Multi-modal optical methods for intra-procedural evaluation of breast specimens

Thesis (Ph.D.)--University of Rochester. Institute of Optics, 2012.Intrasurgical evaluation breast lumpectomy margin status is a challenging but critical job for pathologists, especially for non-palpable lesions. Positive or close margins influence the course of treatment and directly impact patient prognosis. If the margins are close or positive, a second surgery may be necessary to remove any residual cancer, which is costly for the hospital and emotionally and physically taxing for the patient. Re-excision rates nationally are ~40%, demonstrating the impact a tool for rapid evaluation of margin status can have. Microscopic examination of unprocessed breast tissue at the time of the initial surgery would provide pathologists with the ability to fully assess margin status. Optical techniques such as reflectance confocal microscopy (RCM) can provide sub-cellular resolution images of unprocessed breast tissue but have image acquisition times too long for clinical use. This research demonstrates a suite of tools designed to aid pathologists in the evaluation of breast lumpectomy specimens with the goal of reducing the re-excision rate. The first chapter presents the current pathological protocols for the diagnosis and management of breast disease along with an overview of the use of optics in breast cancer detection, diagnosis, and margin evaluation. This chapter clearly outlines the challenges pathologists face when trying to evaluate the margin status of non-palpable lesions. At the end of the chapter, the specific aims of the research are presented. The second chapter discusses mechanical stabilization of the breast specimen and guidewire localization. The mechanical stabilization, achieved through the use of UV curing gelatin, maintains the geometry of the system throughout handling and assists in the initial dissection of the specimen. Two forms of an optical beacon are presented: one based on irradiance values from an embedded fiber optic measured at the surface of the tissue, and one based on the generation of an acoustic pulse by an optical pulse delivered through a fiber optic guidewire. These systems are capable of locating the end of the guidewire to within ± 5 mm and aid the pathologists in initial sectioning by indicating the location of the stereographically placed guidewire. Chapters 3-5 present methods of optical segmentation of the tissue to facilitate microscopic examination. Optical signatures such as color signatures, birefringence, and autofluorescence are used to determine regions of fat and collagen at the tissue surface. Color signatures of fat are used to exclude it from microscopic examination, as fat does not become cancer. Collagen structures are located through the use of intrinsic birefringence and the morphological appearance of the collagen structures can be used by a skilled clinician to guide microscopic imaging to regions likely to contain cancer. Chapter 3 presents the development of the segmentation algorithms along with preliminary tissue trials. Through testing on two systems (a large field of view system made from a modified flatbed scanner in Chapter 4, and a small field of view macro imager coupled to an RCM in Chapter 5), sensitivity and specificity values of up to 80% and 70%, respectively, for fat segmentation. Sensitivity and specificity values up to 60% and 60% were achieved for collagen segmentation. In addition, using the morphological presentation of the tissue, regions of diagnostic significance were correctly identified in every case