HYPERSPECTRAL STIMULATED RAMAN MICROSCOPY FOR THERANOSTICS OF CANCEROUS CELLS AND TISSUE

Abstract

Cancer is a leading cause of human death globally among other diseases. Early cancer detection accompanied with effective treatment is the key to improving the survival rates of cancer patients. To formulate precise and effective cancer treatment strategies for patients and minimize the potential adverse effects during photo-chemical therapies, it is essential to spatiotemporally monitor the therapeutic effect of alternative drugs and screen the optimized dosage and administration time in tissue and cells. Particularly, glioblastoma (GBM) is the most common and aggressive adult brain tumor. A significant characteristic of GBM is the high degree of intratumoral molecular heterogeneity, resulting in postoperative drug resistance and tumor recurrence. At present, single-cell RNA sequencing is the gold-standard method for molecular subtyping (classical (CL), mesenchymal (MES), and proneural (PN)) and heterogeneity assessment of GBM tissues, but not suitable for intraoperative diagnosis due to lengthy and tedious sample preparation and analysis while still lack of spatial information. Therefore, more advanced rapid imaging technologies need to be developed to provide rapid diagnosis and therapeutic guidance in clinical settings. Hyperspectral stimulated Raman scattering (SRS) microscopy is a vibrational spectroscopic imaging technique which has emerged as an appealing tool for label-free tissue and cells imaging with high biochemical specificity and sensitivity at the subcellular level. SRS has found applications in biomedical study, such as stain-free histopathology, pharmacokinetics, and cell metabolism. But the strong scattering of turbid media (e.g., tissue) limits the imaging depth, hindering its further investigations in thick and highly scattering biomedical tissues. To tackle the challenges, the thesis work aims to: (i) develop a rapid hyperspectral SRS imaging platform for biomedical diagnosis of GBM tissue; (ii) develop an optimized photodynamic therapy (PDT) strategy through real-time monitoring of treatment efficacy of cancerous cells with SRS imaging, and (iii) propose a new SRS imaging scheme by using the non-diffracting beams to enhance the imaging depth in highly scattering media. More specifically, we have developed a machine learning-based hyperspectral SRS imaging diagnostic platform for label-free histopathological and spatially-resolved molecular subtyping of GBM tissue. The biochemical distribution mapping of proteins and lipids by hyperspectral SRS imaging unveils crucial histopathological characteristics (e.g., vascular proliferation, demyelination etc.). We found that cell density and myelin integrity show a downward trend in the order of PN-CL-MES, while vascular proliferation is prominent in MES tissues. SRS spectral information was demonstrated to produce significant diagnostic performance for GBM subtypes (diagnostic accuracy rate is 86.3%) to generate the spatial mapping of GBM molecular subtypes depicting the intratumoral heterogeneity in GBM tissue within one hour, which is unprecedented with conventional RNA sequencing technique. We have also combined SRS imaging with transient absorption (TA) microscopy for real-time monitoring of cancerous cells response to mitochondria-targeting up-conversion nanoparticles (UCNPs) during photo-switchable PDT. The time-lapse hyperspectral SRS images of HeLa cells visualized crucial dynamic cell morphological alterations (e.g., membrane blebbing, cytoplasmic contraction, and nuclear volume reduction, etc.), testifying to the cytotoxic effect of PDT. Meanwhile, the multivariate curve resolution (MCR) analysis on SRS imaging shows intracellular compositional variations as well as the diffusion and concentration reduction of intracellular UCNPs, which may be related to elevated mitochondrial fission activities. Furthermore, in the high-wavenumber region (2800–3200 cm-1), a 20-wavenumber red shift in the cell nuclear spectrum was observed after 20 minutes of treatment, demonstrating that the oxidative DNA damage occurred in cell nucleus at a specific time point. Further, we also apply the scalar diffraction theory to investigate the light propagation of non-diffracting beams (e.g., Airy beam, Bessel beam) in highly scattering media. The simulation results reveal that in optical systems with NA of 0.1 and 0.4, the penetration depth of the Airy beam is respectively 1.50-1.87 and 1.35-1.54 times that of the Bessel beam, suggesting the potential of Airy beam-based SRS imaging technique for deeper tissue imaging. In summary, the unique hyperspectral SRS imaging diagnostic platform developed in this work shows a great potential to facilitate the translational research for improving cancer diagnosis and treatment efficacy assessment in biological and biomedical systems.