Date of Graduation

8-2024

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering (MSEE)

Degree Level

Graduate

Department

Electrical Engineering

Advisor/Mentor

El-Shenawee, Magda

Committee Member

Ware, Morgan E.

Second Committee Member

Arnold, Mark

Keywords

Breast cancer; Mueller matrix; Terahertz imaging; Terahertz polarimetry imaging

Abstract

Breast-conserving therapy (lumpectomy) is one of the most performed breast cancer surgeries in the United States. The best outcome of lumpectomy surgery is achieved when the surgical margins are free of cancer. When remnants of cancer are detected at the surgical margins after the initial operation, a second operation will be required to remove the cancer. Unfortunately, a significant number of patients undergo breast conserving surgery (BCS) at local hospitals that do not have access to immediate on-site pathology leading to high rates of re-excision or reoperation (greater than 30%). Therefore, there is a significant need for new intraoperative technology that can be made available for local hospitals and outpatient clinics. Our previous research concludes that while terahertz studies in pre-clinical models have shown strong differentiation between cancerous and fatty tissues, the more clinically relevant differentiation between cancerous and healthy non-fatty tissue remains challenging. To further build upon the successes of our previous project and improve the sensitivity of terahertz imaging cancer detection on the surgical margins, we have identified areas where we can significantly improve the instrumentation. As part of this work, we have re-designed our time-domain terahertz spectroscopy system to develop terahertz polarization sensitive imaging methodology. In this new approach, all four polarizations of the waves are incorporated to increase the spatial and spectra information about different types of tumors tissues. We tested the system on bovine tissue as a biological model to determine the sources of signal generation in the THz images. Finally, we investigated the polarimetry images of human breast cancer surgical specimens and compared co- and cross polarized images. Furthermore, we presented the images in 16 Mueller matrix images. The goal is to better detect the presence of healthy fibrous tissue due to their potential growth of healthy collagen adjacent to cancerous tissues in tumors. In this work, we focused on the co- and cross polarized images presented by vertical-vertical, vertical-horizontal, horizontal-vertical and horizontal -horizontal received and emitted signals. This work allowed us to develop an approach that leverages multiple polarizations. All terahertz images were labeled using the ground truth pathology images. The obtained results demonstrated that co-polarized images of vertical-vertical and horizontal-horizontal polarization provide more features in the image. The obtained results also showed that cross-polarization images are dependent on the interaction between the direction of the electric field and the orientation of the tissue sample with respect to the electric field.

Available for download on Friday, September 12, 2025

Included in

Biomedical Commons

Share

COinS