Linear and Nonlinear Elastic Modulus Imaging: An Application to Breast Cancer Diagnosis

• Published in: IEEE transactions on medical imaging 2012-08, Vol.31 (8), p.1628-1637
• Main Authors: Goenezen, S., Dord, J-F, Sink, Z., Barbone, P. E., Jingfeng Jiang, Hall, T. J., Oberai, A. A.
• Format: Article
• Language: English
• Subjects: Breast Neoplasms - diagnostic imaging; Breast Neoplasms - pathology; Breast tumor; Carcinoma, Ductal, Breast - diagnostic imaging; Carcinoma, Ductal, Breast - pathology; Elastic Modulus; Elasticity Imaging Techniques - methods; Female; Fibroadenoma - diagnostic imaging; Fibroadenoma - pathology; Humans; Image Processing, Computer-Assisted - methods; Image reconstruction; Imaging; inverse problem; Lesions; linear and nonlinear elasticity imaging; Linear Models; Nonlinear Dynamics; Signal-To-Noise Ratio; Strain; Stress; Ultrasonic imaging; ultrasound imaging
• ISSN: 0278-0062; 1558-254X
• DOI: 10.1109/TMI.2012.2201497

We reconstruct the in vivo spatial distribution of linear and nonlinear elastic parameters in ten patients with benign (five) and malignant (five) tumors. The mechanical behavior of breast tissue is represented by a modified Veronda-Westmann model with one linear and one nonlinear elastic parameter. The spatial distribution of these elastic parameters is determined by solving an inverse problem within the region of interest (ROI). This inverse problem solution requires the knowledge of the displacement fields at small and large strains. The displacement fields are measured using a free-hand ultrasound strain imaging technique wherein, a linear array ultrasound transducer is positioned on the breast and radio frequency echo signals are recorded within the ROI while the tissue is slowly deformed with the transducer. Incremental displacement fields are determined from successive radio-frequency frames by employing cross-correlation techniques. The rectangular regions of interest were subjectively selected to obtain low noise displacement estimates and therefore were variables that ranged from 346 to 849.6 mm . It is observed that malignant tumors stiffen at a faster rate than benign tumors and based on this criterion nine out of ten tumors were correctly classified as being either benign or malignant.