Differences between computer-aided diagnosis of breast masses and that of calcifications

BI-RADS 3-5 microcalcifications can preoperatively predict breast cancer HER2 and Luminal a molecular subtype

Purpose

To investigate associations between breast cancer molecular subtype and the patterns of mammographically detected calcifications.

Results

Identified were 93 (19.1%) Luminal A, 242 (49.9%) Luminal B, 108 (22.2%) HER2 and 42 (8.7%) basal subtypes. In univariate analysis, the clinicopathological parameters and BI-RADS 3–5 microcalcifications, which consisted 9 selected features was significantly associated with breast cancer molecular subtype (all P < 0.05). Among subtypes, multivariate analysis showed that calcification >2 cm in range (OR: 1.878, 95% CI: 1.150 to 3.067) and calcification > 0.5 mm in diameter (OR:2.206, 95% CI: 1.235 to 3.323) was independently predictive of HER2 subtype. The model showed good discrimination for predicting HER2 subtype, with a C-index of 0.704. In addition, multivariate analysis showed that calcification morphology (amorphour or coarse heterogenous calcifications OR: 2.847, 95% CI: 1.526 to 5.312) was independently predictive of Luminal A subtype. The model showed good discrimination for predicting Luminal A subtype, with a C-index of 0.74. And we demonstrated that amorphour or coarse heterogenous calcifications were associated with a higher incidence of Luminal A subtype than pleomorphic or fine linear or branching calcifications. There was no significant difference between breast cancer subtypes (Luminal B vs. other; Basal vs. other) and the patterns of mammographically detected calcifications.

Materials and Methods

Mammographic images of 485 female patients were included. The correlation between mammographic imaging features and breast cancer subtype was analyzed using Chi-square test, univariate and binary logistic regression analysis.

Conclusions

This study shows that BI-RADS 3–5 microcalcifications can be conveniently used to facilitate the preoperative prediction of HER2 and Luminal A molecular subtype in patients with infiltrating ductal carcinoma.

Sparse Representation Based Multi-Instance Learning for Breast Ultrasound Image Classification

We propose a novel method based on sparse representation for breast ultrasound image classification under the framework of multi-instance learning (MIL). After image enhancement and segmentation, concentric circle is used to extract the global and local features for improving the accuracy in diagnosis and prediction. The classification problem of ultrasound image is converted to sparse representation based MIL problem. Each instance of a bag is represented as a sparse linear combination of all basis vectors in the dictionary, and then the bag is represented by one feature vector which is obtained via sparse representations of all instances within the bag. The sparse and MIL problem is further converted to a conventional learning problem that is solved by relevance vector machine (RVM). Results of single classifiers are combined to be used for classification. Experimental results on the breast cancer datasets demonstrate the superiority of the proposed method in terms of classification accuracy as compared with state-of-the-art MIL methods.

Diagnosis of lymph node metastasis in uterine cervical cancer: usefulness of computer-aided diagnosis with comprehensive evaluation of MR images and clinical findings

BACKGROUND

Lymph node (LN) status is an important parameter for determining the treatment strategy and for predicting the prognosis for patients with uterine cervical cancer. Computer-aided diagnosis (CAD) can be feasible for differentiating metastatic from non-metastatic lymph nodes in patients with uterine cervical cancer.

PURPOSE

To determine the usefulness of CAD that comprehensively evaluates MR images and clinical findings for detecting LN metastasis in uterine cervical cancer.

MATERIAL AND METHODS

In 680 LNs from 143 patients who underwent radical hysterectomy for uterine cervical cancer, the CAD system using the Bayesian classifier estimated the probability of metastasis based on MR findings and clinical findings. We compared the diagnostic accuracy for detecting metastatic LNs in the CAD and MR findings.

RESULTS

Metastasis was diagnosed in 70 (12%) LNs from 34 (24%) patients. The area under ROC curves of CAD (0.924) was greater than those of the mean ADC (0.854), minimum ADC (0.849), maximum ADC (0.827), short-axis diameter (0.856) and long-axis diameter (0.753) (P < 0.05). The specificity and accuracy of the CAD (86%, 86%) were greater than those of the mean ADC (77%, 77%), maximum ADC (77%, 77%), minimum ADC (68%, 70%), and short-axis diameter (65%, 67%) (P < 0.05).

CONCLUSION

CAD system can improve the diagnostic performance of MR for detecting metastatic LNs in uterine cervical cancer.

NEURAL CLASSIFICATION OF MASS ABNORMALITIES WITH DIFFERENT TYPES OF FEATURES IN DIGITAL MAMMOGRAPHY

Early detection of breast abnormalities remains the primary prevention against breast cancer despite the advances in breast cancer diagnosis and treatment. Presence of mass in breast tissues is highly indicative of breast cancer. The research work presented in this paper investigates the significance of different types of features using proposed neural network based classification technique to classify mass type of breast abnormalities in digital mammograms into malignant and benign. 14 gray level based features, four BI-RADS features, patient age feature and subtlety value feature have been explored using the proposed research methodology to attain maximum classification on test dataset. The proposed research technique attained a 91% testing classification rate with a 100% training classification rate on digital mammograms taken from the DDSM benchmark database.

Computer-aided classification of breast masses: performance and interobserver variability of expert radiologists versus residents

PURPOSE

To evaluate the interobserver variability in descriptions of breast masses by dedicated breast imagers and radiology residents and determine how any differences in lesion description affect the performance of a computer-aided diagnosis (CAD) computer classification system.

MATERIALS AND METHODS

Institutional review board approval was obtained for this HIPAA-compliant study, and the requirement to obtain informed consent was waived. Images of 50 breast lesions were individually interpreted by seven dedicated breast imagers and 10 radiology residents, yielding 850 lesion interpretations. Lesions were described with use of 11 descriptors from the Breast Imaging Reporting and Data System, and interobserver variability was calculated with the Cohen κ statistic. Those 11 features were selected, along with patient age, and merged together by a linear discriminant analysis (LDA) classification model trained by using 1005 previously existing cases. Variability in the recommendations of the computer model for different observers was also calculated with the Cohen κ statistic.

RESULTS

A significant difference was observed for six lesion features, and radiology residents had greater interobserver variability in their selection of five of the six features than did dedicated breast imagers. The LDA model accurately classified lesions for both sets of observers (area under the receiver operating characteristic curve = 0.94 for residents and 0.96 for dedicated imagers). Sensitivity was maintained at 100% for residents and improved from 98% to 100% for dedicated breast imagers. For residents, the computer model could potentially improve the specificity from 20% to 40% (P < .01) and the κ value from 0.09 to 0.53 (P < .001). For dedicated breast imagers, the computer model could increase the specificity from 34% to 43% (P = .16) and the κ value from 0.21 to 0.61 (P < .001).

CONCLUSION

Among findings showing a significant difference, there was greater interobserver variability in lesion descriptions among residents; however, an LDA model using data from either dedicated breast imagers or residents yielded a consistently high performance in the differentiation of benign from malignant breast lesions, demonstrating potential for improving specificity and decreasing interobserver variability in biopsy recommendations.

Computer-aided diagnostic models in breast cancer screening

Mammography is the most common modality for breast cancer detection and diagnosis and is often complemented by ultrasound and MRI. However, similarities between early signs of breast cancer and normal structures in these images make detection and diagnosis of breast cancer a difficult task. To aid physicians in detection and diagnosis, computer-aided detection and computer-aided diagnostic (CADx) models have been proposed. A large number of studies have been published for both computer-aided detection and CADx models in the last 20 years. The purpose of this article is to provide a comprehensive survey of the CADx models that have been proposed to aid in mammography, ultrasound and MRI interpretation. We summarize the noteworthy studies according to the screening modality they consider and describe the type of computer model, input data size, feature selection method, input feature type, reference standard and performance measures for each study. We also list the limitations of the existing CADx models and provide several possible future research directions.

Multi-modality CADx: ROC study of the effect on radiologists' accuracy in characterizing breast masses on mammograms and 3D ultrasound images

RATIONALE AND OBJECTIVES

To investigate the effect of a computer-aided diagnosis (CADx) system on radiologists' performance in discriminating malignant and benign masses on mammograms and three-dimensional (3D) ultrasound (US) images.

MATERIALS AND METHODS

Our dataset contained mammograms and 3D US volumes from 67 women (median age, 51; range: 27-86) with 67 biopsy-proven breast masses (32 benign and 35 malignant). A CADx system was designed to automatically delineate the mass boundaries on mammograms and the US volumes, extract features, and merge the extracted features into a multi-modality malignancy score. Ten experienced readers (subspecialty academic breast imaging radiologists) first viewed the mammograms alone, and provided likelihood of malignancy (LM) ratings and Breast Imaging and Reporting System assessments. Subsequently, the reader viewed the US images with the mammograms, and provided LM and action category ratings. Finally, the CADx score was shown and the reader had the opportunity to revise the ratings. The LM ratings were analyzed using receiver-operating characteristic (ROC) methodology, and the action category ratings were used to determine the sensitivity and specificity of cancer diagnosis.

RESULTS

Without CADx, readers' average area under the ROC curve, A(z), was 0.93 (range, 0.86-0.96) for combined assessment of the mass on both the US volume and mammograms. With CADx, their average A(z) increased to 0.95 (range, 0.91-0.98), which was borderline significant (P = .05). The average sensitivity of the readers increased from 98% to 99% with CADx, while the average specificity increased from 27% to 29%. The change in sensitivity with CADx did not achieve statistical significance for the individual radiologists, and the change in specificity was statistically significant for one of the radiologists.

CONCLUSIONS

A well-trained CADx system that combines features extracted from mammograms and US images may have the potential to improve radiologists' performance in distinguishing malignant from benign breast masses and making decisions about biopsies.

Evaluation of a variable dose acquisition technique for microcalcification and mass detection in digital breast tomosynthesis

In this article the authors evaluate a recently proposed variable dose (VD)-digital breast tomosynthesis (DBT) acquisition technique in terms of the detection accuracy for breast masses and microcalcification (MC) clusters. With this technique, approximately half of the total dose is used for one center projection and the remaining dose is split among the other tomosynthesis projection views. This acquisition method would yield both a projection view and a reconstruction view. One of the aims of this study was to evaluate whether the center projection alone of the VD acquisition can provide equal or superior MC detection in comparison to the 3D images from uniform dose (UD)-DBT. Another aim was to compare the mass-detection capabilities of 3D reconstructions from VD-DBT and UD-DBT. In a localization receiver operating characteristic (LROC) observer study of MC detection, the authors compared the center projection of a VD acquisitioh scheme (at 2 mGy dose) with detector pixel size of 100 microm with the UD-DBT reconstruction (at 4 mGy dose) obtained with a voxel size of 100 microm. MCs with sizes of 150 and 180 microm were used in the study, with each cluster consisting of seven MCs distributed randomly within a small volume. Reconstructed images in UD-DBT were obtained from a projection set that had a total of 4 mGy dose. The current study shows that for MC detection, using the center projection alone of VD acquisition scheme performs worse with area under the LROC curve (AL) of 0.76 than when using the 3D reconstructed image using the UD acquisition scheme (AL=0.84). A 2D ANOVA found a statistically significant difference (p=0.038) at a significance level of 0.05. In the current study, although a reconstructed image was also available using the VD acquisition scheme, it was not used to assist the MC detection task which was done using the center projection alone. In the case of evaluation of detection accuracy of masses, the reconstruction with VD-DBT (AL=0.71) was compared to that obtained from the UD-DBT (AL=0.78). The authors found no statistically significant difference between the two (p-value=0.22), although all the observers performed better for UD-DBT.

A logistic regression model based on the national mammography database format to aid breast cancer diagnosis

OBJECTIVE

The purpose of our study was to create a breast cancer risk estimation model based on the descriptors of the National Mammography Database using logistic regression that can aid in decision making for the early detection of breast cancer.

MATERIALS AND METHODS

We created two logistic regression models based on the mammography features and demographic data for 62,219 consecutive mammography records from 48,744 studies in 18,269 [corrected] patients reported using the Breast Imaging Reporting and Data System (BI-RADS) lexicon and the National Mammography Database format between April 5, 1999 and February 9, 2004. State cancer registry outcomes matched with our data served as the reference standard. The probability of cancer was the outcome in both models. Model 2 was built using all variables in Model 1 plus radiologists' BI-RADS assessment categories. We used 10-fold cross-validation to train and test the model and to calculate the area under the receiver operating characteristic curves (A(z)) to measure the performance. Both models were compared with the radiologists' BI-RADS assessments.

RESULTS

Radiologists achieved an A(z) value of 0.939 +/- 0.011. The A(z) was 0.927 +/- 0.015 for Model 1 and 0.963 +/- 0.009 for Model 2. At 90% specificity, the sensitivity of Model 2 (90%) was significantly better (p < 0.001) than that of radiologists (82%) and Model 1 (83%). At 85% sensitivity, the specificity of Model 2 (96%) was significantly better (p < 0.001) than that of radiologists (88%) and Model 1 (87%).

CONCLUSION

Our logistic regression model can effectively discriminate between benign and malignant breast disease and can identify the most important features associated with breast cancer.

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 breast cancer classification using near-infrared optical tomographic images

An automated procedure for detecting breast cancer using near-infrared (NIR) tomographic images is presented. This classification procedure automatically extracts attributes from three imaging parameters obtained by an NIR imaging system. These parameters include tissue absorption and reduced scattering coefficients, as well as a tissue refractive index obtained by a phase-contrast-based reconstruction approach. A support vector machine (SVM) classifier is utilized to distinguish the malignant from the benign lesions using the automatically extracted attributes. The classification results of in vivo tomographic images from 35 breast masses using absorption, scattering, and refractive index attributes demonstrate high sensitivity, specificity, and overall accuracy of 81.8%, 91.7%, and 88.6% respectively, while the classification sensitivity, specificity, and overall accuracy are 63.6%, 83.3%, and 77.1%, respectively, when only the absorption and scattering attributes are used. Furthermore, the automated classification procedure provides significantly improved specificity and overall accuracy for breast cancer detection compared to those by an experienced technician through visual examination.

Breast ultrasound computer-aided diagnosis using BI-RADS features

RATIONALE AND OBJECTIVES

Based on the definitions in mass category of Breast Imaging Reporting and Data System developed by American College of Radiology, eight computerized features including shape, orientation, margin, lesion boundary, echo pattern, and posterior acoustic feature classes are proposed.

MATERIALS AND METHODS

Our experimental database consists of 265 pathology-proven cases including 180 benign and 85 malignant masses. The capacity of each proposed feature in differentiating malignant from benign masses was validated by Student's t test and the correlation between each proposed feature and the pathological result was evaluated by point biserial coefficient. Binary logistic regression model was used to relate all proposed features and pathological result as a computer-aided diagnosis (CAD) system. The diagnostic value of each proposed feature in the CAD system was further evaluated by the feature selection methods. Additionally, the likelihood of malignancy for each individual feature was also estimated by binary logistic regression.

RESULTS

On each proposed feature, the malignant cases were significantly different from the benign ones. The correlation between the angular characteristic and pathological result was indicated as very high. Three substantial correlations appear in features irregular shape, undulation characteristic, and degree of abrupt interface, but the relationship for orientation feature is low. For the constructed CAD system, the performance indices accuracy, sensitivity, specificity, PPV, and NPV were 91.70% (243 of 265), 90.59% (77 of 85), 92.22% (166 of 180), 84.62% (77 of 91), and 95.40% (166 of 174), respectively, and the area index in the ROC analysis was 0.97. Compared with the significant contribution of angular characteristic, the diagnostic values of posterior acoustic feature and orientation feature were relatively low for the CAD system. When three or more angular characteristics are discovered or the degree of abrupt interface is lower than 18, the likelihood of malignancy could be predicted as greater than 40%.

CONCLUSION

The computerized BI-RADS sonographic features conform to the sign of malignancy in the clinical experience and efficiently help the CAD system to diagnose the mass.

Texture analysis of tissue surrounding microcalcifications on mammograms for breast cancer diagnosis

Diagnosis of microcalcifications (MCs) is challenged by the presence of dense breast parenchyma, resulting in low specificity values and thus in unnecessary biopsies. The current study investigates whether texture properties of the tissue surrounding MCs can contribute to breast cancer diagnosis. A case sample of 100 biopsy-proved MC clusters (46 benign, 54 malignant) from 85 dense mammographic images, included in the Digital Database for Screening Mammography, was analysed. Regions of interest (ROIs) containing the MCs were pre-processed using a wavelet-based contrast enhancement method, followed by local thresholding to segment MCs; the segmented MCs were excluded from original image ROIs, and the remaining area (surrounding tissue) was subjected to texture analysis. Four categories of textural features (first order statistics, co-occurrence matrices features, run length matrices features and Laws' texture energy measures) were extracted from the surrounding tissue. The ability of each feature category in discriminating malignant from benign tissue was investigated using a k-nearest neighbour (kNN) classifier. An additional classification scheme was performed by combining classification outputs of three textural feature categories (the most discriminating ones) with a majority voting rule. Receiver operating characteristic (ROC) analysis was conducted for classifier performance evaluation of the individual textural feature categories and of the combined classification scheme. The best performance was achieved by the combined classification scheme yielding an area under the ROC curve (A(z)) of 0.96 (sensitivity 94.4%, specificity 80.0%). Texture analysis of tissue surrounding MCs shows promising results in computer-aided diagnosis of breast cancer and may contribute to the reduction of unnecessary biopsies.

Bayesian networks of BI-RADStrade mark descriptors for breast lesion classification

We investigated Bayesian network structure learning and probability estimation from mammographic feature data in order to classify breast lesions into different pathological categories. We compared the learned networks to naive Bayes classifiers, which are similar to the expert systems previously investigated for breast lesion classification. The learned network structures reflect the difference in the classification of biopsy outcome and the invasiveness of malignant lesions for breast masses and microcalcifications. The difference between masses and microcalcifications should be taken into consideration when interpreting systems for automatic pathological classification of breast lesions. The difference may also affect use of these systems for tasks such as estimating the sampling error of biopsy.

Breast cancer CADx based on BI-RAds descriptors from two mammographic views

In this study we compared the performance of computer aided diagnosis (CADx) algorithms based on Breast Imaging Reporting And Data System (BI-RADS) descriptors from one or two views. To select cases for the study with different mediolateral (MLO) and craniocaudal (CC) view descriptors, we assessed the agreement in BI-RADS lesion descriptors, BI-RADS assessment, and subtlety ratings for 1626 cases from the Digital Database for Screening Mammogrpahy (DDSM) using kappa statistics. We used 115 mass caseswith different descriptors for the two views to design linear discriminant analysis (LDA) based CADx algorithms. The CADx algorithms used BI-RADS descriptors and patient age as features. Thealgorithms based on BI-RADS descriptors from both the views performed marginally betterthan algorithms based on BI-RADS descriptors from a single view. A system that averaged theresults of two classifiers trained separately on the MLO and CC views displayed the best performance (Az=0.920 +/- 0.027). Thus, some improvement in performance of BI-RADS based CADx algorithms may be achieved by combining information from two mammographic views.

Optimized approach to decision fusion of heterogeneous data for breast cancer diagnosis

As more diagnostic testing options become available to physicians, it becomes more difficult to combine various types of medical information together in order to optimize the overall diagnosis. To improve diagnostic performance, here we introduce an approach to optimize a decision-fusion technique to combine heterogeneous information, such as from different modalities, feature categories, or institutions. For classifier comparison we used two performance metrics: The receiving operator characteristic (ROC) area under the curve [area under the ROC curve (AUC)] and the normalized partial area under the curve (pAUC). This study used four classifiers: Linear discriminant analysis (LDA), artificial neural network (ANN), and two variants of our decision-fusion technique, AUC-optimized (DF-A) and pAUC-optimized (DF-P) decision fusion. We applied each of these classifiers with 100-fold cross-validation to two heterogeneous breast cancer data sets: One of mass lesion features and a much more challenging one of microcalcification lesion features. For the calcification data set, DF-A outperformed the other classifiers in terms of AUC (p < 0.02) and achieved AUC=0.85 +/- 0.01. The DF-P surpassed the other classifiers in terms of pAUC (p < 0.01) and reached pAUC=0.38 +/- 0.02. For the mass data set, DF-A outperformed both the ANN and the LDA (p < 0.04) and achieved AUC=0.94 +/- 0.01. Although for this data set there were no statistically significant differences among the classifiers' pAUC values (pAUC=0.57 +/- 0.07 to 0.67 +/- 0.05, p > 0.10), the DF-P did significantly improve specificity versus the LDA at both 98% and 100% sensitivity (p < 0.04). In conclusion, decision fusion directly optimized clinically significant performance measures, such as AUC and pAUC, and sometimes outperformed two well-known machine-learning techniques when applied to two different breast cancer data sets.

Digital Mammography: Current Capabilities and Obstacles

Digital mammography represents an exciting new technology for breast imaging and possibly breast screening. The decoupling of functional components in digital mammography translates into potential operational efficiencies compared with screen-film mammography (SFM). Digital mammography is a platform for advanced applications not possible with traditional SFM. However, for digital mammography to replace SFM in daily clinical practice, operational and clinical hurdles will have to be overcome.

Tolerance to missing data using a likelihood ratio based classifier for computer-aided classification of breast cancer

While mammography is a highly sensitive method for detecting breast tumours, its ability to differentiate between malignant and benign lesions is low, which may result in as many as 70% of unnecessary biopsies. The purpose of this study was to develop a highly specific computer-aided diagnosis algorithm to improve classification of mammographic masses. A classifier based on the likelihood ratio was developed to accommodate cases with missing data. Data for development included 671 biopsy cases (245 malignant), with biopsy-proved outcome. Sixteen features based on the BI-RADS lexicon and patient history had been recorded for the cases, with 1.3 +/- 1.1 missing feature values per case. Classifier evaluation methods included receiver operating characteristic and leave-one-out bootstrap sampling. The classifier achieved 32% specificity at 100% sensitivity on the 671 cases with 16 features that had missing values. Utilizing just the seven features present for all cases resulted in decreased performance at 100% sensitivity with average 19% specificity. No cases and no feature data were omitted during classifier development, showing that it is more beneficial to utilize cases with missing values than to discard incomplete cases that cannot be handled by many algorithms. Classification of mammographic masses was commendable at high sensitivity levels, indicating that benign cases could be potentially spared from biopsy.

Sensitivity of noncommercial computer-aided detection system for mammographic breast cancer detection: pilot clinical trial

PURPOSE

To evaluate a noncommercial computer-aided detection (CAD) program for breast cancer detection with screening mammography.

MATERIALS AND METHODS

A CAD program was developed for mammographic breast cancer detection. The program was applied to 2,389 patients' screening mammograms at two geographically remote academic institutions (institutions A and B). Thirteen radiologists who specialized in breast imaging participated in this pilot study. For each case, the individual radiologist performed a prospective Breast Imaging Reporting and Data System (BI-RADS) assessment after viewing of the screening mammogram. Subsequently, the radiologist was shown CAD results and rendered a second BI-RADS assessment by using knowledge of both mammographic appearance and CAD results. Outcome analysis of results of examination in patients recalled for a repeat examination, of biopsy, and of 1-year follow-up examination was recorded. Correct detection with CAD included a computer-generated mark indicating a possible malignancy on craniocaudal or mediolateral oblique views or both.

RESULTS

Eleven (0.46%) of 2,389 patients had mammographically detected nonpalpable breast cancers. Ten (91%) of 11 (95% CI: 74%, 100%) cancers were correctly identified with CAD. Radiologist sensitivity without CAD was 91% (10 of 11; 95% CI: 74%, 100%). In 1,077 patients, follow-up findings were documented at 1 year. Five (0.46%) patients developed cancers, which were found on subsequent screening mammograms. The area where the cancers developed in two (40%) of these five patients was marked (true-positive finding) by the computer in the preceding year. Because of CAD results, a 9.7% increase in recall rate from 14.4% (344 of 2,389) to 15.8% (378 of 2,389) occurred. Radiologists' recall rate of study patients prior to use of CAD was 31% higher than the average rate for nonstudy cases (10.3%) during the same time period at institution A.

CONCLUSION

Performance of the CAD program had a very high sensitivity of 91% (95% CI: 74%, 100%).

Self-organizing map for cluster analysis of a breast cancer database

The purpose of this study was to identify and characterize clusters in a heterogeneous breast cancer computer-aided diagnosis database. Identification of subgroups within the database could help elucidate clinical trends and facilitate future model building. A self-organizing map (SOM) was used to identify clusters in a large (2258 cases), heterogeneous computer-aided diagnosis database based on mammographic findings (BI-RADS) and patient age. The resulting clusters were then characterized by their prototypes determined using a constraint satisfaction neural network (CSNN). The clusters showed logical separation of clinical subtypes such as architectural distortions, masses, and calcifications. Moreover, the broad categories of masses and calcifications were stratified into several clusters (seven for masses and three for calcifications). The percent of the cases that were malignant was notably different among the clusters (ranging from 6 to 83%). A feed-forward back-propagation artificial neural network (BP-ANN) was used to identify likely benign lesions that may be candidates for follow up rather than biopsy. The performance of the BP-ANN varied considerably across the clusters identified by the SOM. In particular, a cluster (#6) of mass cases (6% malignant) was identified that accounted for 79% of the recommendations for follow up that would have been made by the BP-ANN. A classification rule based on the profile of cluster #6 performed comparably to the BP-ANN, providing approximately 25% specificity at 98% sensitivity. This performance was demonstrated to generalize to a large (2177) set of cases held-out for model validation.

Development and evaluation of a case-based reasoning classifier for prediction of breast biopsy outcome with BI-RADS lexicon

Approximately 70-85% of breast biopsies are performed on benign lesions. To reduce this high number of biopsies performed on benign lesions, a case-based reasoning (CBR) classifier was developed to predict biopsy results from BI-RADS findings. We used 1433 (931 benign) biopsy-proven mammographic cases. CBR similarity was defined using either the Hamming or Euclidean distance measure over case features. Ten features represented each case: calcification distribution, calcification morphology, calcification number, mass margin, mass shape, mass density, mass size, associated findings, special cases, and age. Performance was evaluated using Round Robin sampling, Receiver Operating Characteristic (ROC) analysis, and bootstrap. To determine the most influential features for the CBR, an exhaustive feature search was performed over all possible feature combinations (1022) and similarity thresholds. Influential features were defined as the most frequently occurring features in the feature subsets with the highest partial ROC areas (0.90AUC). For CBR with Hamming distance, the most influential features were found to be mass margin, calcification morphology, age, calcification distribution, calcification number, and mass shape, resulting in an 0.90AUC of 0.33. At 95% sensitivity, the Hamming CBR would spare from biopsy 34% of the benign lesions. At 98% sensitivity, the Hamming CBR would spare 27% benign lesions. For the CBR with Euclidean distance, the most influential feature subset consisted of mass margin, calcification morphology, age, mass density, and associated findings, resulting in 0.90AUC of 0.37. At 95% sensitivity, the Euclidean CBR would spare from biopsy 41% benign lesions. At 98% sensitivity, the Euclidean CBR would spare 27% benign lesions. The profile of cases spared by both distance measures at 98% sensitivity indicates that the CBR is a potentially useful diagnostic tool for the classification of mammographic lesions, by recommending short-term follow-up for likely benign lesions that is in agreement with final biopsy results and mammographer's intuition.