Artificial Intelligence in Label-Free Optical Imaging Applications

Optical imaging has been indispensable for research in biology. Especially, label-free optical imaging systems are essential in both fundamental and advanced biomedical applications due to their exceptional spatial resolution and enhanced imaging contrast. They leverage intrinsic sample properties for visualization, offering real-time imaging for morphological and molecular information without the use of exogenous contrast agents, which may be hazardous or interfere with normal tissue/sample behavior. For instance, simultaneous label-free autofluorescence multiharmonic (SLAM) microscopy is a nonlinear optical imaging technique using a single excitation source (high-intensity pulsed laser) to simultaneously acquire four different channels: two-photon excited fluorescence (2PEF), three-photon excited fluorescence (3PEF), second harmonic generation (SHG), and third harmonic generation (THG). These four channels can be used to observe microstructures with molecular and functional information, which is highly beneficial for cancer diagnosis. Fluorescence lifetime imaging microscopy (FLIM) is another label-free imaging technique that can be performed with nonlinear two- or three-photon excitation and characterizes endogenous tissue fluorophores based on the time between excitation and de-excitation. It offers comprehensive, environment-sensitive information regarding molecular interaction in biological samples. Therefore, this facilitates the evaluation of cellular conditions, drug effects, protein interactions, and disease progression, which is particularly suitable for biopharmaceutical applications. Apart from the label-free nonlinear optical imaging systems, optical coherence tomography (OCT) is another useful label-free imaging modality based on low-coherence interferometry, providing depth-resolved cross-sectional images with rapid image acquisition. As a result, OCT has been employed in a wide range of clinical applications. In recent years, rapid advancements in artificial intelligence (AI) have significantly altered data analysis and specifically in biomedical imaging. AI technologies have been established as essential tools for extracting meaningful insights from complex data and are applicable across diverse scales (micro to macro) and contrast mechanisms (fluorescence and refractive index). This chapter discusses label-free optical imaging applications with AI-assisted approaches, specifically in cancer diagnosis, cell line selection for biopharmaceuticals, and clinical ear infection diagnosis using SLAM, multimodal nonlinear imaging (SLAM and FLIM), and OCT, respectively.