SAN ANTONIOA computerized palpation device might offer an objective complement to clinical breast examination for detection of suspicious lumps, according to findings reported at the 22nd Annual San Antonio Breast Cancer Symposium. Lesion size as determined by the computerized palpation device better correlated with the palpable extent of the excised lesion than did size as determined by ultrasound, mammography, or clinical breast examination.
Weve just begun a trial that we hope will lead to FDA approval of the device, said Kevin Hughes, MD, director of the breast center at Lahey Clinic, Burlington, Mass. We will be looking at patients who have breast lumps and evaluating the reproducibility of the identification of lumps with the device. Well compare pictures obtained several weeks or months apart to see if we get the same picture with the device on each occasion.
The ultimate goal, he said, is to have a device that gives an objective two-dimensional picture of the lesion, as opposed to a subjective physician exam, which gives results that are difficult to reproduce and quantify. We hope the device provides objective information that we can document and follow over time, he said.
The BreastView Visual Mapping System (Assurance Medical, Inc., Hopkinton, Mass) includes a scan head containing multiple tactile sensors that detect changes in internal pressure as the scan head is moved across the breast.
The sensors transmit electronic signals to a computer, which processes and transforms the signals into tactile maps that provide a graphic representation of breast tissue, including any lumps detected (see Figures 1 and 2).
A pilot study of 100 women led to identification of 112 breast lesions, 30 of which subsequently were excised and found to be cancer.
Each patient was evaluated by the BreastView, mammography, breast ultrasound, and clinical breast examination. The accuracy of each type of evaluation was determined by computing the mean absolute error associated with a modalitys size estimation of the 30 cancers. The mean absolute error was derived by comparing a measurement against the actual lesion size at excision.
The BreastView resulted in a mean absolute error of 22.8%, which was significantly lower than that of ultrasound (35%), mammography (46.3%), or clinical breast exam (44.6%).
The reproducibility of BreastView findings was determined by assessment of at least three technically acceptable tactile maps generated for each of the 30 breast cancers. The analysis showed the device had the least variability (15%) for lesions that were 10 mm to 25 mm in diameter and the greatest variability (24.2%) for lesions greater than 25 mm. Overall, variability averaged 15.9%.
Dr. Hughes said the accuracy of the BreastView is susceptible to some of the same problems encountered during clinical breast examination, in particular, large breasts and firm breast tissue.
The device has the potential to damp out information that is not important, he said. We already do that with our hands during a clinical breast examination. By use of computer simulation techniques, we hope to improve on what we can achieve with a clinical examination.