Search

Although screening x-ray mammography has become a very sensitive method for detecting breast cancer, mammography has low specificity in its diagnostic stage. About 67-85% of breast biopsies are performed on benign lesions. Because of cost and detrimental effects of unnecessary biopsies, the number of biopsies performed on benign lesions needs to be reduced. In this research we are developing a highly sensitive and specific computer-aided diagnosis classifier based on the likelihood ratio, which is designed to aid physicians to identify lesions that should not be sent to biopsy.

The goal of this project is to improve the accuracy of the diagnosis of breast cancer from mammograms by using a computer-based system to provide the physician with a second opinion on whether or not to perform a biopsy. An estimated 2% to 10% of true cancers are not biopsied but are instead followed, while between 60% and 90% of breast biopsies are performed on benign lesions. This report documents progress that has been made in improving the accuracy of diagnoses from mammograms using a Case-Based Reasoning (CBR) approach. The CBR approach predicts the outcome of a biopsy from the known biopsy outcomes for similar cases. The current version of the CBR performs with an accuracy of 61% on a retrospective set of consecutive cases for which the clinical diagnostic accuracy was 35%. The CBR algorithm has four fundamental tasks: (1) specify a reference set of cases, (2) define a metric for the distance between cases, (3) define a rule (based on the distance metric) for selecting "similar" cases from the reference set, and (4) specify a classification technique for predicting the outcome of biopsy from the known outcomes of the selected similar reference cases. The reference database for this study contained about 1500 cases that were referred for biopsy at Duke University Medical Center between 1992 and 2000. Each case included the mammographer's description of the lesion using the BI-RADS (TM) lexicon, known risk factors, and outcomes in the form of benign or malignant status as determined by biopsy. The author concludes that CBR is a promising technique for identifying benign cases. (7 tables, 9 figures, 11 refs.), The original document contains color images. All DTIC reproductions will be in black and white.

The goal of this project is to improve the accuracy of the diagnosis of breast cancer from mammograms by using a computer-based system to provide the mammographer with a second opinion on whether or not to perform a biopsy. An estimated 2% to 10% of true cancers are not biopsied but are instead followed, while between 60% and 90% of breast biopsies are performed on benign lesions. This report documents progress that has been made in improving the accuracy of diagnoses from mammograms using a Case-Based Reasoning (CBR) approach. The CBR approach predicts the outcome of a biopsy from the known biopsy outcomes for similar cases. The current version of the CBR performs with an accuracy of 61% on a retrospective set of consecutive cases for which the clinical diagnostic accuracy was 35%. The CBR algorithm has four fundamental tasks: (1) specify a reference set of cases, (2) define a metric for the distance between cases, (3) define a rule (based on the distance metric) for selecting "similar" cases from the reference set, and (4) specify a classification technique for predicting the outcome of biopsy from the known outcomes of similar reference cases. The reference database for this study contained about 1500 cases that were referred for biopsy at Duke University Medical Center between 1992 and 2000. Each case included the mammographer's description of the lesion using the BI-RADS (TM) lexicon, known risk factors, and outcomes in the form of benign or malignant status as determined by biopsy. At a sensitivity of 0.98 relative to all biopsied lesions, the specificity of CBR was found to be 0.4. Thus, through the use of CBR 40% of the benign biopsies could have been avoided at the cost of delaying diagnosis for 2% of the malignancies. The results demonstrate the feasibility of developing CBR as a decision aid for breast biopsy using the BI-RADS lexicon to index the cases. (5 tables, 6 figures, 11 refs.)

The long-term goal of this work is to develop a computer aid for the decision for breast biopsy. in this -project, a database of BIRADS reports of mammographic findings was developed from five institutions. The focus has been to gather data from multiple sites in order to verify and whether the artificial neural network computer aid to the diagnosis of breast cancer can be translated between locations. In all, cases were acquired from Duke University, University of North Carolina, University of Maryland, University of Pennsylvania, and Sloan-Kettering Cancer Center. An cases included biopsy proof of the presence or absence of'-malignancy. in testing between institutions, the computer -aid -was found to -be robust with -little -loss of performance when cases from other institutions were used as reference cases. These results indicate the possibility that a centrally trained computer aid could provide assistance for the decision to biopsy. The deployment of this system into regional care facilities and into private mammography practices could facilitate transferring the expertise currently present in only a few tertiary care centers to the public at large and to smaller and more rural settings and thus improve access for under-served populations.

The long term goal of this work is to develop a computer aid for the decision for breast biopsy. In this project, an artificial neural network was developed to predict the outcome of biopsy from the mammographic features described in the BIRADS lexicon. The focus has been to optimize the performance of the artificial neural network as well as to investigate alternate decision models. Cases were acquired from Duke University, and all cases included biopsy proof of the presence or absence of malignancy. In testing, the computer aid was found to be able to reduce the number of benign lesions that would be biopsied by 40% while missing 2% of the malignancies. Several alternative classifier designs were found to be offer promise as well. The deployment of this system into regional care facilities and into private mammography practices could facilitate transferring the expertise currently present in only a few tertiary care centers to the public at large and to smaller and more rural settings and thus improve access for under-served populations.

While biopsy is a sensitive and specific test for breast cancer, to achieve a high sensitivity for cancer, many women with mammographic findings due to benign processes undergo biopsy. To improve the specificity of the decision to recommend biopsy and reduce the number of benign biopsies performed, a computer decision aid is being developed. ANN systems for prediction have been evaluated using a database of mammographic cases that were sent to biopsy with the results known. A case findings matching algorithm was implemented using a relational database to simplify and speed the coding. A Case-Based Reasoning approach was selected for this study since we wished to examine the cases and the similarity between them. To classify a given test case as benign or malignant, the case is compared to all previous cases, selecting those cases with were similar with regards to their findings. A decision variable was formed as the "malignancy ratio" computed as the ratio of the number of malignant cases to the total number of similar or "matched" cases. The system performed with an ROC area of 0.77. As described here, the system performs better than chance but poorer than the performance reported for radiologists on these data.

The long range goal of this project is to improve the accuracy and consistency of breast cancer diagnosis by developing a Computer Aided Diagnosis (CAD) system for early prediction of breast cancer from patients' mammographic findings and medical history. Specifically, this system will predict the malignancy of non-palpable lesions that are examined with diagnostic mammography and are considered for biopsy. The goal is to improve the specificity of diagnosis with little loss of sensitivity thus significantly improving the positive predictive value of breast biopsy. Toward this goal, we have developed an artificial neural network (ANN) to predict biopsy outcome from mammographic and history findings. In the first four years of the grant we have 1) developed a user interface for acquiring mammographic findings, 2) acquired 700 cases using the standardized BI-RADS. reporting system, 3) trained and evaluated several ANN predictive models, 4) conducted a small prospective study, 5) examined the inter-and intra-observer variability of the reporting lexicon, 6) investigated reducing the number of active input features, and 7) examined the sensitivity of the system to the techniques used for sampling the data.

We describe an Artificial Neural Network (ANN) approach to computer aided diagnosis of breast cancer from mammographic findings. An ANN has been developed to provide support for the clinical decision to perform breast biopsy. The system is designed to aid in the decision to biopsy those patients who have suspicious mammographic findings. The decision to biopsy can be viewed as a two stage process: 1)the mammographer views the mammogram and determines the presence or absence of image features such as calcifications and masses, 2) the presence and description of these features and the patient's medical history are merged to form a diagnosis. The ANN system is an aid to the second step and is motivated by the large fraction of biopsies that are benign. While mammography is a sensitive procedure for detecting breast cancer, the positive predictive value (PPV) is low. Only 10-34% of women who undergo biopsy for mammographically suspicious nonpalpable lesions actually are found to have malignancy (Kopans 1992) Between 0.5 -2.0% of all mammographic exams result in biopsy; several hundreds of thousands of biopsies are performed on benign lesions each year. The women undergoing biopsy for a benign finding are unnecessarily subjected to the discomfort, expense, potential complications, change in cosmetic appearance, and anxiety that can accompany breast biopsy.