Semester

Spring

Date of Graduation

2013

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committee Chair

Osama Mukdadi

Committee Co-Chair

Ginger Layne

Committee Member

Victor Mucino

Committee Member

Terence Musho

Committee Member

Nithi T. Sivaneri.

Abstract

Breast cancer comprises about 29% of all types of cancer in women worldwide. This type of cancer caused what is equivalent to 14% of all female deaths due to cancer. Nowadays, tissue biopsy is routinely performed, although about 80% of the performed biopsies yield a benign result. Biopsy is considered the most costly part of breast cancer examination and invasive in nature. To reduce unnecessary biopsy procedures and achieve early diagnosis, ultrasound elastography was proposed.;In this research, tissue displacement fields were estimated using ultrasound waves, and used to infer the elastic properties of tissues. Ultrasound radiofrequency data acquired at consecutive increments of tissue compression were used to compute local tissue strains using a cross correlation method. In vitro and in vivo experiments were conducted on different tissue types to demonstrate the ability to construct 2D and 3D elastography that helps distinguish stiff from soft tissues. Based on the constructed strain volumes, a novel nonlinear classification method for human breast tumors is introduced. Multi-compression elastography imaging is elucidated in this study to differentiate malignant from benign tumors, based on their nonlinear mechanical behavior under compression. A pilot study on ten patients was performed in vivo, and classification results were compared with biopsy diagnosis - the gold standard. Various nonlinear parameters based on different models, were evaluated and compared with two commonly used parameters; relative stiffness and relative tumor size. Moreover, different types of strain components were constructed in 3D for strain imaging, including normal axial, first principal, maximum shear and Von Mises strains. Interactive segmentation algorithms were also evaluated and applied on the constructed volumes, to delineate the stiff tissue by showing its isolated 3D shape.;Elastography 3D imaging results were in good agreement with the biopsy outcomes, where the new classification method showed a degree of discrepancy between benign and malignant tumors better than the commonly used parameters. The results show that the nonlinear parameters were found to be statistically significant with p-value <0.05. Moreover, one parameter; power-law exponent, was highly statistically significant having p-value < 0.001. Additionally, volumetric strain images reconstructed using the maximum shear strains provided an enhanced tumor's boundary from the surrounding soft tissues. This edge enhancement improved the overall segmentation performance, and diminished the boundary leakage effect. 3D segmentation provided an additional reliable means to determine the tumor's size by estimating its volume.;In summary, the proposed elastographic techniques can help predetermine the tumor's type, shape and size that are considered key features helping the physician to decide the sort and extent of the treatment. The methods can also be extended to diagnose other types of tumors, such as prostate and cervical tumors. This research is aimed toward the development of a novel 'virtual biopsy' method that may reduce the number of unnecessary painful biopsies, and diminish the increasingly risk of cancer.

Share

COinS