Karen Drukker

Computerized lesion detection on breast ultrasound

We investigated the use of a radial gradient index (RGI) filtering technique to automatically detect lesions on breast ultrasound. After initial RGI filtering, a sensitivity of 87% at 0.76 false-positive detections per image was obtained on a database of 400 patients (757 images). Next, lesion candidates were segmented from the background by maximizing an average radial gradient (ARD) index for regions grown from the detected points. At an overlap of 0.4 with a radiologist lesion outline, 75% of the lesions were correctly detected. Subsequently, round robin analysis was used to assess the quality of the classification of lesion candidates into actual lesions and false-positives by a Bayesian neural network. The round robin analysis yielded an Az value of 0.84, and an overall performance by case of 94% sensitivity at 0.48 false-positives per image. Use of computerized analysis of breast sonograms may ultimately facilitate the use of sonography in breast cancer screening programs.

Computerized detection and classification of cancer on breast ultrasound1

RATIONALE AND OBJECTIVES To develop and evaluate a two-stage computerized method that first detects suspicious regions on ultrasound images, and subsequently distinguishes among different lesion types. MATERIALS AND METHODS The first stage of detecting potential lesions was based on expected lesion shape and margin characteristics. After the detection stage, all candidate lesions were classified by a Bayesian neural net based on computer-extracted lesion features. Two separate tasks were performed and evaluated at the classification stage: the first classification task was the distinction between all actual lesions and false-positive detections; the second classification task was the distinction between actual cancer and all other detected lesion candidates (including false-positive detections). The neural nets were trained on a database of 400 cases (757 images), consisting of complex cysts and benign and malignant lesions, and tested on an independent database of 458 cases (1,740 images including 578 normal images). RESULTS In the distinction between all actual lesions and false-positive detections, Az values of 0.94 and 0.91 were obtained with the training and testing data sets, respectively. Sensitivity by patient of 90% at 0.45 false-positive detections per image was achieved for this detection-plus-classification scheme for the testing data set. Distinguishing cancer from all other detections (false-positives plus all benign lesions) proved to be more challenging, and Az values of 0.87 and 0.81 were obtained during training and testing, respectively. Sensitivity by patient of 100% at 0.43 false-positive malignancies per image was achieved in the detection and classification of cancerous lesions for the testing dataset. CONCLUSION The results show promising performance of the computerized lesion detection and classification method, and indicate the potential of such a system for clinical breast ultrasound.

Quantitative MRI radiomics in the prediction of molecular classifications of breast cancer subtypes in the TCGA/TCIA data set.

Using quantitative radiomics, we demonstrate that computer-extracted magnetic resonance (MR) image-based tumor phenotypes can be predictive of the molecular classification of invasive breast cancers. Radiomics analysis was performed on 91 MRIs of biopsy-proven invasive breast cancers from National Cancer Institute’s multi-institutional TCGA/TCIA. Immunohistochemistry molecular classification was performed including estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2, and for 84 cases, the molecular subtype (normal-like, luminal A, luminal B, HER2-enriched, and basal-like). Computerized quantitative image analysis included: three-dimensional lesion segmentation, phenotype extraction, and leave-one-case-out cross validation involving stepwise feature selection and linear discriminant analysis. The performance of the classifier model for molecular subtyping was evaluated using receiver operating characteristic analysis. The computer-extracted tumor phenotypes were able to distinguish between molecular prognostic indicators; area under the ROC curve values of 0.89, 0.69, 0.65, and 0.67 in the tasks of distinguishing between ER+ versus ER−, PR+ versus PR−, HER2+ versus HER2−, and triple-negative versus others, respectively. Statistically significant associations between tumor phenotypes and receptor status were observed. More aggressive cancers are likely to be larger in size with more heterogeneity in their contrast enhancement. Even after controlling for tumor size, a statistically significant trend was observed within each size group (P=0.04 for lesions ⩽2 cm; P=0.02 for lesions >2 to ⩽5 cm) as with the entire data set (P-value=0.006) for the relationship between enhancement texture (entropy) and molecular subtypes (normal-like, luminal A, luminal B, HER2-enriched, basal-like). In conclusion, computer-extracted image phenotypes show promise for high-throughput discrimination of breast cancer subtypes and may yield a quantitative predictive signature for advancing precision medicine.

Computerized analysis of shadowing on breast ultrasound for improved lesion detection

Sonography is being considered for the screening of women at high risk for breast cancer. We are developing computerized detection methods to aid in the localization of lesions on breast ultrasound images. The detection scheme presented here is based on the analysis of posterior acoustic shadowing, since posterior acoustic shadowing is observed for many malignant lesions. The method uses a nonlinear filtering technique based on the skewness of the gray level distribution within a kernel of image data. The database used in this study included 400 breast ultrasound cases (757 images) consisting of complicated cysts, solid benign lesions, and malignant lesions. At a false-positive rate of 0.25 false positives per image, a detection sensitivity of 80% by case (66% by image) was achieved for malignant lesions. The performance for the overall database (at 0.25 false positives per image) was less at 42% sensitivity by case (30% by image) due to the more limited presence of posterior acoustic shadowing for benign solid lesions and the presence of posterior acoustic enhancement for cysts. Our computerized method for the detection of lesion shadows alerts radiologists to lesions that exhibit posterior acoustic shadowing. While this is not a characterization method, its performance is best for lesions that exhibit posterior acoustic shadowing such as malignant and, to a lesser extent, benign solid lesions. This method, in combination with other computerized sonographic detection methods, may ultimately help facilitate the use of ultrasound for breast cancer screening.

Deciphering Genomic Underpinnings of Quantitative MRI-based Radiomic Phenotypes of Invasive Breast Carcinoma

Magnetic Resonance Imaging (MRI) has been routinely used for the diagnosis and treatment of breast cancer. However, the relationship between the MRI tumor phenotypes and the underlying genetic mechanisms remains under-explored. We integrated multi-omics molecular data from The Cancer Genome Atlas (TCGA) with MRI data from The Cancer Imaging Archive (TCIA) for 91 breast invasive carcinomas. Quantitative MRI phenotypes of tumors (such as tumor size, shape, margin, and blood flow kinetics) were associated with their corresponding molecular profiles (including DNA mutation, miRNA expression, protein expression, pathway gene expression and copy number variation). We found that transcriptional activities of various genetic pathways were positively associated with tumor size, blurred tumor margin, and irregular tumor shape and that miRNA expressions were associated with the tumor size and enhancement texture, but not with other types of radiomic phenotypes. We provide all the association findings as a resource for the research community (available at http://compgenome.org/Radiogenomics/). These findings pave potential paths for the discovery of genetic mechanisms regulating specific tumor phenotypes and for improving MRI techniques as potential non-invasive approaches to probe the cancer molecular status.

Prediction of clinical phenotypes in invasive breast carcinomas from the integration of radiomics and genomics data

Abstract. Genomic and radiomic imaging profiles of invasive breast carcinomas from The Cancer Genome Atlas and The Cancer Imaging Archive were integrated and a comprehensive analysis was conducted to predict clinical outcomes using the radiogenomic features. Variable selection via LASSO and logistic regression were used to select the most-predictive radiogenomic features for the clinical phenotypes, including pathological stage, lymph node metastasis, and status of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). Cross-validation with receiver operating characteristic (ROC) analysis was performed and the area under the ROC curve (AUC) was employed as the prediction metric. Higher AUCs were obtained in the prediction of pathological stage, ER, and PR status than for lymph node metastasis and HER2 status. Overall, the prediction performances by genomics alone, radiomics alone, and combined radiogenomics features showed statistically significant correlations with clinical outcomes; however, improvement on the prediction performance by combining genomics and radiomics data was not found to be statistically significant, most likely due to the small sample size of 91 cancer cases with 38 radiomic features and 144 genomic features.

Automated Method for Improving System Performance of Computer-Aided Diagnosis in Breast Ultrasound

The purpose of this research was to demonstrate the feasibility of a computerized auto-assessment method in which a computer-aided diagnosis (CADx) system itself provides a level of confidence for its estimate for the probability of malignancy for each radiologist-identified lesion. The computer performance was assessed within a leave-one-case-out protocol using a database of sonographic images from 542 patients (19% cancer prevalence). We investigated the potential of computer-derived confidence levels both as 1) an output aid to radiologists and 2) as an automated method to improve the computer classification performance-in the task of differentiating between cancerous and benign lesions for the entire database. For the former, the CADx classification performance was assessed within ranges of confidence levels. For the latter, the computer-derived confidence levels were used in the determination of the computer-estimated probability of malignancy for each actual lesion based on probabilities obtained from different views. The use of this auto-assessment method resulted in the modest but statistically significant increase in the area under the receiver operating characteristic (ROC) curve (AUC value) of 0.01 with respect to the performance obtained using the ldquotraditionalrdquo CADx approach, increasing the AUC value from 0.89 to 0.90 (p -value 0.03). We believe that computer-provided confidence levels may be helpful to radiologists who are using CADx output in diagnostic image interpretation as well as for automated improvement of the CADx classification for cancer.

Breast US Computer-aided Diagnosis System: Robustness across Urban Populations in South Korea and the United States

PURPOSE To evaluate the robustness of a breast ultrasonographic (US) computer-aided diagnosis (CAD) system in terms of its performance across different patient populations. MATERIALS AND METHODS Three US databases were analyzed for this study: one South Korean and two United States databases. All three databases were utilized in an institutional review board-approved and HIPAA-compliant manner. Round-robin analysis and independent testing were performed to evaluate the performance of a computerized breast cancer classification scheme across the databases. Receiver operating characteristic (ROC) analysis was used to evaluate performance differences. RESULTS The round-robin analyses of each database demonstrated similar results, with areas under the ROC curve ranging from 0.88 (95% confidence interval [CI]: 0.820, 0.918) to 0.91 (95% CI: 0.86, 0.95). The independent testing of each database, however, indicated that although the performances were similar, the range in areas under the ROC curve (from 0.79 [95% CI: 0.730, 0.842] to 0.87 [95% CI: 0.794, 0.923]) was wider than that with the round-robin tests. However, the only instances in which statistically significant differences in performance were demonstrated occurred when the Korean database was used in a testing capacity in independent testing. CONCLUSION The few observed statistically significant differences in performance indicated that while the US features used by the system were useful across the databases, their relative importance differed. In practice, this means that a CAD system may need to be adjusted when applied to a different population.

Performance of breast ultrasound computer-aided diagnosis: dependence on image selection.

RATIONALE AND OBJECTIVES The automated classification of sonographic breast lesions is generally accomplished by extracting and quantifying various features from the lesions. The selection of images to be analyzed, however, is usually left to the radiologist. Here we present an analysis of the effect that image selection can have on the performance of a breast ultrasound computer-aided diagnosis system. MATERIALS AND METHODS A database of 344 different sonographic lesions was analyzed for this study (219 cysts/benign processes, 125 malignant lesions). The database was collected in an institutional review board-approved, Health Insurance Portability and Accountability Act-compliant manner. Three different image selection protocols were used in the automated classification of each lesion: all images, first image only, and randomly selected images. After image selection, two different protocols were used to classify the lesions: (a) the average feature values were input to the classifier or (b) the classifier outputs were averaged together. Both protocols generated an estimated probability of malignancy. Round-robin analysis was performed using a Bayesian neural network-based classifier. Receiver-operating characteristic analysis was used to evaluate the performance of each protocol. Significance testing of the performance differences was performed via 95% confidence intervals and noninferiority tests. RESULTS The differences in the area under the receiver-operating characteristic curves were never more than 0.02 for the primary protocols. Noninferiority was demonstrated between these protocols with respect to standard input techniques (all images selected and feature averaging). CONCLUSION We have proved that our automated lesion classification scheme is robust and can perform well when subjected to variations in user input.

A study of the effect of noise injection on the training of artificial neural networks

We studied the effect of noise injection in overcoming the problem of overtraining in the training of artificial neural networks (ANNs) in comparison with other common approaches for overcoming this problem such as early stopping of the ANN training process and weight decay (which is similar to Bayesian artificial neural networks). We found from simulation studies and studies of a computer-aided diagnosis application that noise injection is effective in overcoming overtraining and is as effective as, or even more effective than, early stopping and weight decay.