The rate of re-excision procedures to remove residual tumor left behind after initial resection is high in cancer patients undergoing breast-conserving surgery (BCS) due to the limited intraoperative tools available to the surgeon. The gold standard for BCS specimen margin assessment is histopathological analysis, which is post- operative in nature and can take hours to days to complete. The development of intraoperative margin assessment technology to help the surgeon effectively resect the entire breast tumor during the initial BCS is of paramount importance to reduce patient morbidity, patient disfigurement, and healthcare costs associated with BCS re-excisions. The primary objective of the proposed research project is to further develop a new form of optical imaging, sub-diffuse line scanning (SDLS), to evaluate its value for intraoperative assessment of BCS specimen margins. The first aim is to develop an optimized SDLS system. This aim involves hardware improvements to increase the speed of scanning, enable multi-wavelength acquisition, mitigate specular reflection artifacts, and provide high spatial resolution images. The second aim is to collect SDLS image data of BCS specimens in two phases: first, post-operatively in parallel with standard of care pathological processing, and second, intraoperatively, immediately after excision. In the first phase, line scans will be collected from individual bread loaf slices from each BCS specimen. The image data will be conservatively co-registered with histopathology with guidance from an expert pathologist. In the second phase, line scans of freshly excised, whole specimens will demonstrate the ability to image actual intact BCS specimen margins using SDLS. The third aim is to train a margin classification model using radiomic image features extracted from confirmed tissue subtype regions in the phase 1 bread loaf slice images. The model will be tested on the intraoperative phase 2 images of actual BCS specimen margins. The optimized SDLS system, coupled with the classification model, could eventually be translated into a prospective clinical trial to assess its performance as an intraoperative margin assessment tool. The proposed fellowship training plan involves broadening technical expertise in optical system design, compressive sensing, multivariate data analysis, and machine learning, and includes clinical experiences in breast tissue pathology, radiology, and surgery. Professional development activities include formal presentations to biomedical imaging and clinical audiences, grant writing workshops and R01 grant writing experience, undergraduate mentoring, and teaching. The training plan incorporates publishing peer-reviewed manuscripts from the proposed research. Fellowship sponsors and collaborators are experts with specific expertise in surgery, breast pathology, radiology, biomedical optics, machine learning, and biomedical engineering. The applicant will work closely with the mentori...