High positive margin rates in oncologic breast-conserving surgery are a pressing clinical problem. Volumetric X-ray scanning is emerging as a powerful ex vivo specimen imaging technique for analyzing resection margins, but X-rays lack contrast between non-malignant and malignant fibrous tissues. In this study, combined micro-CT and wide-field optical image radiomics were developed to classify malignancy in breast cancer tissues. Results indicate that this form of multimodal radiomic analysis improves malignancy classification with statistical significance relative to using either X-ray or optical data alone.

We used Diffuse Reflectance spectroscopy (DRS) and Raman spectroscopy to investigate functional differences between murine breast tumor xenografts of varying metastatic potential. Once tumors volume reached 200mm3, in-vivo DRS and ex-vivo Raman spectroscopy were performed. Spectra and optical properties were used to train and test a leave-one-out random forest classifier. Significant differences between metastatic and non-metastatic tumors were observed. The study shows that random forest classifiers coupled with optical spectroscopy provides consistent predictions of metastatic phenotype and metastatic abilities from the primary tumor that can be translated into clinic.

An integrated clinical two photon fluorescence microscope system allows for real-time assessment of freshly excised non-melanoma skin cancer skin biopsies with 2 minutes of preparation and enables imaging of multi-centimeter lesions in under 5 minutes. This system simulates the conventional workflow of a brightfield microscope to minimize pathologist retraining. A blinded study comparing two photon images and coregistered H&E paraffin section images is performed to show degree of concordance between the two modalities.

Our group and others have demonstrated the strong potential of the multiphoton microscopy for a broad range of applications from advancing the understanding of skin biology to non-invasive diagnosis of skin diseases and monitoring therapy effects. We have recently reported on a fast large area multiphoton exoscope for rapidly mapping out macroscopic tissue areas with microscopic resolution and enhanced contrast for selective melanin detection. We will describe the technical abilities of this instrument and demonstrate its feasibility for early melanoma diagnosis based on a pilot study on ex-vivo and in-vivo imaging of pigmented lesions suspicious of melanoma in human skin.

An experimental polarization enhanced laparoscopy (PEL) imaging system was developed to improve the visualization of peritoneal cancer lesions compared to conventional white light laparoscopy (WLL). The design modifications provide sensitivity to backscatter depolarization in tissue. Phantom studies demonstrated the sensitivity of PEL to altered scattering cross section and collagen organization. Implementation of the PEL for biopsy tissue study illustrated the feasibility and potential of PEL to improve the contrast between malignant lesions, and background tissue based on differences in their depolarization properties.

Spatial Frequency Domain Imaging (SFDI) can provide longitudinal label-free, wide field hemodynamic and scattering information of murine tumors. Here we present a longitudinal study utilizing SFDI to monitor a paired immune responsive-resistant model for up to 30 days of treatment. Mice receiving the immunomodulatory treatment had a large increase in the reduced optical scattering throughout treatment compared to the mice receiving an immune-blocking antibody. These results indicate that scattering is sensitive to the immune-mediated apoptosis of tumor cells and capable of discriminating between responsive and resistant tumor models.

Two-photon microscopy was used to investigate responder vs non-responder NSCLC tumors. We received twenty-four NSCLC specimens, resected before therapy, and classified them as responder and non-responder according to subject-detail reports. Two-photon microscopy was used to determine an optical redox ratio and mean NADH lifetime data for each sample. The optical redox ratio revealed a decrease in the responder group when compared to the non-responder tumors, whereas the responder tumor group presented a longer mean NADH lifetime when compared to non-responder group. Our current results point to the potential of metabolic imaging in providing information on treatment response of tumors prior to receiving therapy.

Sculpting light in 3D at high speeds is critical to track and manipulate biological events at cellular scales in real-time. Yet, even the current state of the art, Computer-Generated Holography (CGH), operates with slow algorithms and does not provide enough degrees of control to focus light precisely through deep biological tissue. We address these two challenges with new hardware and algorithms. First, we introduce DeepCGH, a deep learning model that synthesizes 3D holograms in milliseconds, and we demonstrate experimental benefits in multiphoton microscopy applications. We then present new optical instruments that modulate light both in space and time to render 4D light fields with much greater fidelity than coherent CGH techniques.

In hyperspectral imaging, surface reflections and height differences within tissue samples add variations to spectra which are not related to tissue composition. To improve diagnostic accuracy, several pre-processing techniques are used to reduce these variations. However, currently a framework is lacking to choose an optimal pre-processing algorithm technique for a clinical application. We identified 8 pre-processing algorithms and investigated on synthetic data how well each algorithm reduces variations related to surface reflections and height differences and keeps variations between spectra related to differences in tissue optical properties. We demonstrate the use of our framework on colon and breast cancer tissue.

Here, we report on the development of a wide field Raman microscope which significantly improves the speed of acquisition at selected spectral range with spatial resolution in the range of ~200 nm over large field of view. This is achieved by analyzing small fluctuations in a large time-series of Raman images with a stochastic optical reconstruction microscopy (STORM) protocol in order to localize the molecules. We demonstrate the potential of this microscope by analyzing distinct samples such of patterned Silicon, polystyrene microspheres on Silicon wafer and graphene on Silicon/Silicon dioxide substrate.

Breast cancer is a highly heterogenous disease, both phenotypically and genetically. Here, we propose that the spatial context of organelles, specifically their subcellular location and inter-organelle relationships (topology), can be used to inform breast cancer cell classification. Thus, Organelle-Topology-Cell-Classification-Pipeline (OTCCP) was introduced to quantify the topological features of subcellular organelles, remove the bias of visual interpretation, and classify different breast cancer cell lines using a machine learning method. Our goal is to investigate the role of 3D cancer cell growth on the heterogeneity of organelle topology and morphology to increase the understanding of cancer biology on a subcellular level.

Frequency-domain near-infrared spectroscopy (fdNIRS) has been shown to be a promising tool for the diagnosis and monitoring of breast cancer treatment in point-of-care settings. However, current fdNIRS embodiments suffer from poor scalability and high complexity that has slowed their clinical translation. For the first time, we present a handheld, fully-wireless, multi-detector, multi-wavelength, fdNIRS system capable of real-time quantitative noninvasive measurements of optical properties and tissue chromophore concentrations at >10 kHz. High spatial resolution 2D topography images are displayed in real-time on a mobile platform with motion tracking. We characterize the system against prior generations, as well as in-vivo performance in human subjects.

Neural activity patterns span over three orders-of-magnitude in both space and time. This presentation introduces a new versatile all-optical technique for neurophysiology that utilizes a novel label-free optical microscopy system called Superfast Polarization-sensitive Off-axis Full-field Optical Coherence Microscopy (SPoOF OCM). SPoOF OCM captures the changes to both the optical path length and birefringence induced by the millisecond-scale electrical activity of neurons. This multimodal response is imaged in a widefield configuration at 4000 frames per second, with a field-of-view of 200×200 µm^2, 1 µm transverse resolution, and 4.5 µm axial resolution. With an ability to capture responses spanning three orders-of-magnitude in both space and time, SPoOF OCM meets the exacting needs of a comprehensive neurophysiology tool, and overcomes existing limitations of traditional electrophysiology and fluorescence microscopy.

Laser speckle contrast imaging (LSCI) illuminates continuous-wave (CW) laser light on tissue surface. We assembled an integrated LSCI system combining a CW laser at 785 nm and a picosecond pulsed laser at 775 nm. A CMOS camera collected images from mouse head with intact skull. The pulsed laser with engineered diffuser captured more details of brain vessels compared to the CW laser with glass diffusers. The consecutive ligations of left and right common carotid arteries resulted in significant CBF reductions. This research lays the ground to develop multimodal imaging systems integrating LSCI and other imaging techniques with shared pulse illuminations.

Regular blood lipid screening is recommended for early diagnosis of cardiovascular diseases. We present a portable LED based SFDI system as a new clinical tool for non-invasive blood lipid monitoring. The new SFDI system was used to measure changes in optical properties within subcutaneous blood vessels on subjects’ dorsal hands after consumption of a meal. Superficial blood vessels were segmented and a two-layer inverse model was used to incorporate the effects of overlying skin. The results show an increase in μ_s^' and a decrease in μ_a at around 3h after the meal, followed by a gradual return to baseline values.

Diffuse optical spectroscopies are non-invasive, light-based tools that enable real-time bedside monitoring of microvascular hemodynamics. In this talk we highlight the utility of these tools to identify biomarkers of stroke risk in two high-risk patient populations; subarachnoid hemorrhage and sickle cell disease.

In vivo multiphoton microscopy (MPM) provides non-invasive imaging of living tissues in their native state. Melanin, a fluorophore found in the skin, plays a significant role in various physiological and pathological processes. Quantification of cutaneous melanin provides basis for efficient treatment of pigmentary skin disorders and for differentiating melanoma from benign pigmented lesions. MPM imaging has a potential for non-invasive cutaneous melanin evaluation. We recently reported on the development of a fast large area multiphoton exoscope (FLAME) for in vivo macroscopic imaging with microscopic resolution and enhanced molecular contrast for selective melanin detection. In this work, we demonstrate the benefit of sampling a large volume, using FLAME in vivo, for enhancing the accuracy of the melanin content assessment in human skin.

Second-harmonic generation microscopy and macro-scale imaging were combined to enable multiscale assessments of mouse skin undergoing uniaxial mechanical testing. Skin from old and young mice experienced a substantial micro-scale volume reduction during uniaxial tension. A non-affine relationship between the 3D collagen fiber kinematics and local deformation was also observed. Aged skin was found to have a lower stiffness but increased collagen fiber realignment during mechanical loading. These results are being used to develop multiscale models of skin mechanics and obtain a more complete understanding of age-related changes in skin structure-function relationships.