34254 research outputs found
Sort by
Automated Structuring Transformation and Tumor Detection Based on CT Liver-Tumor Reports
No full textIn the field of bioinformatics, textual and visual recognitions are two main techniques for disease diagnosis. This thesis aims to design methods for automated report-structure transformation and tumor detection based on Computed-Tomography (CT) liver-tumor reports. Automatic report-structure transformation is motivated by the concept that a readable structure is essential to doctors for re-usage because much knowledge is contained in CT liver-tumor reports. For this purpose, a two-stage approach is proposed in this thesis, which combines Natural Language Processing (NLP) and Machine Learning (ML). In the first stage, NLP and feature selection techniques are used to extract relevant feature keywords. Then in the second stage, a multi-label classification approach is implemented to structure the reports. In addition to automated report-structure transformation, this thesis also intends to detect liver tumors automatically by simultaneously considering CT images and textual descriptions in reports. We thus propose a two-stage method including liver partition and target tumor detection to attack this problem. First, nine 3D U-Net models are adopted to partition a liver into nine corresponding regions, which were defined by Couinaud. Second, the Mask R-CNN is used to detect the tumors in the regions specified in textual reports. In the experiments, the textual structure transformation and the tumor detection are evaluated on real data from a specific hospital. The experimental results achieved an accuracy of 90.99% in automatic report-structure transformation, a dice coefficient of 75.41% for liver segmentation, and a hit rate of 80.35% for target tumor detection. The evaluation results of the proposed method outperform those of other comparative methods
Studying Heterogeneous Reaction of Sodium Sulfite/Nitrite Aerosols with Ozone via Aerosol Optical Tweezers
No full textAqueous-phase oxidation reactions within aerosols is an important process in atmospheric chemistry, including the conversion of sulfur dioxide and nitrogen oxides into sulfate and nitrate ions. These ions can act as condensation nuclei to form larger aerosols, further affecting visibility and air quality. Water-phase oxidation reactions may promote the generation and growth of aerosols, affecting photochemical reactions or precipitation patterns in the atmosphere, thereby affecting the atmospheric chemical cycle.
This study used optical tweezers to capture single aerosol particles containing either nitrite or sulfite ions and subjected them to heterogeneous oxidation reactions with ozone, monitoring with Raman spectroscopy. The aim was to observe whether these aerosols exhibited different reaction kinetics under varying relative humidity and ozone concentration conditions. By analyzing the decay of the reactant signals in Raman spectrum of single aerosol droplets, the reaction rate constants (k) were calculated.
Additionally, in low pH conditions, sulfite ions can be converted to bisulfite, so experiments are also conducted using a solution with pH of 5.0. Furthermore, the kinetic analysis model KM-SUB is used to fit the reactions of ozone and ions under different conditions, comparing the effects of surface and bulk reactions. According to the KM-SUB simulation, the surface rate constant at high humidity conditions is found to be 4.63\uc2\ub13.29 \uc3 10-11 and 3.11\uc2\ub13.22 \uc3 10-10 cm2 s\ue21 for the respective samples. Interestingly, as the ozone concentration decreases, there seems to be a trend of higher rate constants influenced by competitive adsorption. We also discuss the impact of changing desorption lifetime on the reaction kinetics. These findings provide valuable insights for a deeper understanding of aerosols and their impact on the atmosphere
Investigation of SiGe/Si Heterojunction Line Tunneling Inductive Tunnel Field-Effect Transistor with Source Schottky Contact
No full textThis thesis presents a Silicon-Germanium/Silicon Heterojunction Line Tunneling Inductive Tunnel Field-Effect Transistor with Source Schottky Contact (SC HJLT-iTFET) and a Source Schottky Contact Silicon-Germanium/Silicon Heterojunction Line Tunneling Inductive Tunnel Field-Effect Transistor with Charge Enhancement Layer (SC HJLT-CEL iTFET). The SC HJLT-iTFET device employs group 4 elements of Silicon-Germanium and Silicon heterojunctions as the channel, which are compatible with silicon-based CMOS processes. The channel thickness is designed to be 5 nm, and the ultra-thin body (UTB) is used for improving gate control ability. The source region channel material is Si0.5Ge0.5, incorporating germanium to achieve a narrow energy bandgap characteristic, while the drain region channel employs pure silicon with a wider energy bandgap for improved performance. The unique structure enhances line tunneling between the source and channel regions while minimizing point tunneling between the channel and drain, leading to increase on-state current (ION) and reduced off-state current (IOFF). The P+ source region forms a Schottky contact with appropriate metal work function to induce carrier inversion, and the full overlapping between the source electrode and gate regions can improve ON-state current and obtain a steep subthreshold swing, by line tunneling mechanism dominated. The gate consists of a bilayer structure comprising a high-k dielectric layer, Hafnium dioxide (HfO2), and a lower-k dielectric of Silicon dioxide (SiO2), serving as a blocking layer to enhance gate control over the channel region and reduce lateral electric fields between the gate and drain, respectively. Furthermore, the SC HJLT-iTFET device replaces the P-I-N junction in traditional TFETs by utilizing an M-S junction, this design eliminates the need for multiple ion implantation and annealing steps, overcomes self-alignment and random dopant fluctuations(RDF) during ion implantation, and significantly reduces thermal budget. Various analyses and optimizations are performed on the SC HJLT-iTFET to enhance its switching performance and subthreshold characteristics. The SC HJLT-iTFET device achieves a minimum subthreshold swing (SSmin) of 6.8 mV/dec and an average subthreshold swing (SSavg) of 19.8 mV/dec at a bias voltage VD of 0.2 V. Additionally, the ON/OFF current ratio (ION/IOFF ratio) is 1.02 \uc3 1010.
A second device, the SC HJLT-CEL iTFET, main structure shares the same as the previously introduced SC HJLT-iTFET. However, in the SC HJLT-iTFET, the source region is formed by a metal-semiconductor contact. In practical applications, the junction between metal and semiconductor exhibits various undesirable phenomena, including defects at the interface, dangling bonds, electric dipoles, and metal induced gap state (MIGS) leading to Fermi Level Pinning. These unfavorable effects prevent the estimation of the Schottky barrier height by using the Schottky-Mott model for the metal-semiconductor interface.
To address the Fermi Level Pinning issue and enhance the charge density induced by the Schottky contact in the source region, a charge enhancement layer (CEL) is introduced between the semiconductor channel and source metal. The CEL consists of a 1 nm thickness oxide layer (SiO2) with a length of 65 nm. Simulation and analysis results demonstrate that the addition of the charge enhancement layer effectively suppresses the Fermi Level Pinning effect and enhances the charge in the source region. In this study, the CEL section of the SC HJLT-CEL iTFET device is analyzed and optimized. Under a bias voltage of VD = 0.2 V, the SC HJLT-CEL iTFET device achieves SSmin of 7.2 mV/dec and SSavg of 21.2 mV/dec. Additionally, the device exhibits an ION/IOFF ratio of 1.28 \uc3 1011.
These features enable the device to possess excellent electrical performance, achieving high current drivability and high switching speed at extremely low supply voltage. These attributes align with the objectives of low-power semiconductor device design in Industry 4.0, which include high current drivability, low process cost, and low power consumption.
Therefore, these devices are suitable for applications in low-power consumption fields such as the Internet of Things (IoT), artificial intelligence (AI), and related domains
Accelerating Golden-Angle Radial Sampling: A Probability Density Function-Based Approach
No full textMagnetic Resonance Imaging (MRI) is a commonly used imaging technique for clinical diagnosis, which can image the whole body or localized parts of the body to obtain non-invasive images of human anatomy or physiological functions. However, its clinical use is limited by the scanning time, which takes a longer time. The main limitation is the imaging principle of MRI, which is based on the use of gradient coils to generate a gradient magnetic field in a homogeneous primary magnetic field for spatial encoding, frequency encoding and phase encoding, respectively. Phase encoding, which uses different magnetic field gradient strengths, requires multiple excitations before scanning for measurement, and therefore becomes a limiting factor in scanning time. In this study, three probability density functions are proposed for sparse sampling of Golden-Angle radial sampling. It combines the concepts of parallel imaging and compressed sensing to reconstruct sparse k-space data. This study hopes to reduce the scanning time while maintaining the image quality.
MRI 3D images are usually sampled in k-space in the frequency encoding (represented in this study by the kx direction) and phase encoding (represented in this study by the ky direction directions, but it can also be sampled in the other two phase encodings (ky,kz). Use golden angle radial sampling on two phase encoding, in addition to the acceleration method of reducing the number of spokes, reducing the number of sampling points on the spokes can also reduce the scan time. Therefore, in the study, the golden angle radial trajectory is on two phase encodings, and k-space is sparse sampled by using the probability density function.
However, when sampling by probability, low spatial frequency information may not be collected due to random sampling that misses the near-center of k-space. This affects the results of coil sensitivity estimation. It is therefore important to obtain the low spatial frequency components around the k-space origin. Meanwhile in the compressed sensing reconstruction procedure, the regularization weight between data consistency and sparsity constraints affects the image quality. In this study, a series of simulation was performed for the evaluation of the above two parameters in different image reconstruction settings. Simulated k-space data are generated with three different types of probability density functions, and the reconstructed images were then evaluated.
The results of this study show that the image can be successfully reconstructed using acceleration by reducing the number of sampling points on the spokes. Compared with the reconstruction results of full sampling, values of SSIM of these sparsity sampled methods are more than 0.9, and values of RMSE of them are less than \ue310\ue3^(-4). And among the reconstructed images of the three probability density sampling modes, the uniform mode is the best sampling method. In the comparison with the common way of reducing spokes, although the results of the fewer spokes were better in both assessment indicators (SSIM and RMSE). However, the method proposed in this study has better performance in slice analysis of different structures. In complex structural slice, the uniform mode is better one at reconstructing image details because the number of sampling points at high spatial frequencies is more than that of the fewer spokes method. There is not much difference between the two sparsity sampling methods in slices of fully uniform and tube sections. Therefore, reconstructing images using uniform mode may be advantageous for slices with more complex structures. It is possible to reduce the scan time by reducing the number of spokes and the number of sampling points on the spokes at the same time in the future
Digital Management of Intellectual Property Rights- A Case Study in Research and Development Industry
No full textNowadays, companies are very concerned about digital transformation because digital technology is changing the way and scale of global business operations. Digital transformation is a broad concept that encompasses all aspects of digital technology in business, such as automated processes, data analytics, artificial intelligence, the Internet of Things, etc. Companies are concerned about digital transformation because successful digital transformation can lead to increased competitiveness, improved customer experience, new markets, optimized internal management processes, reduced costs, and adaptation to market trends.
In most common cases of digital transformation, companies focus on cost (e.g., accelerating the digitization and automation of internal processes), sales and marketing (e.g., using tools such as cloud computing and the Internet of Things) management, but ignore the fact that intellectual property, which is an important intangible asset of the company, also needs to be managed in a digital way. In the event that digitalization becomes the market trend, enterprises that neglect the digital management of intellectual property may not be able to keep up with the pace of digitalization and lag behind their competitors, and in serious cases, this may lead to negligence in the management of intellectual property and leakage of confidential information. With the high level of digitalization, how should Taiwan enterprises manage their intellectual property?
This study takes the R&D manufacturing industry as an example to analyze and discuss the intellectual property management problems faced by the case company, and the results of the study show that since most companies (including the case company) are involved in the production process of purchasing raw materials from suppliers, internal product development and manufacturing, and collaborative development with external partners, the production process relies on cooperation with external partners, and intellectual property and confidential information must be shared with external partners in most cases. Under such circumstances, if an enterprise only focuses on internal intellectual property management without paying attention to the risk that intellectual property may be leaked by external third parties, it may expose the company's intellectual property to risk. This study hereby focuses on the intellectual property management problems faced by most companies, and uses case company as examples, and to formulate improvement strategies for intellectual property management as a reference for the future management of enterprises
Queer Literature and Feminine Consciousness : On Ye Qing's Poetry
No full textModern society often deliberates over issues of women\ue2s consciousness. Over different time periods, women\ue2s rights have gone through distinctive transformations, from parental rule and the matchmaker\ue2s authority to the pursuit of freedom to love as well as modern issues of advocating for sexual autonomy. These transformations are like every girl\ue2s development, in that each growth phase can bring about different struggles. Struggles can be around self-exploration from the inner essence to the exterior physical body, or a rush to seek societal approval, or struggles around experiences of trauma and depression. The process of reading queer literature is comparable to reflecting on the hardships and bumpy roads encountered on the path to women\ue2s equality.Is it possible that that the passions mobilized in striving for equality actually burn down the life-force of queer subjects down to their embers? Furthermore, how do outsider eyes view the queer individual\ue2s pursuit of self?
In the process of searching for the form of women\ue2s consciousness that appears in queer literature I hope to discuss queer issues, women\ue2s issues as well as issues of the self. Through the course of life exploration, we can discover that perhaps every person is capable of finding a microcosm of their self that exists within the texts of queer literature. It may be that we are all those authors who have been suppressed by their families, yearning for love and approval. Maybe we are all characters from those erotic novels who at once enjoy being unrestrained while also feeling self-conscious. And we may also all be contemporary patriarchs who clearly love our family members even while we brutally push them into the abyss. This thesis will discuss these topics. It is hoped that through balancing an inquiry into textual form and the emotions such form expresses, with a keen attention on the rhetorical dimensions of the author\ue2s work, we can reveal the imagery presented within.
This thesis mainly uses the works of Ye Qing as the focus of discussion, and through an examination of her multiple identities, investigates the social and cultural experiences of lesbian poets as well as lesbians. Ye Qing once said that one could become closer to her through the reading of her poems. Thus, looking at society through her poems, I hope to provide understanding of how members of the queer community can find love and be loved within hegemonic heterosexual society. I also hope to provide understanding of how members of the queer community find their place within their families, as well as how they view their female identity when facing their own self. This research will break new ground in this field of study, as there is not yet sufficient research done on Ye Qing\ue2s works. Thus, it can be said that the in-depth understanding of Ye Qing has value for the field of queer Taiwanese literature. Through the viewpoint of Ye Qing\ue2s poetry we can understand and analyze, with a hope to deeply explore women\ue2s consciousness within the context of queer issues. Along with a broad discussion of seminal studies including those of influential pioneers of women's consciousness such as Simone de Beauvoir as well as those of Qiu Miaojin, we will probe and examine the mysterious literary grace of queer literature and women\ue2s consciousness as they are expressed in Ye Qing\ue2s poetry
A Study on the Four Anthologies of \ue2Chu Hsi versus Lu Hsiang-shan\ue2(Chu Lu I-T\ue2ung) in the Ming Dynasty
No full textAs a major issue in the history of Chinese philosophy, \ue2Chu Hsi versus Lu Hsiang-shan\ue2(Chu Lu I-T\ue2ung\ue6\ub1\ue9\ub8\ue7\ub0\ue5) has been discussed by many scholars in different dynasties. This thesis focuses on the four anthologies of \ue2Chu Lu I-T\ue2ung\ue2 in the Ming Dynasty(\ue6\ue6), including Cheng Ming-Cheng's(\ue7\ua8\ue6\ue6\ubf,1445-1499) Tao-Yi-Pien(\ue9\ue4\ub8\ue7\ub7\ua8), Wang Yang-Ming's(\ue7\ue9\ubd\ue6, 1472-1529) The Later Chu Hsi: The Final Conclusion(\ue6\ub1\ue5\uad\ue6\ue5\ub9\ub4\ue5\uae\ue8\uab), Cheng Tung's(\ue7\ua8\ue7\ub3, 1480-1560) Hsien-Pi-Lu(\ue9\ue9\ua2\ue9), and Chen Chien's(\ue9\ub3\ue5\ubb\uba, 1497-1567) Hsueh-Pu-Tung-Pien(\ue5\uad\ub8\ue8\ue9\ue8\ube\ua8), and analyzes their viewpoints, summarizes the characteristics of \ue2Chu Lu I-T\ue2ung\ue2 in Ming Dynasty.
We find that the four anthologies can be classified into two models, including \ue2Chu and Lu were like-minded\ue2(\ue6\ub1\ue9\ub8\ue5\ue9) and \ue2Chu and Lu were non-like-minded\ue2 (\ue6\ub1\ue9\ub8\ue9\ue5\ue9). And we also discovery the friendship history of Chu Hsi and Lu Hsiang-shan became clearer along with the debate between the two models, which inspired the later generations of scholars to study Chu Hsi\ue2 whole life, making the writing of chronicle of Chu Hsi more and more complete. Furthermore, through the debate between the two models, it not only promoted the link of "Lu-Wang system"(\ue9\ub8\ue7\ue4\ub9\ue5\uad\ub8), but also inferred the distinction standard to the Philosophy of Chu Hsi and Lu Hsiang-shan, that is, \ue2The Philosophy of the Mind\ue2(Hsin-hsueh\ue5\ubf\ue5\uad\ub8) and \ue2Neo-Confucianism\ue2( Li-hsueh\ue7\ue5\uad\ub8). Last but not least, we can point out two characteristics of \ue2Chu Lu I-T\ue2ung\ue2 in the Ming Dynasty. First, those discussions of \ue2Chu Lu I-T\ue2ung\ue2 in the Ming were actually political issues, which meant that the argument between the two models pointed to two deep-rooted ideologies, although the scholars who supported that \ue2Chu and Lu were like-minded\ue2 hoped to break ideologies in an academic way but with little success; second, all of them paid more attention to build the theory of mind(\ue5\ubf\ue6\ua7\ue8\uab) and to practice in daily life, and abide by \ue2honoring the virtuous nature but also following the path of knowledge\ue2.(\ue5\ub0\ue5\ube\ub7\ue6\ua7\ue8\ue9\ue5\ue5\uad\ub8) However, the differences between Chu Hsi and Lu Hsiang-shan can still be found in the debate along with the two models: Chu Hsi more emphasized the \ue2intelligence\ue2 subject of virtue(\ue6\uba\ue6\ua7), while Lu Hsiang-shan emphasized the \ue2good nature\ue2 subject of virtue(\ue5\ue6\ua7), this inspired the different ways of learning
A 0.5 V 10-bit 2MS/s SAR ADC With Merge\ue3Split Capacitor Switching Method in 90 nm
No full textAnalog to Digital Converters (Analog to Digital Converters) are building blocks required for a variety of applications, one such application is the System On Chip for biological systems. The output signal speed of biosensors is usually very slow and occupies a small frequency range. Since physically small batteries with long-term reliable operation must be used in biomedical systems, and for ease of integration, ADCs need to be designed with low power consumption and compact layout. CMOS biomedical signal sensing device is the main component of the overall biomedical signal system and is widely used in mobile biomedical systems. In most of these architectures, ADCs are used because they provide the right trade-off of read speed and area. Single Slope ADC (SSADC), Successive Approximation Register Successive Approximation Register ADC (SAR ADC), Cyclic ADC, and sigma-delta ADC are the most popular structures used in ADCs.
This paper realizes the use of split-merge SAR ADC, which greatly reduces the power consumption of capacitor switching compared with the traditional switching method. On this basis, the capacitor with the largest weight is removed, and the extra Vref is used to compensate the capacitance weight, the circuit includes a two-stage bootstrapped S&H circuit, a strong arm comparator circuit, a clock generator for distributing clock pulses in the circuit, a voltage generation circuit and other analog circuits, and other digital circuits for SAR logic control
Examining the Impact of Cultural Values on Brand Loyalty: A Study of Collectivism and Individualism
No full textThis study examines how cultural values\ue2precisely, collectivism and individualism\ue2shape the phenomena of brand loyalty in the setting of an increasingly linked global economy. The study investigates how people from various cultural backgrounds display different levels of brand loyalty using a complex research strategy that incorporates surveys, consumer behavior analysis, and interviews. The results provide insights that are helpful for marketers who want to customize their tactics to appeal to various cultural segments, improving their capacity to cultivate long-lasting brand-consumer connections and traverse the challenges of a culturally varied customer landscape
Green Chemistry Approaches to Catalytic Conversion of Lignocellulosic Biomass and Biomass-Derived Carbohydrates into High-Value Chemicals via Sustainable and Eco-Friendly Deep Eutectic Solvents Technique
No full textThe valorization of lignocellulosic biomass and carbohydrate-based feedstock using a microwave heating system was investigated in this study. The research focused on the conversion of various biomass feedstocks, including corncob, sugar bagasse, and Thai rice straw, into important chemical compounds such as 5-HMF, furfural, and levulinic acid. The experiments were conducted under specific reaction conditions, utilizing different solvents and catalysts. For the valorization of lignocellulosic biomass, the DMSO/water solvent system with H2SO4 catalyst yielded favorable results. Corncob residues and Thai rice straw exhibited 5-HMF yields of 34.15% and 37.75%, respectively, along with furfural yields around 24.35% and 20.38% for both feedstocks, respectively. Levulinic acid yields reached 39.45% and 40.21% for these respective biomass sources. When n-butanol/water was used as the solvent with ZSM-5 as the catalyst, Thai rice straw produced 5-HMF yields of 42.25%, furfural yields of 20.36%, and levulinic acid yields of 37.19%. The recyclability of the ZSM-5 catalyst was also evaluated, demonstrating its potential for cost-effective utilization.
Regarding carbohydrate-based feedstock valorization, D-glucose was converted into 5-HMF using the microwave heating system. The experiments utilized ZSM-5, DESs, and TiO2 catalysts, with ZSM-5 yielding the highest 5-HMF yields of 89.5%, 93.6%, and 90.3% under specific reaction conditions. The high yields of 5-HMF obtained indicate the potential of D-glucose as a feedstock for the production of this valuable chemical. The recyclability of the ZSM-5 and TiO2 catalysts was also confirmed, further enhancing the economic feasibility and sustainability of the process.
The study demonstrates the process's efficiency and effectiveness, as evidenced by the high yields obtained under specific reaction conditions. These conditions include a ZSM-5 catalyst dosage of 1.5g, a temperature of 140 \uc2\ub0C, and a reaction time of 30 min. Moreover, the catalysts' recyclability enhances the economic feasibility and sustainability of the valorization approach. Additionally, the determination of the activation energy provides valuable insights for optimizing the process and facilitating scale-up efforts.
Overall, the results highlight the microwave heating system's potential for biomass valorization and the production of valuable chemicals. The efficient yields achieved under specific reaction conditions, coupled with catalyst recyclability, contribute to the process's economic and environmental sustainability. These findings pave the way for future research and industrial applications, positioning the microwave heating system as a promising technology for sustainable production from renewable biomass resources