GIST Scholar
Not a member yet
30271 research outputs found
Sort by
D2C-Morph: Brain regional segmentation based on unsupervised registration network with similarity analysis
No full textBrain regional segmentation is an image-processing approach widely used in brain image analyses. Deep learning models that perform segmentation alone play an important role in medical fields such as automatic diagnosis and prognosis prediction. This method is effective for rapid diagnosis and large-scale processing. However, spatial alignment between image data is required for accurate segmentation. We proposed D2C-Morph, which can jointly perform registration and segmentation through unsupervised learning. The proposed model emphasizes the features of each input through a dual-path network and is designed to use contrastive learning twice. In addition, we demonstrated that the performance of the decoder can be improved by using a correlation feature map that enhances the similarity of the feature maps between two inputs through a correlation layer. Our study demonstrates that the deformation field of the registration network can be utilized for segmentation to jointly perform image processing pipelines.FALSEsciescopu
SWIR-LightFusion: multi-spectral semantic fusion of synthetic SWIR with thermal IR (LWIR/MWIR) and RGB
No full textEnhancing scene understanding in adverse visibility conditions remains a critical challenge for surveillance and autonomous navigation systems. Conventional imaging modalities, such as RGB and thermal infrared (MWIR/LWIR), when fused, often struggle to deliver comprehensive scene information, particularly under conditions of atmospheric interference or inadequate illumination. To address these limitations, Short-Wave Infrared (SWIR) imaging has emerged as a promising modality due to its ability to penetrate atmospheric disturbances and differentiate materials with improved clarity. However, the advancement and widespread implementation of SWIR-based systems face significant hurdles, primarily due to the scarcity of publicly accessible SWIR datasets. In response to this challenge, our research introduces an approach to synthetically generate SWIR-like structural/contrast cues (without claiming spectral reproduction) images from existing LWIR data using advanced contrast enhancement techniques. We then propose a multimodal fusion framework integrating synthetic SWIR, LWIR, and RGB modalities, employing an optimized encoder-decoder neural network architecture with modality-specific encoders and a softmax-gated fusion head. Comprehensive experiments on public RGB-LWIR benchmarks (M3FD, TNO, CAMEL, MSRS, RoadScene) and an additional private real RGB-MWIR-SWIR dataset demonstrate that our synthetic-SWIR-enhanced fusion framework improves fused-image quality (contrast, edge definition, structural fidelity) while maintaining real-time performance. We also add fair trimodal baselines (LP, LatLRR, GFF) and cascaded trimodal variants of U2Fusion/SwinFusion under a unified protocol.The outcomes highlight substantial potential for real-world applications in surveillance and autonomous systems. Details of synthetic SWIR generation and fusion methodology will be publicly available at https://github.com/MI-Hussain/SynthSWIRNet_2.FALSEsciescopu
Exploring dynamic transcriptome variation in thyroid cancer and aged kidneys through single-cell RNA sequencing
No full textSingle-cell RNA sequencing (scRNA-seq) has revolutionized transcriptomic research by enabling the detailed examination of gene expression at the individual cell level, thereby overcoming the limitations of bulk RNA sequencing. This high-resolution approach enhances our understanding of cellular diversity, dynamic responses, and intricate heterogeneity across various biological contexts, including disease progression and immune modulation. scRNA-seq is particularly useful for studying complex tissues, such as the kidney, brain, and tumor microenvironments, as it elucidates previously elusive cell-cell interactions and responses. Despite its advantages, significant gaps remain in understanding the mechanisms underlying diseases, leading to limitations in treatment options across various diseases, including cancer. In particular, the dedifferentiation process in cancer, associated with poor prognosis and high recurrence rates, is poorly understood. Additionally, the detection of low abundance transcripts, such as long noncoding RNAs (lncRNAs), poses challenges for current RNA sequencing methodologies. To address these challenges, this study employed scRNA-seq to investigate the dedifferentiation process in thyroid cancer and its associated tumor microenvironment. By integrating scRNA-seq with whole exome sequencing, I identified copy number amplification of CREB3L1 as closely associated with dedifferentiation. Furthermore, I demonstrated that elevated expression levels of CREB3L1 were linked to decreased survival rates. I also introduced a novel targeted single-cell RNA sequencing approach designed to identify rare transcripts, including lncRNAs, leading to the discovery of tissue- and age-specific lncRNAs. I explored the functional roles of these lncRNAs by constructing gene regulatory networks and observed dynamic expression changes during aging, particularly in glomerular cells. Collectively, my findings provided insights into the heterogeneity and molecular evolution of thyroid cancer, highlighting the potential driver role of CREB3L1 in its dedifferentiation process. Additionally, I illuminated the comprehensive landscape of lncRNA expression and function, offering a valuable resource for future analyses. This work aims to deepen our understanding of critical biological mechanisms across various diseases and tissues, leveraging single-cell transcriptomic analysis to reveal molecular changes and cellular dynamics.DoctorAbstract∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ⅰ
List of Contents∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ⅲ
List of figures∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ⅴi
List of original publication∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ⅴiii
I. INTRODUCTION∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 1
I.1. Background of single cell RNA sequencing ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 1
I.2. Thyroid cancer dedifferentiation ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 3
I.3. Long non coding RNA∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 4
II. METHOD AND MATERIALS∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
II.1. Clinical information∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
II.2. Tissue dissociation and purification∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
II.3. Library preparation and sequencing (thyroid cancer)∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 6
II.4. Single-cell bioinformatics analysis (thyroid cancer)∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 7
II.5. Deconvolution analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 8
II.6. Cell-cell communication analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 8
II.7. Cell trajectory analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 8
II.8 InferCNV analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 8
II.9 SCENIC analysis and ssGSEA∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 9
II.10 WES and somatic alteration calling ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 9
II.11 ChIP sequencing data analysis ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 10
II.12 IHC and IF staining∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 10
II.13 Cell culture and overexpression of CERB3L1∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 11
II.14 Transwell assay∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 11
II.15 Western blotting∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 12
II.16 Experimental Animals∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 12
II.17 Cell Culture and Preparation for Single-Cell RNA-Seq∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 12
II.18 Single-Cell Preparation from Mouse Tissues∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 13
II.19 scRNA-Seq Library Construction∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙ 13
II.20 Enrichment of Single-Cell lncRNA∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 14
II.21 Nanopore Sequencing and Data Analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 14
II.22 Single-cell bioinformatics analysis (long non coding RNA)∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 15
II.23 Transcription Factor Selection∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 16
II.24 Gene Regulatory Networks Construction in Each Cell Type∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 17
II.25 Cellular Connectivity between On-Target Clusters∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 17
II.26 On-Target lncRNA Calculation∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 18
II.27 Primary Cell Culture for Small Interfering RNA Knockdown and RNA-Seq∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 18
II.28 In Situ Hybridization∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 19
III. RESULTS
III.1 Characterizing dedifferentiation of thyroid cancer by integrated analysis∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 20
III.1.1 scRNA-seq analysis and cellular profile of thyroid cancer∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 20
III.1.2 Dynamic transcriptional changes in thyroid cancer dedifferentiation∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 21
III.1.3 CNAs in thyroid cancer dedifferentiation∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 22
III.1.4 CREB3L1 as a master transcriptional regulator in ATC∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 23
III.1.5 Correlation of CREB3L1-driven pathways with thyroid cancer progression∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 24
III.2 Cell Type– and Age-Specific Expression of lncRNAs across Kidney Cell Types∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 26
III.2.1 Targeted scRNA-Seq significantly enriched lncRNAs∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 26
III.2.2 Constructing a single-cell lncRNA atlas of various mouse tissues∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 26
III.2.3 Cell type–specific lncRNAs showing a high correlation to known cell-type marker genes in
the kidney cells∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 27
III.2.4 Gene regulatory networks of lncRNAs and transcription factors in the kidney cells∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 28
III.2.5 Exploring Age- and Cell Type–Specific Expression of lncRNAs in Glomerular Cells∙∙∙∙∙∙∙∙∙∙ 29
III.2.6 Deciphering aging-associated changes in lncRNA expression patterns across kidney cell
types∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 30
IV. DISCUSSION∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 96
V. ABSTRACT IN KOREAN∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 100
VI. REFERENCES∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 102
ACKNOWLEDGEMENT∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ 11
Development of an Open-Sided Magnetic Particle Imaging System With Amplitude Modulation-Based Reconstruction
No full textMagnetic particle imaging (MPI) is a rapid and sensitive imaging technique that can be used to measure the spatial distribution of superparamagnetic iron oxide (SPIO) nanoparticles, which can be applied to vascular and perfusion imaging, cancer imaging, hyperthermia, drug treatments, and intraoperative navigation. Although closed-bore conventional MPI scanners offer the advantage of high coil efficiency, they significantly increase the risk of claustrophobia and panic attacks, may not accommodate all body shapes and sizes, are challenging to integrate with other medical modalities, and face several limitations when scaled for human use. Open-sided MPI with amplitude modulated (AM) has the potential to overcome these above issues due to the accessibility of imaging objects from sides without coil size constraints and using a low-amplitude high-frequency excitation field combined with a low-frequency high-amplitude drive field. In this article, we develop a compact open-sided MPI system utilizing a mechanically rotated 3-D fieldfree line (FFL) with permanent magnets for energy efficiency. The AM MPI method along with open gradiometer coils with a packaged configuration enhances uniform sensitivity. The developed system can achieve a field of view (FOV) of 40 × 40 × 40 mm3. To obtain guidance on working parameters for the AM method and compare the performance of parameters with the X-space method, theory was developed, and results were simulated and experimentally verified by a developed scanner using AM for high resolution and sensitivity, avoiding heating issues. © 2025 IEEE.FALSEsciescopu
Maximizing oxygen permeation via catalytic functionalization under oxyfuel conditions
No full textOxygen transport membranes (OTMs) offer a promising route for high-efficiency, cost-effective oxygen supply in energy and chemical industries, with the potential to significantly reduce CO2 and NOx emissions when integrated into oxy-combustion processes. However, conventional OTMs suffer from poor chemical stability in CO2-rich environments, prompting the development of dual-phase membranes that, while more stable, typically exhibit lower oxygen permeation rates. In this study, we address this limitation by enhancing the surface exchange kinetics of Fe2NiO4-Ce0.8Tb0.2O2-delta (NFO-CTO) membranes by surface modification with various oxygen oxidation-reduction reaction (OORR) catalysts, including Ce, Pr, Sm, Tb, Co, Nb, Zr, and Al oxides, and Pr-based binary oxides. Comprehensive characterization using electrochemical impedance spectroscopy, oxygen isotopic exchange, and gas permeation measurements revealed a substantial improvement in surface reaction kinetics. Catalyst activation led to a six-fold increase in oxygen flux under standard conditions and up to a 2.5-fold enhancement under harsh environments containing CO2 and SO2 at 850 degrees C, mimicking oxyfuel combustion conditions. This work demonstrates that rational catalyst selection and integration can overcome fundamental surface limitations in dual-phase membranes, offering a viable strategy to advance oxygen separation technologies for sustainable energy applications.TRUEsciescopu
High C-rate Li-NMC/graphite pouch cell end-of-life prediction via cycle-dependent variations and machine learning
No full textThe accurate prediction of end-of-life for lithium-ion batteries is crucial for enhancing safety, reliability, and cost-efficiency in electric vehicles and energy storage systems. This study investigates the degradation characteristics of Li-NMC/graphite pouch cells under high C-rate conditions and introduces a machine learning-based predictive model for EoL estimation. Incremental capacity analysis is integrated with ensemble models such as Random Forest, Gradient Boosting, and CatBoost to extract electrochemical degradation features. Our model accurately predicts the cycle number at which state of health reaches 80%, with the Gradient Boosting algorithm achieving the highest prediction accuracy, with a root mean squared error of 17.63 and a mean absolute percentage error of 3.11. These findings demonstrate the potential of data-driven approaches for reliable battery health monitoring. The proposed framework can significantly contribute to the advancement of predictive maintenance strategies in battery management systems. © 2025 The Royal Society of Chemistry.sciescopu
Robustness and tunability of spin dynamics in FeRh films under magnetic inhomogeneity control
No full textUnderstanding and controlling spin precession dynamics are crucial for ultrafast magnetic switching in spintronic devices and next-generation memory technologies. In this work, we investigate how the coherent spin precession in the ferromagnetic state of FeRh is influenced by various types of spatial magnetic inhomogeneity, including a deficiency of ferromagnetic ions and two- or three-dimensional mixtures of ferromagnetic (FM) and antiferromagnetic (AFM) phases. H+-ion irradiation is widely used to stabilize the FM state of FeRh at room temperature, and our results reveal that it has little effect on the main characteristics of spin precession dynamics. In the coexisting FM(-)AFM phases, we found that FM resonance is strongly influenced by exchange interactions across the magnetic phase boundaries, resulting in significant changes in the precessional frequency and the damping coefficient. These findings highlight both the robustness and the tunability of spin dynamics using the material design and the magnetic phase control, which can be useful for spintronic applications.FALSEsciescopu
Development of a lateral control system for autonomous vehicles by integrating quasi-static and dynamic control methods
No full textRecent advancements in lateral control systems for autonomous vehicles have focused on achieving high tracking precision and stable control performance. Building on these efforts, this study designs and implements an integrated lateral control system using quasi-static Stanley method and dynamic Proportional, Integral, Derivative (PID) controller.The proposed system works in three stages: positioning, quasi-static control and dynamic control. In the positioning stage, precise positioning is achieved by applying Real-Time Kinematic enhanced positions and headings with velocity obtained from Inertial Measurement Unit. Additionally, a method is proposed to collect and process path coordinate data in environments where predefined path coordinates are unavailable. In the quasi-static control stage, an efficient path tracking control algorithm based on the Stanley method is implemented and applied to the collected ground truth path coordinate data. Finally, in the dynamic control stage, the PID controller is utilized to enhance control precision by minimizing the difference between the vehicle's actual heading and the target heading response time required to reach the target heading through adjusting the steering wheel angle.The experiment was conducted by driving along two target paths within the GIST campus. On the first target path, Mean Absolute Error (MAE) was recorded at 0.0786 m, and Root Mean Square Error (RMSE) was 0.1140 m. Additionally, the average error between the vehicle's target heading and actual heading was recorded as 2.1411 degrees. On the second target path, MAE was 0.0427 m, and RMSE was 0.0802 m. The average error between the target heading and actual heading was 2.5000 degrees, demonstrating the performance of the lateral control system.To prove the superiority of the proposed method, thorough collections of prior real-world lateral control systems were exercised, and direct comparisons were made. To the best of our knowledge, the proposed system outperforms previous literature on real-world experiments. We also provide the download link to the target path coordinate data so that future competitions on the best lateral control systems can be held. We believe our work can serve as a foundation for future lateral control systems in the autonomous driving community.TRUEsciescopu
Improving the OER Performance of NiFe-LDH electrocatalyst via Heterostructure Formation with NiB in Alkaline Media
No full textNickel-iron layered double hydroxides (NiFe-LDH) present a promising alternative to precious metal electrocatalysts for the oxygen evolution reaction (OER) in alkaline media. This potential stems primarily from the Ni-O-Fe bond, which facilitates the formation of catalytically active sites. However, Fe sites in NiFe-LDH are prone to leaching during the OER process, resulting in diminished activity and stability. This study investigates the catalytic activity of heterostructures NiFe-LDH with NiB (NiFe@NiB) compared pristine nickel boride (NiB) and NiFe-LDH. The properties of NiB, such as electrical conductivity, surface area, and coordination number, can be finely tuned through annealing temperature. Notably, boron, acting as an electron acceptor, induces changes in the nickel oxidation state (Ni³⁺ → Ni(³⁺δ)⁺), which drives a phase transformation from β-NiOOH to γ-NiOOH. The γ-NiOOH structure in the NiFe@NiB enhances the intrinsic catalytic activity compared to the β-NiOOH structure of NiFe-LDH. The NiFe@NiB composite exhibits an approximately fivefold increase in surface roughness and a 9.95-fold enhancement in electrochemical surface area (ECSA) relative to NiFe-LDH. Additionally, the intrinsic activity, as represented by turnover frequency, increased by 6.31%. Consequently, the NiFe@NiB achieved a significantly reduced overpotential of 280 mV at 10 mA cm⁻², compared to 355 mV for NiFe-LDH under alkaline OER conditions. Furthermore, the potential of NiFe-LDH increased by 39 mV over 130 hours, whereas NiFe@NiB exhibited only a 17 mV increase, reflecting a 41% improvement in stability. The NiFe@NiB catalyst demonstrated negligible changes in charge transfer resistance (Rct) and overall catalytic activity during durability testing, highlighting its robust performance and long-term stability.MasterAbstract i
Contents ii
List of Tables iii
List of Figures iv
1. Introduction 1
2. Theoretical Background 3
2.1.Oxygen evolution reaction (OER) catalyst 3
2.2. Nickel-iron layered double hydroxide (NiFe-LDH) 3
2.3. Strategies for increasing layer distance of nickel-iron LDH: formation of heterostructure with nickel boride· 4
3. Experiments 8
3.1. Preparation of NiFe@NiB heterostructure catalyst 8
3.2. Materials characterization 9
3.3. Electrochemical analysis 9
4. Results and Discussion 10
4.1. Nickel boride (NiB) substrate 8
4.2. Effect of heterostructure with NiB substrate on OER performance of NiFe-LDH catalyst 8
5. Conclusion 29
6. References 30
Acknowledgement 3
A point cloud-based mesh-independent convolutional neural network frameworks for flow field prediction on variable geometries
No full textDespite advancements in high-performance computing and numerical algorithms, Computational Fluid Dynamics (CFD) remains challenging for practical real-time applications, particularly in analysis and design tasks such as digital twin implementations. While traditional Reduced-Order Methods offer efficient and accurate predictions of entire flow fields, autoencoder Convolutional Neural Networks (CNNs) have demonstrated success in reconstructing CFD solutions due to their exceptional local feature extraction capabilities and scalability. However, their applicability is constrained to simple geometries because of the reliance on Cartesian or pixel-like grid structures. In this study, we propose a novel Point-based U-Net (PointUNet) framework incorporating Local Point Encoding (LPE) as a mesh-independent autoencoder model. The key functionality of LPE lies in its ability to transform point cloud data into a standard input array for conventional CNNs using a Virtual Reference Grid. This approach avoids data loss typically associated with interpolation or extrapolation, enabling greater flexibility in mesh generation and complex geometry handling. Verification was conducted using airfoil flows at transonic speeds and cylinder flows at low Reynolds numbers with various cross-sectional shapes, achieving minimal verification errors. The results were compared directly with other point cloud methods, demonstrating superior accuracy and efficiency in predicting highly nonlinear flows involving separation and shock waves, showing better agreement with full-order CFD solutions.FALSEsciescopu