Biophotonics & Optical Bioimaging
Our research is dedicated to developing novel optical technologies for imaging of living systems to improve understanding, diagnosis, and intervention of cancer and neuro-oncology diseases.
Stimulated Raman Imaging for Label-Free Pathology and Image-Guided Neurosurgery
Stimulated Raman scattering (SRS) microscopy is a hybrid technology between multiphoton and Raman scattering. SRS microscopy can do rapid, high-resolution Raman imaging of the chemical bonds in the native biomolecules. The stain-free biomolecular contrasts of SRS imaging enables numerous biomedical applications. In particular, two-color SRS imaging of fresh brain tumor tissue with contrasts of proteins at 2940 cm-1 (attributed to the CH3 chemical bond; pseudocolor blue) and lipids at 2854 cm-1 (attributed to the CH2 chemical bond; green) has been used to render stain-free digital neuropathology for rapid diagnosis. This methodology can potentially replace intraoperative histopathology consultation, which requires long turn-back time and intense labor from the pathology division.
The essential diagnostic feature in oncologic pathology is the cancer cell nuclei. In stain-free SRS pathology images, the contrast of cell nuclei is lower than in H&E stained images. To achieve rapid segmentation and quantification of cell nuclei from the heterogeneous tissue background, collaborating with computer and data science experts, we are using deep neural networks (DNN) to analyze SRS images of cancer cells. We developed a two-photon fluorescence imaging-based automated labeling approach to provide large-scale labeled data as the ground truth for the training of U-NET and MASK RCNN networks. The trained machine learning models are able to rapidly identify cell nuclei from two-color SRS images with high accuracy. The AI-powered SRS rapid fresh tissue digital pathology will be eventually used for image-guided brain tumor surgery in the operating room.
Live-Cell Imaging with SRS and Multiphoton Microscopy for Cancer Lipid Metabolism Studies
We are developing live-cell imaging technologies for long-term, time-lapse tracking of lipid droplets (LDs) in cancer cells using SRS & multiphoton microscopy. Abnormal lipid metabolism is a newly identified hallmark of cancer, such as gliomas (GBM) or ovarian cancer. Stain-free imaging of the native LDs in live cells with SRS has a few unique advantages over fluorescence-stained imaging. Unlike fluorescence, SRS imaging has no photobleaching issue. In addition, SRS can image all newly synthesized LDs, which may not carry any fluorescence staining. More importantly, hyperspectral SRS imaging can decode the biochemical composition of the LDs, which cannot be achieved with two-photon fluorescence imaging. Imaging the quantification and dynamics of LDs in cancer cells may help discover new anti-cancer therapeutics.
Imaging Drug Delivery Using SRS Microscopy with Raman Tags
Collaborating with pharmaceutical experts, we are developing novel Raman tags for SRS imaging of drug delivery with improved sensitivity. As a proof-of-principle study, we synthesized a non-fluorescent lysosome tracker with a bright Raman tag. Overlay imaging of stain-free lipid droplets (LDs) and lysosomes with the Raman Lyso-tracker reveals active interaction between the two cellular organelles. Current work focuses on imaging the transmembrane trafficking and the dynamics of Raman-tagged drug molecules in live cancer cells.