Soo-Yeon Cho

Although chimeric antigen receptor T-cell (CAR-T) therapy has substantially improved outcomes in diffuse large B-cell lymphoma (DLBCL), disease relapse remains a major clinical challenge. Because treatment-associated immune remodeling evolves dynamically over time, complementary biomarkers capable of longitudinally monitoring immune-cell state changes are needed to evaluate therapeutic response and relapse following CAR-T therapy. In this study, we developed a label-free, high-throughput image cytometry platform for longitudinal, multiparametric profiling of peripheral blood mononuclear cells (PBMCs) collected from a cohort of 17 DLBCL patients undergoing CAR-T therapy. Using bright-field single-cell imaging, we quantified four representative biophysical parameters, including cell size, eccentricity, refractive index, and mean pixel intensity, from PBMC populations at base and post-treatment time points. These parameters were analyzed at both population and patient levels to characterize treatment-associated remodeling trajectories and biophysical heterogeneity. Static comparisons revealed population-level biophysical differences between the 10 relapsed (R) and 7 nonrelapsed (NR) patients. Longitudinal and patient-level analyses, performed by tracking serial PBMC samples collected from the same patients across treatment time points, demonstrated that treatment outcome was associated with coordinated remodeling trajectories in NR patients and greater patient-specific heterogeneity in R patients. Patient-level effect size and heterogeneity analyses further demonstrated that relapse was associated with altered remodeling dynamics that were not apparent from population-averaged measurements alone. These findings demonstrate that longitudinal biophysical profiling of PBMCs enables quantitative monitoring of treatment-associated immune remodeling and provides a complementary biomarker for assessing therapeutic response following CAR-T therapy.

Microalgae offer a promising platform for carotenoid manufacturing due to their productivity, tunable selectivity, and compatibility with large-scale bioprocessing. Yet, their cellular responses exhibit highly heterogeneous and asynchronous pigment and morphological transitions under fluctuating light stress, making precise prediction and process optimization challenging. Existing analytical methods cannot resolve these transitions in living cells because they either require destructive sampling or lack spectral specificity. In this study, we introduce near-infrared (nIR) image cytometry (NIC), a label-free, nondestructive, and high-throughput platform that quantifies biophysicochemical heterogeneity at single-cell resolution. By positioning chlorophyll detection within the nIR window, NIC structurally removes spectral interference from visible pigments and achieves precise quantification of chlorophyll content comparable to liquid chromatography. Simultaneously, NIC extracts biophysical parameters such as cell size and shape together with biochemical states including carotenoid accumulation in a fully label-free manner, enabling direct observation of their mechanistic coupling in living cells. Using representative industrial microalgae including D. salina and H. pluvialis as test beds, NIC resolves species-specific photoprotective strategies through 3D heterogeneity mapping. Furthermore, we customized a machine learning model that enables the prediction of cultivation stages for completely unknown batches with up to 89.0% stage-variable accuracy, driven by the single-cell data and multivariate biophysicochemical features quantified by NIC. This generalizable framework links cellular heterogeneity to process behavior and provides a basis for real-time monitoring in next-generation microalgal biomanufacturing.

Neuromodulators such as dopamine (DA) operate through volume transmission in the brain, creating spatially distributed fields that shape circuit excitability. However, conventional sensing modalities, including voltammetry and fiber photometry, inherently collapse these multidimensional dynamics into one-dimensional, single-point signals, thereby obscuring spatial heterogeneity and propagation behavior. Here, we introduce the Neural Remote Imaging Fiber Interface (NeuRIFI), a dual-readout near-infrared (NIR) sensing platform that bridges localized neurochemical detection and field-level spatial interpretation. By functionalizing implantable optical fibers with DNA-functionalized SWCNT nanosensors, this platform leverages NIR fluorescence emission within the tissue-transparent optical window to enable synchronous fiber-guided photometry and remote NIR imaging without genetic labeling. In this configuration, the photometric channel provides a spatially anchored validation reference, while the remote imaging channel preserves the spatial information on neurochemical signals. We validated the interface across tissue phantoms and ex vivo mouse brains, demonstrating reliable deep-brain signal acquisition at a 3 mm depth. Furthermore, by computing frame-to-frame intensity derivatives, we resolved transient DA flux at a temporal resolution of 20 ms, a spatial resolution of ∼1.5 μm per pixel, and a concentration sensitivity of 90 nM, enabling high spatiotemporal resolution mapping of neurochemical transport across the sensing surface. Collectively, this dual-readout architecture provides a framework linking point-validated sensing to remotely observable spatial structures, offering spatial accessibility for studying complex neurochemical landscapes in deep-brain structures and extending neurochemical field mapping toward in vivo settings.

Individual cells sense and transition between functional states, and the distribution of these states over time determines drug response, disease progression, and cell manufacturing outcomes. However, repeated measurement is difficult with label‐based acquisition, as photobleaching, phototoxicity, and probe‐induced perturbation accumulate. Label‐free monitoring that leverages intrinsic physical signals circumvents these constraints, shifting the analytical burden from label chemistry to instrumental drift. In this review, we organize this field into three measurement modalities, including imaging‐based, vibrational spectroscopy‐based, and electrical sensing‐based, each linked to distinct intrinsic state variables. Anchoring each modality to biological state variables such as structural organization, molecular composition, or dielectric architecture enables a physics‐grounded framework that clarifies how intrinsic signals map onto functional cellular phenotypes in longitudinal monitoring. We describe the measurement principle, drift sources, and feature space for each modality, and evaluate representative platforms against a frame spanning design, feature definition, quantitative performance, and validation practice. Dominant analytical constraints differ systematically across modalities, motivating integrative architectures in which complementary modalities resolve ambiguities that no single modality can disentangle. We further discuss shared requirements for calibration, standardized reporting, and multimodal integration, and outline requirements in hardware miniaturization, edge inference, and artificial intelligence‐guided molecular attribution to support scalable quantitative single‐cell phenotyping.

High-dimensional, cross-reactive sensor arrays enable powerful chemical fingerprinting, but practical use is often limited by the difficulty of identifying a small, reliable subset of sensors from screening data that contain only a few replicates per analyte. In many experimental settings, collecting large data sets is infeasible, and sensor selection must be performed under conditions of substantial variability and partial overlap between analyte responses. Here, we present a transparent and data-efficient analysis framework that enables rational reduction of sensor arrays using limited experimental data. Rather than relying solely on black-box accuracy metrics, our approach constructs simple, analyte-specific decision regions from measured responses and evaluates how individual sensors contribute to the separation of analytes that are most difficult to distinguish. Sensors are ranked by their contribution to resolving these difficult cases, producing stable and reproducible selections even when replicate numbers are small. An intuitive trade-off curve directly identifies the smallest sensor subset required to reach a desired classification performance. The robustness of the approach is demonstrated using controlled inflation alteration of experimental variability and by application to three independent fluorescence-based sensor libraries. In all cases, we show that only a few selected sensors are sufficient to achieve low classification error while retaining clear, interpretable decision maps that reveal how sensor responses give rise to analyte discrimination. This work provides an open-source, general, platform-agnostic strategy for screening data analysis and sensor-array design, offering compact and interpretable solutions suited to data-limited conditions common in chemical and biological sensing.

Cellular efflux provides a direct readout of homeostasis, metabolic activity, and pathological changes. As biological systems progress from single cells to 2D planar models, 3D constructs, and whole organs, rising biological complexity reshapes efflux behavior through shifts in ionic fluctuations, diffusion patterns, matrix composition, and tissue mechanics. Because each level introduces distinct efflux information that reflects individual cell behavior, collective signaling, or integrated physiological activity, dedicated sensing strategies are required to capture these signals with fidelity across the full hierarchy. In this review, we outline recent sensing strategies and interface designs that enable efflux monitoring across these biological hierarchies, and then describe how these concepts emerge in the major transducing modalities, including optical, electrochemical, and electrical approaches that define which efflux signatures can be measured. We highlight implementations that use photonic nanomaterials, nanostructured electrodes, and field effect architectures to sustain quantitative readout in increasingly complex microenvironments. We further discuss how diffusion behavior, reaction kinetics, and sensor response characteristics shape efflux feature extraction across scales, and highlight the key directions this field must advance toward. Combined with advances in materials and interface engineering, these developments establish efflux sensing as a central analytical framework for tissue engineering, drug evaluation, and bioprocess control.

Point mutations in circulating tumor DNA (ctDNA) represent critical biomarkers for minimally invasive cancer management. However, their detection remains challenging because single-base mismatches impose only a modest energetic penalty within long ctDNA fragments and ensemble signal averaging in conventional transducer configurations obscures such subtle differences. Here, we present a kinetically tuned single-walled carbon nanotube nanosensor that selectively resolves the single-nucleotide KRAS G12D mutation against wild-type backgrounds through near-infrared (nIR) fluorescence modulation. The DNA corona was rationally engineered by computational anchoring-domain selection to maximize nanotube affinity and transduction gain, followed by capture-domain length tuning to suppress wild-type hybridization while preserving mutant complementarity. Stronger anchoring yielded amplified and reproducible spectral shifts, and systematic probe truncation revealed that the site of shortening governs the balance between mutant affinity and wild-type exclusion. By fitting single-analyte and cotitration assays to a competitive binding model, we extracted quantitative kinetic parameters, and these kinetic binding models rationalize the observed selectivity trends in terms of mismatch-dependent hybridization thermodynamics, thereby defining generalizable design rules for point mutation-selective DNA/single-walled carbon nanotube (SWCNT) sensors. Guided by this framework, the optimized construct achieved a limit of detection (LOD) of 151.6 nM in serum spiked with wild-type DNA, demonstrating robustness in complex biofluids. This kinetic-model-driven nanosensor strategy introduces a principled route for precise point mutation detection directly in liquid biopsy samples, providing a disruptive alternative to sequencing-based assays for portable, real-time cancer monitoring.

Reactive oxygen species (ROS) bursts from photoaging cause skin damage and chronic conditions. Understanding spatiotemporal ROS dynamics is critical for developing therapies and cosmetic strategies to enhance skin health. Conventional assays and fluorescence microscopy lack the resolution for real-time ROS quantification due to photobleaching and labeling issues. Here, we developed a label-free, real-time monitoring platform with high spatiotemporal resolution using a near-infrared (nIR) fluorescent single-walled carbon nanotube (SWNT) nanosensor array to quantify ROS bursts from daily photoaging. The SWNT array, dual-functionalized with DNA and poly- l -lysine, achieved selective H ₂ O ₂ recognition and skin cell compatibility. The skin cell–friendly nanosensor interface (SNI) enabled attomole-level detection of H ₂ O ₂ bursts in a two-dimensional keratinocyte model under natural ultraviolet exposure, revealing photoadaptation behavior. Distinct oxidative stress wave profiles were identified via nIR data and numerical modeling. Using SNI, we introduced the anti-ROS score to evaluate skin care antioxidants, providing insights into photoaging pathways and cosmetic advancements.

Aging heterogeneity in tissue-regenerative cells leads to variable therapeutic outcomes, complicating quality control and clinical predictability. Conventional analytical methods relying on labeling or cell lysis are destructive and incompatible with downstream therapeutic applications. Here we show a label-free, nondestructive single-cell analysis platform based on nanosensor chemical cytometry (NCC), integrated with automated hardware and deep learning. nIR fluorescent single-walled carbon nanotube arrays in a microfluidic channel, together with photonic nanojet lensing, extract four key aging phenotypes (cell size, shape, refractive index, and H₂O₂ efflux) from flowing cells in a high-throughput manner. Approximately 10⁵ cells are quantified within 1 h, and NCC phenotype data were used to construct virtual aging trajectories in 3D space. The resulting phenotypic heterogeneity aligns with RNA-sequencing gene-expression profiles, enabling reliable prediction of therapeutic efficacy. The platform rapidly identifies optimally aged cells without perturbation, providing a robust tool for real-time monitoring and quality control in regenerative-cell manufacturing.