Date of Award
Spring 2022
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Biomedical Engineering
First Advisor
Yu, Bing
Second Advisor
Joshi, Amit
Third Advisor
Gilat-Schmidt, Taly
Abstract
Positive margins after breast-conserving surgery (BCS) or lumpectomy may lead to a twice higher risk of breast cancer recurrence. Additional surgery, which is therefore recommended to women with positive margins, is associated with significant psychological, cosmetic, and financial burdens for patients and their caregivers. Since the release of the 2014 SSO-ASTRO guidelines, the odds of re-excision have declined significantly but remain in the range of 14-18% and vary substantially among surgeons. Because existing intraoperative margin assessment tools are not efficient enough or routinely used for reasons such as time and labor cost, the definitive pathologic margin status is not available until several days after surgery provided by histopathology. Therefore, a device with large tissue coverage, microscopic resolution, and fast speed is highly desired for assessing entire surgical margins rapidly during the lumpectomy, which is expected to reduce the re-excision rates. However, no emerging technology in development has demonstrated this capability.In this research, a deep-ultraviolet scanning fluorescence microscope (DUV-FSM) imaging system has been developed and evaluated for rapid and accurate detection of cancer cells on the surface of fresh, unprocessed breast tissues. The DUV-FSM has achieved a spatial resolution of ~3 µm and an imaging speed of 1.0 min /cm2. A total of 66 fresh human breast tissues were imaged and the fluorescence images showed excellent visual contrast in color, tissue texture, cell density and morphology between invasive carcinomas and their normal counterparts (such as adipose and stroma). Visual interpretation of the fluorescence images by non-medical evaluators distinguished invasive carcinoma from normal samples with excellent sensitivity (97.62%) and specificity (92.86%). Using regional-defined nuclear-cytoplasmic ratio alone was able to differentiate patch-level (2 x 2 mm2) invasive carcinoma from normal breast tissues with reasonable accuracy. Automated tumor/normal binary classification model based on texture and other features achieved high patch-level (0.5 x 0.5 mm2) accuracy (89.8%) and sample-level accuracy (91.5%). To further accelerate the imaging speed, a dual-scanning DUV-FSM that is capable of simultaneously surveying two surfaces of a lumpectomy specimen is proposed and some preliminary results are presented. The DUV-FSM has shown great potential to be utilized for intraoperative breast margin assessment.