Hyungsoon Im

Gold nanostructures with branched morphologies exhibit strong plasmonic properties, enabling advanced biosensing and biochemical applications. Here, we report a rapid and scalable synthesis of gold nanoflowers (AuNFs) using modified benzenethiols as both reducing and capping agents. Benzenethiols bearing electron‐donating substituents promoted rapid Au ³⁺ reduction and anisotropic branch growth, completing particle formation in less than 10 s without seed preparation or organic‐aqueous transfer steps. The resulting AuNFs had abundant plasmonic hot spots and intrinsic Raman reporters, serving as dual‐modality probes for colorimetry and surface‐enhanced Raman scattering (SERS). When integrated into lateral flow assays, AuNFs markedly enhanced analytical sensitivity. In oxycodone detection, limits of detection reached 4.48 pg/mL (colorimetry) and 0.38 fg/mL (SERS), representing 700‐fold and 1,200‐fold improvements, respectively, over spherical gold nanoparticles. This performance enabled accurate quantification of oxycodone in plasma samples from murine addiction models. The developed method provides a practical route for preparing high‐performance plasmonic nanomaterials for applications in biosensing, catalysis, and analytical chemistry.

Extracellular vesicles (EVs) have emerged as key analytes in liquid biopsy analysis, as they inherit the molecular characteristics of their parental cells. For cancer diagnosis and treatment monitoring, tumor-derived EVs can serve as surrogate markers of their originating tumor cells. However, technical challenges arise from EV heterogeneity, as molecular signals from a rare EV subpopulation are often neglected in bulk-EV analysis. This limitation underscores the importance of single-EV analysis, enabling the detection and analysis of individual tumor EVs. While multiple single-EV technologies have been introduced, accurately detecting and quantifying rare target EVs in clinical samples remains challenging in practice. Here, we developed a set of protocols that improve the selectivity and accuracy of single-EV analysis by minimizing false-positive signals and enhancing assay robustness. Specifically, we improved the validity of data processing for single-EV imaging, optimized EV labeling procedures, and applied colocalization analysis between cancer-specific markers and EV signals. We demonstrated that the approach is applicable to two orthogonal single-EV technologies, fluorescence imaging and nanoflow cytometry, yielding consistent results across the two methods and significantly improving the detection accuracy of tumor-derived EVs spiked in plasma samples. Collectively, we demonstrated precise single-EV profiling with high accuracy and selectivity, with evident benefits for the applicability of EV analysis in clinical studies.

Quantitative fluorescence analysis of single extracellular vesicles (EVs) is often complicated by heterogeneous particle loading on continuous surfaces, obscuring where and how a signal should be counted. This study presents a simple, array‐based method that couples site‐specific capture into optically resolvable nanowells with mask‐gated image analysis to obtain unambiguous, single‐site fluorescence measurements. EVs settle onto nanowell arrays, and excess particles on the inter‐well surface are cleared by a PDMS translation step, resulting in the capture of single EVs in discrete nanowells and accurate detection of in‐well signals that are registered to a bright‐field nanowell mask. This format yields > 99% EV capture at the designed location in grids and enables the accurate detection of EVs’ fluorescence signals with a low background. Compositional analysis on a single substrate accurately tracks programmed EV mixture ratios with near‐unity slopes and HER2 positivity in breast cancer EVs. Finally, the patterning method is readily adapted to plasma‐derived EVs, expanded to multi‐channel fluorescence imaging, and applied to mixed cargos of co‐patterned EVs and nanoparticles. These results establish a reliable pathway for array‐based EV detection and precise manipulations of EVs and other nanoparticles.

The necessity for early diagnosis of pancreatic ductal adenocarcinoma (PDAC) has been continuously emphasized due to its high mortality rate and limited treatment efficacy. Liquid biopsy, which enables early detection of this aggressive cancer, has been considered one of the most promising methods to improve patient survival outcomes. Among detectable biomarkers in liquid biopsy, tumor-derived extracellular vesicles (EVs) are particularly recognized for their clinical significance in early screening, diagnosis, and prognosis of cancer. However, the applicability of PDAC-derived EVs has been highly limited due to the lack of a suitable detection platform, which remains the main difficulty hindering further application. In this study, we present an electrochemical (EC)-based PDAC-derived EV sensing platform using gold nanostructure-modified ITO (AuNS@ITO) electrodes with sufficient sensitivity and specificity for clinical samples. The electrode modified with AuNS@ITO enhanced the EC signal by 6.6-fold compared to the AuNS-modified electrode, based on more efficient electron transfer achieved through high surface-to-volume ratio and superior electrical conductivity. Moreover, modified ELISAs (enzyme-linked immunosorbent assays) were utilized on the developed EC sensing platform, maximizing EV capture by targeting multiple cancer-specific markers (EpCAM/EGFR/MUC1) and labeling captured EVs with HRP-conjugated antibodies targeting EV markers (CD63/CD9/CD81). The developed EC sensing platform demonstrated a linear dynamic concentration range between 10⁴ and 10⁷ particles/mL, with a detection limit calculated as 33 particles/mL in Plasma. Furthermore, healthy control and PDAC patient-derived EV samples were successfully distinguished based on this analysis, and the platform could even separate PDAC patients by cancer stages. Therefore, this newly developed EC sensing platform for PDAC-derived EVs can contribute to realizing EV-based liquid biopsy for early cancer diagnosis and assessing cancer stage as an auxiliary diagnostic tool.

Ovarian cancer stem cells (CSCs) can seed recurrent drug-resistant disease. Likewise, non-CSCs can acquire CSC phenotypic properties. How this process is orchestrated is of interest to inform how it might be prevented. We tested the hypothesis that ovarian CSC and/or drug-resistant tumor cells confer stem-like properties via extracellular vesicles (EVs). We focused our investigation on how EVs might mediate EZH2 signaling to promote a phenotypic change in drug-sensitive, non-CSCs. To accomplish this, we utilized paired PARP inhibitor-sensitive and - resistant ovarian cancer (OvCa) cell lines, EZH2 knockdown lines, and patient-derived organoids (PDOs) originating from recurrent high-grade serous OvCa. Small EVs isolated from drug-sensitive, CSC and/or drug-resistant enriched cultures, PARP inhibitor (olaparib) resistant lines, or drug-treated (olaparib or carboplatin) lines were cultured with treatment naïve or sensitive lines for defined time points. The impact of small EV exposure was determined by assessing cell number, metabolic activity, viability, sphere and colony-forming capacity, ALDH activity, DNA damage, and changes in associated signaling pathways. We found that EVs from CSC or drug-resistant enriched cell fractions communicate CSC-like phenotypes to the more sensitive tumor cells via EZH2 canonical and non-canonical signaling pathways, promoting stemness. We conclude that EV-mediated activation of EZH2 signaling represents a targetable mechanism contributing to stemness-associated drug resistance in OvCa. Graphical

Cholangiocarcinoma (CCA) is a highly lethal malignant tumor with a five‐year survival rate below 10%. Clinical challenges include the absence of effective screening tools, the gradual progression until symptoms manifest, and the low diagnostic accuracy of CCA tissue due to the paucicellular nature of specimens obtained during an intervention. To address these challenges, an integrated multi‐channel automatic cellular profiling (iMAP) system is developed for robust analysis of endoscopically obtained biliary specimens. Brushing cytology samples are introduced into a microfluidic chip to enrich and capture suspicious epithelial cells while eliminating blood and fibrotic cells. The captured cells are subsequently immunostained with fluorescent antibodies for further molecular analysis. The entire process, from sample loading to diagnosis, is fully automated and completed within 30 min. After validating the system with in vitro cell lines, iMAP is applied to analyze clinical brush specimens. In a pilot clinical study with 56 patient samples, good clinical diagnostic accuracy (AUC = 0.85) is demonstrated for CCA using a four‐marker combination (MUC1, EpCAM, EGFR, and MARS1). The fully automated iMAP can serve as a complementary technology to the existing clinical cytopathology workflow, thereby minimizing sample loss and enhancing diagnostic accuracy through multiplexing.

Extracellular vesicles (EVs) play a crucial role in intercellular communication, signaling pathways, and disease pathogenesis by transporting biomolecules such as DNA, RNA, proteins, and lipids derived from their cells of origin, and they have demonstrated substantial potential in clinical applications. Their clinical significance underscores the need for sensitive methods to fully harness their diagnostic potential. In this comprehensive review, we explore EV heterogeneity related to biogenesis, structure, content, origin, sample type, and function roles; the use of EVs as disease biomarkers; and the evolving landscape of EV measurement for clinical diagnostics, highlighting the progression from bulk measurement to single vesicle analysis. This review covers emerging technologies such as single-particle tracking microscopy, single-vesicle RNA sequencing, and various nanopore-, nanoplasmonic-, immuno-digital droplet-, microfluidic-, and nanomaterial-based techniques. Unlike traditional bulk analysis methods, these methods contribute uniquely to EV characterization. Techniques like droplet-based single EV-counting enzyme-linked immunosorbent assays (ELISA), proximity-dependent barcoding assays, and surface-enhanced Raman spectroscopy further enhance our ability to precisely identify biomarkers, detect diseases earlier, and significantly improve clinical outcomes. These innovations provide access to intricate molecular details that expand our understanding of EV composition, with profound diagnostic implications. This review also examines key research challenges in the field, including the complexities of sample analysis, technique sensitivity and specificity, the level of detail provided by analytical methods, and practical applications, and we identify directions for future research. This review underscores the value of advanced EV analysis methods, which contribute to deep insights into EV-mediated pathological diversity and enhanced clinical diagnostics.

Lens-free digital in-line holography (LDIH) provides a large field-of-view at micrometer-scale resolution, making it a promising tool for high-throughput cellular analysis. However, the complexity of diffraction images (holograms) produced by LDIH presents challenges for human interpretation and requires time-consuming computational reconstruction, often leading to artifacts and information loss. To address these issues, we present HoloNet, a novel deep learning architecture specifically designed for direct analysis of diffraction images in cellular diagnostics. Tailored to the unique characteristics of diffraction images, HoloNet captures multi-scale features, enabling it to outperform conventional convolutional neural networks in recognizing well-defined regions within complex holograms. HoloNet classifies breast cancer cell types with high precision and quantifies molecular marker intensities using raw diffraction images of cells stained with ER/PR and HER2. Additionally, HoloNet has proven effective in transfer learning applications, accurately classifying breast cancer cell lines and discovering previously unidentified subtypes through unsupervised learning. By integrating computational imaging with deep learning, HoloNet offers a robust solution to the challenges of holographic data analysis, significantly improving the accuracy and explainability of cellular diagnostics.

INTRODUCTION: Artificial intelligence (AI) models have been applied to differential dementia detection tasks in brain images from curated, high‐quality benchmark databases, but not real‐world data in hospitals. METHODS: We describe a deep learning model specially trained for disease detection in heterogeneous clinical images from electronic health records without focusing on confounding factors. It encodes up to 14 multimodal images, alongside age and demographics, and outputs the likelihood of vascular dementia, Alzheimer's, Lewy body dementia, Pick's disease, mild cognitive impairment, and unspecified dementia. We use data from Massachusetts General Hospital (183,018 images from 11,015 patients) for training and external data (125,493 images from 6,662 patients) for testing. RESULTS: Performance ranged between 0.82 and 0.94 area under the curve (AUC) on data from 1003 sites. DISCUSSION: Analysis shows that the model focused on subcortical brain structures as the basis for its decisions. By detecting biomarkers in real‐world data, the presented techniques will help with clinical translation of disease detection AI. Highlights: Our artificial intelligence (AI) model can detect neurodegenerative disorders in brain imaging electronic health record (EHR) data. It encodes up to 14 brain images and text information from a single patient's EHR. Attention maps show that the model focuses on subcortical brain structures. Performance ranged from 0.82 to 0.94 area under the curve (AUC) on data from 1003 external sites.