Investigating the dynamic activities of protein expression and signaling in living organisms is a crucial focus of intense research aimed at elucidating the processes that underlie disease progression and improving treatments and drug development. Resolving these activities from each individual provides particularly valuable information, as it helps decipher cell heterogeneity and enhances our understanding of numerous biological processes. However, quantifying protein secretion from single cells in real-time remains a challenging task due to the limitations of existing technologies. This is an area where nanophotonics offers promising opportunities, as it allows the detection of minute protein molecules in a label-free manner due to the strong light-matter interaction. Integration with microfluidic technology further enables high-throughput analysis, which is key to enabling potential screening applications. This thesis reports on the development of a high-throughput spectroscopic imaging platform and advanced integrated optofluidic plasmonic biosensor arrays that enable the real-time monitoring of single-cell and single-organoid secretion in a label-free manner. The platform incorporates automated stage scanning and data collection of both spectroscopic and bright-field images, which allows the simultaneous measurement of a hundred individual cells/organoids at a time. The plasmon-mediated extraordinary optical transmission of the gold nanohole arrays enables the ultrasensitive detection of the secreted protein analytes. Additionally, a unique fabrication technique was developed for an open-top microwell membrane with polydimethylsiloxane (PDMS) to isolate single cells for on-chip measurements. For organoids, a special design of two-layer microwell structures was implemented to monitor the organoids while keeping the escaping cells and debris from entering the sensing area and disturbing the signal. The novel platforms have been used to measure a large number of single cells and single organoids under various conditions, yielding a statistical distribution obtained from extracting the kinetic behavior of the signals of Interleukin-2 (IL-2) secretion from EL4 cells and vascular endothelial growth factor A (VEGFA) secretion from various colorectal tumor organoids. The optical images of the tumor organoids have been processed with machine-learning-based image analysis for an automated segmentation and size evaluation, revealing the organoid size reduction in the condition of drug treatment, consistent with previously reported studies. Furthermore, we leveraged the flexibility of PDMS to design microwell arrays with volumes down to ~65 pL to confine fluorescently labelled individual cancer cells and different types of immune cells. By coupling this chip with time lapse fluorescence microscopy and deep neural network algorithm, we studied quantitively cellular interactions in a high-throughput manner for elucidating the cytotoxity of CD4 and CD8 cells. This research has addressed the challenges of quantifying protein secretion from single cells in real-time, offering a label-free approach with nanophotonics. These novel platforms make powerful tools to understand the biology of individual living organisms and paves the way for both fundamental biological studies and clinical translations. We anticipate that such interdisciplinary work will open new avenues in biomedical research.