Susan G Komen  
I've Been Diagnosed With Breast Cancer Someone I Know Was Diagnosed Share Your Story Join Us And Stay Informed Donate To End Breast Cancer
    Home > Research & Grants > Grants Program > Research Grants > Research Grants Awarded > Abstract

    Research Grants Awarded

    Identification And Treatment-Monitoring Using Advanced Ultrasonic Imaging Methods

    Grant Mechanism:
    Investigator Initiated Research

    Scientific Abstract:
    This proposed research project seeks to make significant improvements distinguishing cancerous from non-cancerous breast tissue. Success will result in a reduction in unwarranted breast biopsies and will provide an effective means of monitoring non-surgical treatments of breast cancer. The proposed program is based on successful prior pilot studies supported by a BCRP IDEA grant (DAMD17-98-1-8331, PI: FL Lizzi) and prior breast-elastography studies by the proposed PI at University of Texas supported by an NIH P01 grant (CA64597, PI: J Ophir). The proposed PI was an active participant in both prior studies, which convincingly demonstrated the feasibility of concepts to be fully exploited under the proposed grant. The proposed study involves two important clinical applications. 1) For detected lesions, we will use the three methods (elastography, spectrum analysis, and morphometric analysis) to identify benign lesions that do not have to be biopsied. Spectrum analysis and morphometric analysis, in combination with elastography, will provide us with reliable lesion identification. 2) For patients undergoing pre-surgical treatment to reduce tumor size, we will monitor cancerous tumors using the three methods described above. Although some volume-estimation methods are available, our proposed methods will provide us with a more accurate and precise, inexpensive, quantitative means of tumor-volume estimation. We will develop specialized approaches that beneficially exploit the unique sensitivity of elastography, spectrum analysis, and morphometric analysis, to be used individually or in combination, as needed. These will be designed specifically to take advantage of the differences among various properties of benign and malignant lesions, including elasticity and acoustic properties. Recent clinical studies have shown that a combination of specific B-mode features can be effective for differentiating benign and malignant breast lesions. We currently have the expertise to quantify these features for reliable discrimination. In addition, elastography, i.e., strain imaging, has demonstrated great potential for differentiating benign and malignant breast masses. Stiffer masses, such as cancers, deform less than surrounding non-cancerous tissue and thus appear darker in strain images. Benign lesions also may be stiffer than normal breast tissue, but because malignant breast lesions are stiffer than benign lesions, they appear darker (have higher contrast) in strain images. Furthermore, malignant lesions typically appear larger on strain images than in B-mode images. Finally, benign and malignant lesions have different nonlinear elastic properties. As a result, strain contrast typically increases with increasing strain for malignant lesions, whereas it decreases with increasing strain for benign lesions. The reliability of distinguishing benign from malignant lesions can be significantly improved if all criteria are used in combination. Initially, the performance of the proposed methods will be tested using tissue-mimicking phantoms that will contain rigid lesions. We will evaluate how accurately our methods can estimate lesion volumes. Subsequently, the efficacy of our proposed methods in differentiating benign and malignant lesions will be validated using biopsy results, and performance will be assessed by evaluating specificity and sensitivity as well as by statistical methods including ROC analyses. Finally, our proposed treatment-monitoring approach will be validated by comparing the estimated tumor volume prior to lumpectomy surgery with the volume measured using 3-D renderings generated from histology sections of the lumpectomy specimen.

    Lay Abstract:
    We will study an advanced ultrasound-based approach for identification and treatment-monitoring of breast cancer. We will use three methods, either individually or in combination, as needed: 1) elastography, 2) spectrum analysis, and 3) morphometric analysis. 1) Cancerous tissues generally are stiffer than normal tissues. Physicians routinely use palpation for detecting cancers in the breast, prostate, etc. However, manual palpation is useful only for shallow lesions and when lesions are much stiffer than their surrounding tissues. Breast elastography can be though of as a quantitative and more-sensitive implementation of manual breast palpation. In elastography, an external compression is applied to the tissue surface. The induced internal tissue deformations are estimated by analyzing ultrasonic echo signals acquired before and after compression. Because malignant tumors are significantly stiffer than benign lesions, cancers are expected to produce lower relative strain with respect to the surrounding tissue. Elastography studies already have shown great potential in the detection and identification of lesions in the breast. 2) Physicians demonstrated that certain features in B-mode ultrasonic images can be used to reliably differentiate cancerous breast lesions from the benign ones. These features include descriptors of appearance of the lesions itself (acoustic features) and descriptors of the lesion border (morphometric features). Our group at RRI developed a spectrum-analysis method for tissue characterization. The processing used for generating the (B-mode) image displayed on clinical ultrasonic scanners discards information that may be useful for describing tissue characteristics. The spectrum analysis method exploits this information, and we have been successfully investigating and applying this method for many years to classify tumors in several organs. We have performed a breast-cancer study that used spectrum analysis to quantify acoustic features. 3) In the study mentioned above, we have used the quantified tumor margin/border descriptors to automatically classify breast lesions. We achieved an ROC-curve area of 0.9164 ñ 0.0346 for 130 patients; however, we believe the proposed studies will result in marked improvements in our ability to distinguish cancerous from non-cancerous breast lesions and to monitor the size of lesions accurately. The proposed study involves two important clinical applications. 1) For detected lesions, we will use the above three methods (elastography, spectrum analysis, and morphometric analysis) to identify benign lesions that do not have to be biopsied. Spectrum analysis and morphometric analysis, in combination with elastography will provide us with reliable lesion identification. 2) For patient undergoing pre-surgical treatment to reduce tumor size, we will monitor cancerous tumors using the three methods described above. Although some volume-estimation methods are available, our proposed methods will provide us with a more accurate and precise, inexpensive, quantitative means of tumor-volume estimation.