing set features. We then used a known relation between three-class ideal observer decision variables and that used by a two-class ideal observer when two of three classes are grouped into one class, giving one decision variable for distinguishing malignant from nonmalignant detections, and a second for distinguishing true-positive from false-positive computer detections. For comparison, we grouped the training data into two classes in the same two ways and trained two-class BANNs for these two tasks. The three-class BANN decision variables were essentially identical in performance to the specifically trained two-class BANNs, with the average difference in area under the ROC curves being less than 0.0035 and no differences in area being statistically significant. Thus, the BANN outputs obey the same theoretical relationship as do the three-class and two-class ideal observer decision variables, which is consistent with the claim that the three-class BANN output can provide good estimates of the decision variables used by a three-class ideal observer.