34254 research outputs found
Sort by
Application of 3D convolutional neural network with brain magnetic resonance imaging for Alzheimer's Disease classification
No full textAlzheimer's disease is a brain disorder commonly known as dementia in the elderly. This disease leads to memory, thinking, and behavioral problems, representing an abnormal aging phenomenon. With the aging population, the number of people affected by Alzheimer's disease is increasing every year. The Taiwan Alzheimer's Disease Association estimates that by the year 2031, there will be nearly 460,000 dementia patients in Taiwan, which means that there will be two dementia patients for every 100 Taiwanese individuals.
In medical imaging, structural magnetic resonance imaging (MRI) provides detailed information about various regions of the brain and can display specific multi-directional images. In modern clinical examinations, it is possible to directly observe the overall brain structure and detect the morphology of specific brain regions, making it a cornerstone in the detection of Alzheimer's disease. Therefore, this study aims to establish a binary (normal and abnormal) and ternary (normal, mild cognitive impairment, and abnormal) model for Alzheimer's disease, assisting physicians in diagnosing and predicting the presence of Alzheimer's disease.
Structural MRI provides 3D images that are constructed from a stack of 2D images. In terms of data preprocessing, image processing techniques such as skull stripping and registration are applied to each participant's data. For data collection, preprocessed 2D slices are used as the dataset, which helps save memory and accelerate model computations. The dataset is then divided into training and testing sets, and one of two types of 3D convolutional neural networks, such as the reference 3D ResNet architecture, is trained. Data augmentation techniques are employed during training to enhance the model's generalization ability. The probabilities for each class are calculated using the softmax function, and the model's performance is evaluated based on accuracy. Additionally, the Gradient-weighted Class Activation Mapping (Grad-CAM) technique is used to visualize the regions of the patient's brain that are most relevant according to the 3D convolutional neural network
Study on Chemical Compositions of the Sponge Neopetrosia proxima and the Octocorals Briareum stechei and B. violaceum
No full textIn this study, the chemical compositions of the sponge Neopetrosia proxima and the octocorals Briareum stechei and B. violaceum were screened, which led to the isolation of 58 metabolites, including 18 new compounds. From the sponge N. proxima, a new manzamine alkaloid, manzamine A 27-N-oxide (1), and five known analogues (2\ue26) were obtained. From two samples of the coral B. stechei, ten new briarane diterpenoids, briarenols U (7) and V (8), briastecholides A\ue2C (9\ue211), G\ue2J (19\ue222) and M (23), along with other 17 reported briaranes (12\ue218 and 24\ue233) were isolated. From the cultured coral B. violaceum, seven new briarane diterpenoids, briavioids A\ue2G (34\ue240), together with 18 known briaranes (41\ue258) were obtained. The structures of isolated compounds were elucidated using spectroscopic methods. The absolute configurations of 1, 6, 13, 17, 24, 34, 37 and 56 were established by the single-crystal X-ray diffraction analysis. The stereochemistry of solenolide B (15) was revised based on its NMR data.
Manzamines 1\ue23 were tested in regard to their cytotoxic effects on Hep3B, SC-M1 and MCF-7 tumor cell lines. In the results, manzamines 1 and 3 moderately reduced cell viability of these cell lines, while manzamine 2 showed significant cytotoxicity against these cell lines with the IC50 values ranging from 3.70 to 4.10 \uce\ubcM. The inhibitory ability of briarane diterpenoids against the iNOS and COX-2 proteins were tested at 10 \uce\ubcM. The results showed that briaranes 19, 20, 24, 27, 29, 30 and 57 significantly suppressed the iNOS protein to 8.34, 17.01, 37.95, 28.76, 1.21, 26.66 and 28.60%, respectively; while briaranes 8, 11, 15, 22, 23, 33\ue236, 38, 39, 41, 42, 51, 54 and 56 moderately reduced this protein with the expression ranging from 51.60 to 83.29%. And briaranes 19 and 29 reduced the COX-2 protein to 63.52 and 79.00%, respectively
A Study on Return Factors in Initial Coin Offerings
No full textThis study collected data on tokens issued between September 2015 and January 2023, resulting in a sample size of 3,527 tokens. The main focus of the research was to analyze the impact of token issuance characteristics and team member features on the success of token issuance, as well as return rates and volatility of token prices. The empirical results revealed that the number of top management team members had a positive influence on the completion and amount of funds raised during the token issuance process. Additionally, tokens that set a minimum fundraising threshold had a lower likelihood of being successfully completed. Regarding return rates, the study found that the return rates of mainstream cryptocurrencies significantly influenced the short-term and long-term returns of tokens. Furthermore, it was observed that the first day return of tokens had a negative impact on their long-term returns, suggesting a long-term reversal effect in token returns. The volatility of tokens was influenced by the racial composition of team members and the volatility of mainstream tokens. Issuing a white paper and raising a larger amount of funds were found to reduce long-term price volatility, thus mitigating risks
Joint Channel Estimation and Reflection Optimization for Energy Harvested RIS-assisted OFDM Systems with Reinforcement Learning
No full textThe reconfigurable intelligence surface (RIS) is a low-cost and low-power consumption structure used to implement a full-duplex relaying system. In this thesis, we address the energy harvesting (EH) RIS-assisted orthogonal frequency division multiplexing (OFDM) system for achieving green communications. In the considered network, both the User Equipment (UE) and the RIS node are equipped with EH devices that can harvest energy from the environment to power consumption. Our goal is to jointly optimize the transmit power and reflection strategy to maximize the overall transmission rate. Particularly, we consider the channel estimation of the active reflectors in the reflection strategy. The associated optimization problem is challenging due to causality constraints and the non-convex formulation. Therefore, we adopt the reinforcement learning (RL) algorithm for the optimization process. To effectively handle the continuous action space and accelerate the training speed, we develop an actor-critic policy gradient (ACPG) RL algorithm to quickly learn suboptimal solutions. Simulation results validate the proposed design and demonstrate its ability to adaptively allocate transmit power, gradually improving the total transmission rate based on the observed channel qualities, battery status, and the harvested energy at the UE and RIS nodes
A harmony or a chaos? Relationship between family control and financial performance by using kinship ties in public Taiwanese family enterprises as moderators
No full textFrom the perspective of socioemotional wealth theory, it is found that kinship ties in family board can positively reinforce positive effects brought by family control. However, from the perspective of conflicts, the board can be turned into chaos when moderated by kinship ties such as transgenerational ties, sibling ties and in-law ties with effect of negative effects, decreased but positive effects and increased but positive effects, respectively. This study closes the gap by using hand-collecting data of family identity of board members in 499 Taiwanese public family firms. Results from hierarchical analysis reveal critical information. Transgenerational ties indeed significantly turn positive effects from family control to financial performance represented by ROA into negative relationship, while negative and positive effects by sibling ties and in-law ties are unsupportive accordingly. Future research is suggested that panel data from multiple years and in-depth interviews on the board members can be interactively applied from the aspect of SEW in \ue2E\ue2 to investigate relevance among kinship ties, family members\ue2 emotional attachments and firm performance
Reengineering of Search and Seizure Process\ue2A case of Embedding the Evidence Chain of Custody into the original Information System
No full textThis study aims to explore the application of information systems in the search and seizure processes of judicial agencies, and elucidate their impact on business process reengineering and efficiency improvement. Firstly, the necessity and motivation for process reengineering in judicial law enforcement agencies are addressed. Subsequently, central concepts such as search and seizure and evidence management are introduced, and the impacts of information system implementation on users, managers, and information system personnel are analyzed.
Users are able to handle cases more efficiently through the information system, avoiding duplicate work caused by paper-based records. Managers can supervise and analyze processes through system logs and records, providing more accurate decision-making support. Information system personnel need to consider the flexibility and scalability of system design, as well as factors related to information security.
Lastly, the paper emphasizes the importance of information systems in the judicial system, promoting cross-agency data communication and enhancing case-related processes. However, effective operation requires organizations to prioritize grassroots education, ensure intuitive system design, and comply with laws and regulations.
The application of information systems in judicial agencies significantly impacts business process reengineering and efficiency improvement. Attention should also be given to information security, outsourcing vendor influence, and data communication. The study provides plans and recommendations to help judicial agencies adapt to information system implementation and achieve optimal outcomes
Beam Search Design in MISO-OFDM OTA Test Systems
No full textThis thesis aims to design beam tracking algorithms in multiple input single output-orthogonal frequency division multiplexing (MISO-OFDM) systems and measure the beamforming performance of a 4x4 millimeter-wave RF module in Over-The-Air (OTA) testing. Due to hardware limitations and considering a mobile communication scenario, a single-user transmission mode is adopted in the communication environment. In the system, the transmit-side codebook is designed based on different beamforming patterns corresponding to different number of antennas. The entire codebook is then searched hierarchically through time division to find the best analog precoding vectors corresponding to different direction of beams by calculating the average signal energy of different codewords at the receiver. The optimal codeword information is then fed back to the transmitter through a feedback link, and then transmitter use the best beam for signal transmission. In the subsequent process, if the signal energy falling below a threshold, an local beam search is performed immediately and then switch to neighboring beams to ultimately achieve the goal of beam tracking in mobile communication. This design is implemented on a M3force platform to test its feasibility. In addition, computer simulations are conducted under different environments and parameter settings to compare the time consumption and system performance between the algorithms and exhaustive codebook search method in beam tracking. The results show that the average channel capacity of low-resolution hierarchical codebook search proposed in this paper can reach 91.94% of the upper bound of channel capacity, while the exhaustive codebook search method only achieves 56.42% of the performance under a scenario with 16 antenna configurations, a user moving speed of 40 km/hr, and a signal-to-noise ratio of 20dB. The corresponding beam search costs are 6.14% and 11.21%, respectively. The algorithm presented in this paper exhibits better search efficiency and achieves more accurate beam direction matching
Investigating Business Strategy and Management in Rural Health Care: A Case Study of a Hospital in Pingtung Country
No full textThe existence of medical centers in rural areas is necessary, even if there are factors such as uneven distribution of medical resources, low medical accessibility, and inconvenient transportation. In the spirit of maintaining the health and well-being of all people, regardless of the level of economic development in the region, the primary goal of a healthcare system is to meet the health needs of the population it serves. The existence of hospitals in remote areas has a unique and sacred mission. In recent years, due to the aging of Taiwan's population structure and the emergence of new medical technologies, Taiwan's medical industry environment has changed to a certain extent. In addition, there are fundamental medical difficulties in rural areas. Only by adjusting can the organization remain competitive. This study takes a remote hospital in Pingtung County, Taiwan as the research subject, and explores how the current external environment changes and how to use the existing advantages and disadvantages within the organization to generate the direction of future business strategies. The study found that the case hospital can continue to operate locally, such as: providing professional medical services and 24-hour emergency medical care; providing complete medical services for elderly patients; and continuing to organize community care activities. Through these methods, the case can build the sense of identity and trust of the local people. Furthermore, the case needs to look forward to the next 30 years, through the establishment of satellite clinics, the continuous upgrading of medical equipment, the development of more diverse and comprehensive self-funded projects, and the development of employee education and training programs and training systems to continuously optimize and improve medical services
Economic, China Factors, and Democratic Support: An Analysis on Taiwan Young Netizens (2018-2019)
No full textIn recent years, democratic politics globally have faced unprecedented challenges. Whether it is the democratic recession phenomenon pointed out by Diamond or the concept of democratic deconsolidation proposed by Foa and Mounk, they undoubtedly serve as warnings to democratic politics. However, have the findings of Western scholars in recent years also emerged in Taiwan? Do the Taiwanese people have unwavering faith in democracy? What are the possible factors influencing the support for democracy among the Taiwanese population?
This study aims to explore whether the support for democracy among young Taiwanese netizens between 2018 and 2019 is influenced by economic factors and the China factor. Online surveys are used as the data source for empirical verification and hypothesis construction. The first stage employs chi-square tests to examine the correlation between independent variables and absolute support for democracy. In the second stage, binary logistic regression analysis is conducted to validate the hypotheses of this study, examining whether the variables have the expected impact on Taiwanese voters' absolute support for democracy.
The research findings reveal that the evaluation of economic conditions, post-unification economic life evaluation, and acceptance of conditional unification are all significant factors influencing Taiwanese people's absolute support for democracy. These findings once again confirm the influence of economic and China-related factors on the support for democracy among the Taiwanese population
Discussion on the Competitive Strategy of Taiwan's PCB Industry-A Case Study of Company Z
No full textPrinted Circuit Board (PCB) is a substrate used in the assembly of electronic components and is often referred to as the "mother of electronic products." Taiwan's PCB industry holds a significant position in the global electronic supply chain and has made substantial contributions to the domestic economy and the advancement of the electronics industry.
This study investigates the competitive strategies in Taiwan's PCB industry, with Company Z chosen as the representative case due to its leading position in the industry. Through case study methodology and analysis of secondary data, further research is conducted on the industry and the selected company. Utilizing Resource-Based View and Porter Five Forces Analysis as frameworks, the study explores how the case company creates competitive advantages by examining internal resources and external environments.
Business strategies and competitive advantages are mutually interrelated and influential. Companies that formulate and execute appropriate business strategies aligned with their goals and market environments can achieve competitive advantages and stand out in the market. Simultaneously, maintaining and enhancing competitive advantages require companies to adjust and optimize their business strategies based on changing external and internal conditions to sustain their competitiveness