Broad and safe access to ultrafast laser technology has been hindered by the absence of optical fiber-delivered pulses with tunable central wavelength, pulse repetition rate, and pulse width in the picosecond–femtosecond regime. To address this long-standing obstacle, we developed a reliable accessory for femtosecond ytterbium fiber chirped pulse amplifiers, termed a fiber-optic nonlinear wavelength converter (FNWC), as an adaptive optical source for the emergent field of femtosecond biophotonics. This accessory empowers the fixed-wavelength laser to produce fiber-delivered ∼20 nJ pulses with central wavelength across 950 to 1150 nm, repetition rate across 1 to 10 MHz, and pulse width across 40 to 400 fs, with a long-term stability of >2000 h. As a prototypical label-free application in biology and medicine, we demonstrate the utility of FNWC in real-time intravital imaging synergistically integrated with modern machine learning and large-scale fluorescence lifetime imaging microscopy.
The biopharmaceutical industry relies on selecting high-performing cell lines to meet quality and manufacturability criteria. However, this process is time- and labor-intensive. To address this, label-free multimodal multiphoton microscopy techniques were employed to characterize biopharmaceutical cell lines in early passages. Using a machine learning-assisted single-cell analysis pipeline, over 95% accuracy for monoclonal cell line classification was achieved in all passages. Additionally, Open Set Recognition allowed the differentiation of desired cell lines in polyclonal pools. The study offers a promising solution to expedite the cell line selection process, reducing time and resources while ensuring the identification of high-performance biopharmaceutical cell lines.
Understanding drug fingerprints in complex biological samples is essential for drug development. We demonstrate a deep learning-assisted hyperspectral coherent anti-Stokes Raman scattering (HS-CARS) imaging approach for identifying drug fingerprints at single-cell resolution. The attention-based deep neural network, Hyperspectral Attention Net (HAN), highlights informative spatial and spectral regions in a weakly supervised manner. Using this approach, drug fingerprints of a hepatitis B virus therapy in murine liver tissues was investigated. Higher classification accuracy was observed with increasing drug dosage, reaching an average AUC of 0.942. Results demonstrate the potential for label-free profiling and localization of drug fingerprints in complex biological samples.
Efficient cell line development is crucial for optimizing biopharmaceutical production. We demonstrate the potential of SLAM and FLIM microscopy to optimize this process by correlating metabolism-related features with measured productivity in early CHO cell passages. Eight CHO cell lines were imaged using SLAM and FLIM microscopy, and a pipeline was developed to classify the cells. A linear SVM achieved 95% accuracy in predicting productivity. Important features and their channel affiliations were identified, revealing optical metabolic characteristics from NAD(P)H and FAD associated with productivity. SLAM features correlated with growth and viability, while FLIM features correlated with protein production, highlighting the importance of multimodal label-free imaging.
Fluorescence lifetime imaging microscopy (FLIM) provides valuable insights into molecular interactions and states in complex cellular environments. Conventional FLIM analysis methods struggle with accurate lifetime estimation with low photons-per-pixel (PPP). We propose DeepFLR, a self-supervised deep learning framework for robust FLIM signal restoration with limited photons. By exploiting the spatiotemporal dependencies of FLIM signals, DeepFLR reconstructs the fluorescence decay curves, leading to accurate lifetime estimations using existing lifetime estimation methods. The results demonstrate that DeepFLR enables reliable lifetime estimation with less than 10 PPP for a diverse set of biological samples. The proposed approach significantly reduces the photon budget of FLIM and opens up numerous low-light FLIM applications.
Quality control in molecular optical sectioning microscopy is indispensable for transforming acquired digital images from qualitative descriptions to quantitative data. Although numerous tools, metrics, and phantoms have been developed, accurate quantitative comparisons of data from different microscopy systems with diverse acquisition conditions remains a challenge. Here, we develop a simple tool based on an absolute measurement of bulk fluorophore solutions with related Poisson photon statistics, to overcome this obstacle. Demonstrated in a prototypical multiphoton microscope, our tool unifies the unit of pixelated measurement to enable objective comparison of imaging performance across different modalities, microscopes, components/settings, and molecular targets. The application of this tool in live specimens identifies an attractive methodology for quantitative imaging, which rapidly acquires low signal-to-noise frames with either gentle illumination or low-concentration fluorescence labeling.
KEYWORDS: Optical coherence tomography, Biological research, Ear, In vitro testing, Image analysis, Principal component analysis, In vivo imaging, Feature extraction, Binary data
Otitis media (OM) is a prevalent disease among children worldwide. Antibiotic-resistant bacterial biofilms can develop in the middle ear during recurrent/chronic ear infections. OCT was used to compare microstructural texture features from primary bacterial biofilms in vitro. From 1200 ROI images of each biofilm class, 934 texture features were extracted. Principle component analysis and five-fold cross-validation were performed using Support vector machines (SVMs). Currently, the model has achieved 0.97 AUC (cubic kernel function) and an average classification accuracy of 89%. Texture analysis of bacterial biofilm OCT images with SVM may enable real-time visualization and differentiation of OM-causing bacterial biofilms in vivo.
In the production of biotherapeutics, Chinese hamster ovary (CHO) cells are known as the gold standard. One challenge in the development of these cell lines is the identification of high expressing, yet stable CHO cells. Here we apply simultaneous label-free autofluorescence multi-harmonic (SLAM) microscopy to four CHO cell lines of varying levels of productivity and stability. With the assistance of machine learning, we were able to classify the CHO cell lines into their respective categories with an accuracy of 85%. Application of this CHO cell characterization technology to upstream bioprocessing can potentially improve workflows such as high-throughput screening and monitoring.
Label-free multimodal optical bioimaging allows non-perturbative profiling of biological samples based on their intrinsic optical molecular properties. In this study, we utilized SLAM and FLIM microscopy to identify CHO cell lines with favorable process performance for the production of therapeutic monoclonal antibodies and proteins. Here, a single-cell analysis pipeline was developed to quantitatively characterize CHO cell lines based on their phenotypes. To perceive the rich information in the multi-modal bioimages, a custom-built multi-task deep neural network was built, which can extract features from different aspects of the optical and molecular properties of the sample. This work demonstrated the potential of ML-assisted multi-modal optical imaging in the identification of cell lines with desirable characteristics for biopharmaceutical production at earlier time points.
We present a weakly-supervised deep learning framework for human breast cancer-related optical biomarker discovery based on label-free autofluorescence multiharmonic (SLAM) microscopy. This framework consists of three stages: self-supervised consistency training for image representation learning at multiple scales; cancer region identification by weakly-supervised Multiple Instance Learning (MIL); optical biomarker discovery based on channel-wise attribution maps. Currently, the model has achieved an average AUC of 0.86 on the breast cancer global detection task. The attribution maps on different scales highlight distinct structures in SLAM which facilitate new insights into tumor micro-environment and field cancerization.
Self-amplifying mRNA (SAM), a synthetic RNA vaccine which self-replicates upon delivery into the cytoplasm encapsulated with lipid nanoparticles (LNPs), leads to a strong and sustained immune response. In this study, we investigated SAM-LNP uptake and subsequent SAM release and distribution in baby hamster kidney (BHK-21) cells using coherent anti-Stokes Raman scattering (CARS) and multiphoton imaging techniques. This work demonstrates the significance of multimodal imaging techniques to capture the successful delivery of SAM and the subsequent production of proteins within cells. Our study can be further extended to label-free detection techniques to investigate targeted drug-delivery.
Chinese hamster ovary (CHO) cells are the most widely used cell line for the recombinant expression of human therapeutics. To investigate a select cell line monoclonal antibody production, we monitor NAD(P)H, a crucial enzymatic cofactor, and an auto-fluorescent bio-marker, with two-photon fluorescence lifetime imaging microscopy (2P-FLIM). This represents a high-resolution, label-free technique for longitudinally characterizing a changing environment (if any) during metabolic transitions. 2P-FLIM analysis of NAD(P)H in four different CHO cell lines helps us predict productive cell types from others. A detailed single cell analysis is also presented that can separate cell types based on optical and morphological classification.
The primary goal of this study was to track PS-ASO and GalNAc-PS-ASO uptake in two cell cultures as the first step to understand the observations from the clinical studies. The multimodal imaging setup of CARS and 2PF modalities in conjunction with the image analysis pipeline made it uniquely possible to address these challenges. We report here the time-dependent uptake, internalization, and localization differences between GalNAc-PS-ASOs and PS-ASOs in liver cells. We believe our findings will help us form the basis for further investigations with more complex cellular co-cultures and with tissue and animal models.
Simultaneous label-free autofluorescence multiharmonic (SLAM) microscopy produces multimodal high-resolution images with rich functional and structural information from living tissue. Here we present a deep learning framework for human breast cancer-related optical biomarker discovery based on SLAM. This framework consists of three stages: self-supervised consistency training for image representation learning at multiple scales; cancer region identification by Multiple Instance Learning; optical biomarker discovery based on channel-wise attribution maps. This study demonstrates the capability of AI-assisted SLAM microscopy in capturing rich information from living tissue and extracting relationships between optical features with human breast cancer, which can be extended to various types of diseases and treatment conditions.
Liver-on-a-chip is a 3D in vitro hepatic microphysiological system aiming to recreate the conditions of liver tissue on a microscopic scale. CN Bio microphysiological system (CN Bio Innovations, UK) is one of the advanced liver-on-a-chip models. In this study, a multimodal optical imaging platform incorporating nonlinear optical imaging techniques such as multiphoton microscopy (MPM), fluorescence lifetime imaging microscopy (FLIM), coherent anti-Stokes Raman scattering (CARS) microscopy, and simultaneous label-free autofluorescence multiharmonic (SLAM) microscopy was used for characterizing the structural and functional changes associated with inflammation, lipid accumulation and drug uptake in the CNBio liver-on-a-chip model.
An efficient and automated image analysis pipeline is essential for extracting quantitative information from multimodal image datasets. In this study, a multimodal optical imaging platform was used to capture CARS, 2PF, and FLIM images from control and drug-treated cells. Images were collected using both fluorescent label-based and label-free approaches. Here we present a single-cell analysis pipeline for the multimodal cellular image analysis. The results demonstrate the capability of our single-cell analysis pipeline for quantitatively measuring the intracellular drug distribution and its longitudinal uptake using a multimodal optical imaging platform, which can provide novel insights into the uptake pathways and target-sites.
Recent advances in tissue engineering and microfabrication have led to development of novel Complex In Vitro models (CIVMs) that more closely mimic pathophysiological functions of human tissues and organs. CIVMs can provide deeper insights into the mechanisms of human disease and pharmacological properties of new drug candidates during early stages of development. In this study, a multimodal optical imaging platform was used for characterizing the structural and functional features of a liver-on-a-chip model (CN Bio Innovations, UK).
Antisense oligonucleotides (ASOs), a novel paradigm in modern therapeutics, modulate cellular gene expression by binding to complementary RNA sequences. Triantennary N-acetyl galactosamine (GN)-conjugated ASOs show greatly improved potency via Asialoglycoprotein receptor (ASGR)-mediated uptake in hepatocytes. Here, we compare the uptake kinetics and subsequent distribution of untargeted ASOs to that of GN-ASOs in mouse macrophages and hepatocytes using simultaneous coherent anti-Stokes Raman scattering (CARS) and two-photon excited fluorescence imaging. While the CARS modality captured the changing lipid distributions and overall morphology of the cell, two-photon fluorescence imaging measured the uptake and the subsequent distribution of the fluorescently labeled (Alexa-488) ASOs inside the cells.
Antisense oligonucleotides (ASOs) are single stranded negatively charged molecules which downregulate the translation of specific target messenger RNA (mRNA). Chemically modified ASOs with phosphorothioate (PS) linkages have been extensively studied as research tools and as clinical therapeutics and nine oligonucleotide-based drugs have been approved by regulatory agencies. While several cell surface proteins that bind PS-ASOs and mediate their cellular uptake have been identified, the mechanisms leading to productive internalization of PS-ASOs are not well understood. We demonstrate the potential of hyperspectral CARS imaging to detect the intracellular presence of ASOs in a label-free manner.
Fluorescence Lifetime Imaging Microscopy (FLIM), providing unique quantitative functional information, has gained popularity in various biomedical and molecular biology studies. Here we present an open-source Python package, FlimTK, a toolkit that enables state-of-the-art functions for FLIM image analysis and visualization. It contains comprehensive functionalities for reading FLIM raw files, fluorescence lifetime estimation, heterogeneity analysis, and spatial distribution analysis. FlimTK package is optimized for high performance and ease of use for integration into custom Python-based analysis workflows. FlimTK source code, demo analysis workflows, and tutorial documentation are available for download from GitHub.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.