Niigata University Medical and Dental Hospital
Niigata University Academic RepositoryNot a member yet
33813 research outputs found
Sort by
子宮内膜症に対する遺伝的感受性とE-カドヘリン遺伝子の関与の可能性
Get PDF新潟大学Niigata University博士（医学）Background: Endometriotic cells display invasive characteristics despite their benign histological appearance. Recently, the epithelial-mesenchymal transition,in which epithelial cells acquire mesenchymal and migratory properties, has attracted attention as a mechanism of tumor invasion. We aimed to investigate the association between endometriosis and polymorphisms of the E-cadherin gene, a central player in the epithelial-mesenchymal transition,in Japanese women. Methods: Twelve single nucleotide polymorphisms (SNPs) in the E-cadherin gene were identified by real-time PCR using a TaqMan assay in 511 women with endometriosis (the majority in stages III and IV) and 498 healthy controls. Results: Allel e frequency analysis indicated that there was a marginally higher frequency of the rs4783689 C allele in women with endometriosis compared wit h controls (corrected p = 0.007 ; odds ratio = 1.37; 95% confidence interval, 1.14 to 1.64). No significant associations with endometriosis were found for the other eleven SNPs. Conclusions: Although this study was limited by sample size , the E-cadherin gene polymorphism rs4783689 was marginally associated with endometriosis in the Japanese population, suggesting that E-cadherin might be involved in genetic susceptibility to endometriosis.Kunihiko Yoshida, Kosuke Yoshihara, Sosuke Adachi, Kazufumi Haino, Koji Nishino, Masayuki Yamaguchi, Nobumichi Nishikawa, Katsunori Kashima, Tetsuro Yahata, Hideaki Masuzaki, Hidetaka Katabuchi, Kenichiro Ikuma, Hiroshi Suginami, Kenichi Tanaka, Possible involvement of the E-cadherin gene in genetic susceptibility to endometriosis, Human Reproduction, Volume 27, Issue 6, June 2012, Pages 1685–1689, https://doi.org/10.1093/humrep/des080This article has been accepted for publication in Human Reproduction Published by Oxford University Press.新大博（医）第1829
C57BL/6NマウスにおけるVps13a遺伝子を欠損した脳、血液、精巣の表現型の解析
Get PDF新潟大学Niigata University博士（医学）The Vps13a gene encodes a lipid transfer protein called VPS13A, or chorein, associated with mitochondria-associated endoplasmic reticulum (ER) membranes (MAMs), mitochondria-en-dosomes, and lipid droplets. This protein plays a crucial role in inter-organelle communication and lipid transport. Mutations in the VPS13A gene are implicated in the pathogenesis of chorea-acan-thocytosis (ChAc), a rare autosomal recessive neurodegenerative disorder characterized by chorea, orofacial dyskinesias, hyperkinetic movements, seizures, cognitive impairment, and acanthocytosis. Previous mouse models of ChAc have shown variable disease phenotypes depending on genetic background. In this study, we report the generation of a Vps13a flox allele in a pure C57BL/6N mouse background and the subsequent creation of Vps13a knockout (KO) mice via Cre-recombina-tion. Our Vps13a KO mice exhibited increased reticulocytes but not acanthocytes in peripheral blood smears. Additionally, there were no significant differences in GFAP- and Iba1-positive cells in the striatum, the basal ganglia of central nervous system. Interestingly, we observed abnormal spermat-ogenesis leading to male infertility. These findings indicate that Vps13a KO mice are valuable mod-els for studying male infertility and some hematological aspects of ChAc.Pinyomahakul, J.; Ise, M.; Kawamura, M.; Yamada, T.; Okuyama, K.; Shibata, S.; Takizawa, J.; Abe, M.; Sakimura, K.; Takebayashi, H. Analysis of Brain, Blood, and Testis Phenotypes Lacking the Vps13a Gene in C57BL/6N Mice. Int. J. Mol. Sci. 2024, 25, 7776. https://doi.org/10.3390/ijms25147776新大院博（医）第1224
脊髄後角ニューロンの抑制性伝達に対するレミマゾラムの作用
Get PDF新潟大学Niigata University博士（医学）Remimazolam is an ultra-short benzodiazepine that acts on the benzodiazepine site of γ-aminobutyric acid (GABA) receptors in the brain and induces sedation. Although GABA receptors are found localized in the spinal dorsal horn, no previous studies have reported the analgesic effects or investigated the cellular mechanisms of remimazolam on the spinal dorsal horn. Behavioral measures, immunohistochemistry, and in vitro whole-cell patch-clamp recordings of dorsal horn neurons were used to assess synaptic transmission. Intrathecal injection of remimazolam induced behavioral analgesia in inflammatory pain-induced mechanical allodynia (six rats/dose; p 0.05). This study suggests that remimazolam induces analgesia by enhancing GABAergic inhibitory transmission in the spinal dorsal horn, suggesting its potential utility as a spinal analgesic for in-flammatory pain.Rintaro Hoshino, Nobuko Ohashi, Daisuke Uta, Masayuki Ohashi, Hiroyuki Deguchi, Hiroshi Baba,
Actions of remimazolam on inhibitory transmission of rat spinal dorsal horn neurons,
Journal of Pharmacological Sciences,
Volume 155, Issue 2,
2024,
Pages 63-73,
ISSN 1347-8613,
https://doi.org/10.1016/j.jphs.2024.04.002.
(https://www.sciencedirect.com/science/article/pii/S134786132400032X)新大院博（医）第1228
妊娠中に施行された広汎子宮頸部摘出術が胎児発育及び血流波形に与える影響に関する研究
Get PDF新潟大学Niigata University博士（医学）Background: Our previous study demonstrated the safety and effectiveness of abdominal radical trachelectomy during pregnancy but did not focus on the fetus. This study aimed to clarify the influence of abdominal radical trache-lectomy performed during pregnancy on the fetus. Methods: Eight cervical cancer patients who underwent abdominal radical trachelectomy at our hospital between February 2013 and August 2020 were enrolled in this study. To assess the peri- and postoperative influence on the fetus, we performed fetal heart monitoring at 30-min intervals during abdominal radical trachelectomy and calcu-lated the estimated fetal body weight and resistance indexes of the middle cerebral artery and umbilical artery from postsurgery until delivery. Results: Four out of eight patients had preterm birth due to chorioamnionitis in one case and consideration of the recurrent risk of cervical cancer in three cases. Fetal heart monitoring during abdominal radical trachelectomy revealed deceleration just once in one case but no abnormal findings in the other cases. In all cases, the fetal growth after abdominal radical trachelectomy was normal until delivery. No abnormal Doppler findings were detected in the middle cerebral artery or umbilical artery. Conclusion: Our findings clarified that abdominal radical trachelectomy performed for the treatment of early-stage cervical cancer during pregnancy has no obvious influence on fetal growth. Next, it is necessary to evaluate the growth and development of children delivered from mothers who have undergone abdominal radical trachelectomy during pregnancy.Shima, E., Itsukaichi, M., Yoshihara, K. et al. Fetal biometric and Doppler measurements following abdominal radical trachelectomy in the second trimester of the pregnancy. BMC Pregnancy Childbirth 22, 343 (2022). https://doi.org/10.1186/s12884-022-04671-6新大博（医）第1831
機械学習に基づくデルタ地域の洪水氾濫モデルの開発
Get PDF新潟大学Niigata University博士（学術）Due to the limitations of numerical simulations in real-time flood management, this study focused on developing a machine learning-based flood inundation model utilizing Long Short-Term Memory (LSTM), particularly for delta regions. Furthermore, in response to challenges associated with acquiring high resolution elevation data required for the flood modeling, remote sensing-based digital terrain models (DTMs) were developed. The models showed reliability in three basins in Japan (Kameda, Niitsu and Shirone) located in delta regions, when compared with light detection and ranging (LiDAR) elevation data; this emphasized the applicability for flood modeling. The research further integrated these detailed topographic models into numerical simulation-based flood model and applied them to the data-scarce lowland watershed in Xuan Thuy, Vietnam, validating the model's applicability in capturing flood events. LSTM-based flood modeling was then developed for Kameda and Xuan Thuy. Simulated Monte Carlo approach was utilized to acquire various rainfall pattern data. The data were then inputted for the training process. The model was optimized and LSTM parameters were analyzed to yield optimum performance; resulting the most effective training data, number of epochs and hyperparameters. The LSTM approach outperformed;conventional numerical models in terms of computational efficiency and real-time prediction. To validate the approach, statistical metrics (R², Nash-Sutcliffe efficiency (NSE), percent bias (PBIAS)) were calculated. The results acquired for various rainfall scenarios were categorized as 'very good', thus confirming the accuracy of the model. Afterwards, the LSTM and numerical simulation-based flood models were further integrated into a web-based platform, offering educational resources and real-time predictions. Since the platform served as an accessible flood predictions and visualizations with user-friendly interface, it is potential for an effective flood preparedness and response. In conclusion, these study approaches succeeded in bridging data gaps, improving flood management strategies, and potentially increasing resilience against flood hazards, specifically within the delta regions.Rau, Maulana Ibrahim, Development of Machine Learning-Based Flood Inundation Model in Delta Regions (September 20, 2024). Available at SSRN: https://ssrn.com/abstract=5165669 or http://dx.doi.org/10.2139/ssrn.5165669新大院博(学)第234
効率的な接空間埋込みのための局所構造化ユニタリーネットワーク
Get PDF新潟大学Niigata University博士（工学）This thesis presents a novel linear transform, the Locally Structured Unitary Network (LSUN), designed to effectively capture the tangent space of manifolds latent in high-dimensional data. Chapter 1: High-dimensional data has become increasingly abundant, enabling enhanced decision-making and more precise calculations across various novel applications. However, the volume and dimensionality of this data present significant challenges. Efficient utiliza-tion requires dimensional reduction methods that can represent data in a lower-dimensional form while preserving as much information as possible. Various approaches for dimensional reduction exist, but linear tangent space embedding methods are particularly valued for their simplicity and interpretability. Nonetheless, each linear method has unique issues that affect its ability to provide an accurate tangent space representation. Chapter 2: This chapter discusses principal component analysis (PCA) and convolu-tional dictionary learning (CDL) as primary linear methods for capturing the tangent space. Localized PCA captures tangent spaces with its shift-variant structure, while CDL employs a shift-invariant structure with redundant filter kernels and sparsity-aware approximation methods. Chapter 3: High-dimensional dynamical system data naturally lying on manifold struc-tures can benefit significantly from tangent space embedding. This chapter focuses on dy-namical system modeling, emphasizing Dynamic Mode Decomposition (DMD), a widely used method. Various enhancements to DMD are explored to achieve better results. Chapter 4: The proposed LSUN structure effectively captures the tangent space of manifolds in high-dimensional data. LSUN can be interpreted as a specialized class of filter banks based on cascaded primitive block operations, where the filters are adaptively variable during training to fit the data. This approach extends the concept of localized PCA by introducing overlap and shares similarities with the convolutional layer by incorporating locality and overlap. However, unlike the shift-invariance of convolutional layers, which maintain a fixed set of filter kernels, LSUN employs shift-variance with adaptively variable filter kernels that fit the given dataset. This transform is both linear and unitary, promoting simple construction and interpretability. Additionally, the unitarity enables self-supervised training, enhancing computational efficiency. The construction of LSUN is based on M-D LPPUFB, utilizing a cascaded arrangement of block operations, including Givens rotation, shift, and butterfly operations. The angle parameters of the novel system are changed block-by-block, making them adaptively variable, while all constraints and properties embedded in the base systems are preserved despite these changes. Chapter 5: To validate our claims, we conducted a series of experiments focused on dynamical system modeling tasks. Preliminary experiments, including 1-D and 2-D data approximations, image denoising, and video denoising tasks, demonstrated LSUN’s ability to capture the underlying low-dimensional structure, outperforming conventional methods. The video denoising experiments further validated LSUN’s capability to handle time series data. In dynamical system modeling tasks, LSUN enhances the DMD process by embedding the input dataset in a low-dimensional structure. This approach leverages the natural low-dimensional structure of dynamical system data, often manifesting as a manifold. LSUN’s locality, overlap, and ability to directly correspond its learned filter kernels to the tangent space effectively reduce the computational burden and improve the accuracy of the DMD process. Chapter 6: This chapter concludes the thesis by emphasizing LSUN’s novel approach to capturing tangent spaces in high-dimensional data. Preliminary experiments validated LSUN’s performance in data approximation, image denoising, and video denoising. The study successfully applied LSUN in dynamical system modeling, reducing dimensionality and increasing the efficiency of the DMD process. Future research directions include in-vestigating the application of LSUN in modeling nonlinear conservative dynamic systems, exploring its potential to enhance interpretability in dynamical system modeling, and exper-imenting with LSUN’s behavior in complex nonlinear systems and higher-order dynamical systems. Further research in these areas could significantly advance LSUN’s capabilities and applications across various fields.新大院博(工)第565