Skip to main content
Utah's Foremost Platform for Undergraduate Research Presentation
2015 Abstracts

Peak Density Histograms and Pathology Interpretations for High-frequence Ultrasonic Testing of Breast Cancer Surgical Specimens

Robyn Omer, Utah Valley University

Physical Sciences

Removal of all malignant tissue during lumpectomy is critical for preventing local recurrence of the breast cancer. Failure to remove all cancer results in 20-40% of lumpectomy patients returning for additional surgery. At Utah Valley University, a method is being developed to detect cancer during the initial surgery to ensure all of the cancer has been removed. Peak density, which is the number of peaks and valleys in a specified spectral range of a high-frequency (HF) ultrasound signal, correlates to breast pathology in lumpectomy specimens. The objective of this study was to determine if the histograms of peak density versus the number of measurements provide information on corresponding breast tissue pathology. High-frequency ultrasonic data were obtained from a blind study of surgical specimens obtained from 73 lumpectomy patients at the Huntsman Cancer Institute in Salt Lake City, Utah, and South Jordan, Utah. The data were normalized to remove bias between patients. The ultrasonic signals were converted to spectra using a Fourier transform. Peak densities were calculated from the spectra by counting the number of peaks and valleys in the 20-80 MHz range. This was achieved by counting where the slopes of the spectra (their derivatives) crossed zero. A histogram was created by assigning each peak density value to a bin, and then counting the number of measurements that fell within that bin. The histogram of the peak densities produced an asymmetric Gaussian-type distribution with a range of peak density values from 0 to 27 and a mode of 5. Using threshold values determined from a pilot study for differentiating pathology with peak density, it was determined that the peak of the distribution (5-6) corresponded to normal tissue pathology, the shoulders of the distribution (0-4 and 7-10) corresponded to abnormal pathologies, and the tail of the distribution (11-27) corresponded to malignant tissue types. These correlations matched the types of specimens tested, specifically tumors, margins, and lymph nodes. The correlations also provide a measure of the success of removing malignant tissue and achieving negative margins during lumpectomy procedures. Using histograms to analyze the data not only provides a new approach for differentiating tissue pathology, but also provides a statistical measure of the success of lumpectomy procedures performed by a specific surgeon or at a specific institution.