https://scholarworks.unist.ac.kr/handle/201301/12 Wed, 08 Apr 2026 00:41:18 GMT 2026-04-08T00:41:18Z https://scholarworks.unist.ac.kr/handle/201301/90912 Title: Fusion-Based Digital Profiling of Extracellular Vesicles for Cancer Diagnosis Author(s): Clarissa, Elizabeth Maria Abstract: Cancer is among the leading causes of death worldwide, accounting for approximately one in nine deaths in men and one in 12 deaths in women, according to the global cancer statistics (GLOBOCAN) 2022, with an estimate of 32.6 million incidence by the year 2045. Tissue biopsy has been held as the gold standard for cancer diagnosis; however, it presents various drawbacks, as it is invasive, painful, and poses infection risk. Tissue biopsy is also rather inappropriate for capturing tumor heterogeneity, which is considered as one of the key factors of treatment resistance, and inadequacy in detecting early-stage tumor, or residual lesions, constraining its effectiveness. In order to address these challenges that hamper the effectiveness of cancer diagnosis, a non-invasive alternative termed the liquid biopsy, has emerged. Liquid biopsy involves the sampling of body fluids such as urine and blood, to detect various biomarkers such as circulating tumor cells (CTCs), circulating tumor DNA (ctDNA), circulating tumor RNA (ctRNA) and tumor-derived extracellular vesicles (EVs). Extracellular vesicles (EVs) are considered as a promising liquid biopsy marker due to their stability and abundance, and due to the fact that they play roles in intercellular communication system, carrying various biomarkers such as DNA, RNA, protein, lipid, sugar structures, and metabolites. Various approaches in both EVs isolation and detection have been made to employ EVs as the target for liquid biopsies, and one approach that has been suggested for EVs detection is fusion. The process involves fusing the EVs with membrane vesicles containing targeting probe, allowing the detection of EV internal cargo such as nucleic acids. However, most fusion methods are done in bulk scale, compromising the system’s sensitivity. This thesis describes the combination of the two approaches of fusion and compartmentalization, aiming to provide an EV biomarker detection method in a digital manner, termed the EV-CLIP. EV-CLIP stands for EVs fusion with charged liposomes (CLIPs), which does not require the need for labor-intensive sample pre-processing, avoiding loss of the EVs, further simplifying the detection process and increasing the detection sensitivity. The EV-CLIP method described in this thesis allows a charge-based, protein-independent fusion method to incorporate molecular beacon (MB) into EVs. Through the adjustment of ratios between positive and negatively charged lipids, CLIP surface charge could be fine-tuned for efficient fusion with EVs, where variation of lipid composition yields different fusion efficiency ranging from 4.6% to 60.2%. Following the optimization of CLIP to EV ratio, coupling of EV-CLIP fusion with compartmentalization was achieved through microfluidic droplet reactor. The EV-CLIP platform could digitally profile the presence of miRNA and mRNAs within individual EVs, while preventing uncontrolled aggregation that would otherwise happen in the conventional bulk setting. This EV-CLIP approach simplifies RNA detection using a minimal sample volume of 20 µL, while eliminating the need for prior EV isolation or RNA preparation, further preserving the sample integrity. The thesis also discusses detailed aspects of the EV-CLIP system including the kinetics of the reaction, and other detailed characterization. Lastly, the thesis also highlights the clinical implementation of EV-CLIP. The approach was tested with 83 patient samples and detected EGFR L858R and T790M mutations with high AUC values of 1.0000 and 0.9784, respectively. Additionally, EV-CLIP’s performance in quantifying EV-derived mRNA for specific mutations in serial monitoring samples taken during course of chemotherapy highlights its potential for precise quantification of rare EV subpopulations, facilitating the exploration of single EV RNA content and enhancing understanding of diverse EV populations in various disease states. underlining the translational potential of EV-CLIP to be integrated into clinical systems. In conclusion, EV-CLIP method allows an unbiased comprehensive analysis of heterogeneous EV population, contributing to a more nuanced understanding of EV biology and biomarker applications and offering a reliable platform for the detection of EV-derived biomarkers, especially RNA–allowing a precise and informative liquid biopsy application. Major: Graduate School of UNIST (2013-2020) Department of Biomedical Engineering Sat, 31 Jan 2026 15:00:00 GMT https://scholarworks.unist.ac.kr/handle/201301/90912 2026-01-31T15:00:00Z https://scholarworks.unist.ac.kr/handle/201301/90902 Title: Blood-Based Precision Epigenomics for Elucidating the Molecular Basis of Psychiatric Disorders Author(s): Kwon, Yoonsung Abstract: This dissertation demonstrates that blood-based DNA methylation changes provide crucial insights into the biological mechanisms underlying psychiatric disorders and represent an initial step toward defining the personal epigenome. Given the global increase in psychiatric disorders, elucidating their molecular basis is of growing importance. To understand and discover predictive biomarkers of mental diseases, we identified and validated methylation biomarkers associated with anxiety and suicide attempt in Koreans. In anxiety disorders, validated epibiomarkers implicated pathways involved in neurosignaling, synaptic regulation, apoptosis, and mitochondrial dysfunction. In suicide attempts, epibiomarkers primarily linked to synaptic signaling were consistently detected in both blood and postmortem brain tissues, indicating cross-tissue and cross-ethnic relevance. To address the limitations of GRCh38 in epigenomic analyses, we identified CpG-creating variants specific to the Korean population using the Korea10K Genome Project, which revealed an increased number of CG sites near genes associated with psychiatric disorders in Koreans. Additionally, suicide-associated epibiomarkers were discovered from novel CpG sites created by DNA variants, emphasizing the significance of ethnicity-informed epigenetic variation. Collectively, these findings suggest that psychiatric disorders exhibit systemic epigenetic signatures detectable in blood, providing a basis for personalized and precision mental health diagnostics. Major: Department of Biomedical Engineering Sat, 31 Jan 2026 15:00:00 GMT https://scholarworks.unist.ac.kr/handle/201301/90902 2026-01-31T15:00:00Z https://scholarworks.unist.ac.kr/handle/201301/90900 Title: Color Compensation for Visibility Enhancement in Optical See-Through Displays across Ambient Light Author(s): Park, Hyunhee Abstract: Optical see-through (OST) displays present digital imagery over a real scene. Because the light reaching the eye is an additive combination of displayed light and ambient light transmitted from the environment, changes in ambient luminance level and ambient chromaticity (summarized by correlated color temperature, CCT) can alter the appearance of overlaid content, including reduced contrast, reduced color strength, and shifts in overall color balance. Increasing SR is associated with compression of tone differences in low and mid levels and a reduction in CIELAB chroma (C*). Across SR levels, changes in ambient CCT are associated with hue-dependent deviations and chroma changes relative to a 6500 K darkroom reference condition. Based on these observations, an image visibility enhancement algorithm is formulated in the BT.709 YCbCr domain. SR-dependent processing adjusts Y(Luma) using a power-law tone parameter α expressed as a function of SR and image-mean luma Ȳ, and adjusts color magnitude using a power-law chroma parameter β that varies with SR within the tested range. In addition, CCT-dependent adjustments are organized as a lookup table of hue shift and chroma weight indexed by SR, CCT, and YCbCr hue, applied through rotation and scaling of the (Cb, Cr) vector while preserving Y(Luma). Pairwise preference evaluations with natural images on an OST display indicate that SR-dependent tone and chroma processing tends to improve visibility preference as SR increases under 6500 K ambient light. When ambient chromaticity differs from 6500 K, adding the CCT-dependent adjustment tends to provide additional preference gains. Qualitative analysis suggests that observer judgments reflect multiple criteria beyond shadow recovery alone, including highlight separation, texture preservation, scene impression in dark or backlit content, and the legibility of informative elements such as text and icons. Major: Department of Biomedical Engineering Sat, 31 Jan 2026 15:00:00 GMT https://scholarworks.unist.ac.kr/handle/201301/90900 2026-01-31T15:00:00Z https://scholarworks.unist.ac.kr/handle/201301/90901 Title: Quantifying geroelasticity through transcriptomic aging clocks during infectious stress Author(s): An, Kyungwhan Abstract: This dissertation advances the perspective that aging is dynamic and reversible rather than strictly unidirectional. Biologically meaningful age (or transcriptomic age) is measured here using RNA sequencing data from whole blood of human individuals. Transcriptomic age is far more flexible than chronological age, as shown by reversible age shifts observed during infection. To establish the primary objective of managing aging within theoretical multiomic domains, this work reconceptualizes the reversibility of aging by introducing two new terms: geroelasticity, which draws biophysical analogies from elasticity and plasticity to describe the adaptive properties of multiomic aging against stress, and gerostasis, which models aging as an information-driven process in which entropy accumulation can be computationally minimized and actively maintained via multiomic control. In summary, this dissertation reframes aging as a modifiable process, offering both methodological and conceptual foundations for future research aimed at extending both healthspan and lifespan. Major: Department of Biomedical Engineering Sat, 31 Jan 2026 15:00:00 GMT https://scholarworks.unist.ac.kr/handle/201301/90901 2026-01-31T15:00:00Z