Search results for: mage segmentation

Authors: Ja-Keoung Koo, Kensuke Nakamura, Hyohun Kim, Dongwha Shin, Yeonseok Kim, Ji-Su Ahn, Byung-Woo Hong

Abstract:

The machine learning techniques based on a convolutional neural network (CNN) have been actively developed and successfully applied to a variety of image analysis tasks including reconstruction, noise reduction, resolution enhancement, segmentation, motion estimation, object recognition. The classical visual information processing that ranges from low level tasks to high level ones has been widely developed in the deep learning framework. It is generally considered as a challenging problem to derive visual interpretation from high dimensional imagery data. A CNN is a class of feed-forward artificial neural network that usually consists of deep layers the connections of which are established by a series of non-linear operations. The CNN architecture is known to be shift invariant due to its shared weights and translation invariance characteristics. However, it is often computationally intractable to optimize the network in particular with a large number of convolution layers due to a large number of unknowns to be optimized with respect to the training set that is generally required to be large enough to effectively generalize the model under consideration. It is also necessary to limit the size of convolution kernels due to the computational expense despite of the recent development of effective parallel processing machinery, which leads to the use of the constantly small size of the convolution kernels throughout the deep CNN architecture. However, it is often desired to consider different scales in the analysis of visual features at different layers in the network. Thus, we propose a CNN model where different sizes of the convolution kernels are applied at each layer based on the random projection. We apply random filters with varying sizes and associate the filter responses with scalar weights that correspond to the standard deviation of the random filters. We are allowed to use large number of random filters with the cost of one scalar unknown for each filter. The computational cost in the back-propagation procedure does not increase with the larger size of the filters even though the additional computational cost is required in the computation of convolution in the feed-forward procedure. The use of random kernels with varying sizes allows to effectively analyze image features at multiple scales leading to a better generalization. The robustness and effectiveness of the proposed CNN based on random kernels are demonstrated by numerical experiments where the quantitative comparison of the well-known CNN architectures and our models that simply replace the convolution kernels with the random filters is performed. The experimental results indicate that our model achieves better performance with less number of unknown weights. The proposed algorithm has a high potential in the application of a variety of visual tasks based on the CNN framework. Acknowledgement—This work was supported by the MISP (Ministry of Science and ICT), Korea, under the National Program for Excellence in SW (20170001000011001) supervised by IITP, and NRF-2014R1A2A1A11051941, NRF2017R1A2B4006023.

Keywords: deep learning, convolutional neural network, random kernel, random projection, dimensionality reduction, object recognition

Procedia PDF Downloads 290

Authors: Scott M. Black, Craig Maclean, Pauline Hall Barrientos, Konstantinos Ritos, Asimina Kazakidi

Abstract:

Aortic aneurysms and dissections are leading causes of death in cardiovascular disease. Both inevitably lead to hemodynamic instability without surgical intervention in the form of vascular stent-graft deployment. An accurate description of the aortic geometry and blood flow in patient-specific cases is vital for treatment planning and long-term success of such grafts, as they must generate physiological branch perfusion and in-stent hemodynamics. The aim of this study was to create patient-specific computational fluid dynamics (CFD) models through a multi-modality, multi-dimensional approach with boundary condition optimization to predict branch flow rates and in-stent hemodynamics in custom stent-graft configurations. Three-dimensional (3D) thoracoabdominal aortae were reconstructed from four-dimensional flow-magnetic resonance imaging (4D Flow-MRI) and computed tomography (CT) medical images. The former employed a novel approach to generate and enhance vessel lumen contrast via through-plane velocity at discrete, user defined cardiac time steps post-hoc. To produce patient-specific boundary conditions (BCs), the aortic geometry was reduced to a one-dimensional (1D) model. Thereafter, a zero-dimensional (0D) 3-Element Windkessel model (3EWM) was coupled to each terminal branch to represent the distal vasculature. In this coupled 0D-1D model, the 3EWM parameters were optimized to yield branch flow waveforms which are representative of the 4D Flow-MRI-derived in-vivo data. Thereafter, a 0D-3D CFD model was created, utilizing the optimized 3EWM BCs and a 4D Flow-MRI-obtained inlet velocity profile. A sensitivity analysis on the effects of stent-graft configuration and BC parameters was then undertaken using multiple stent-graft configurations and a range of distal vasculature conditions. 4D Flow-MRI granted unparalleled visualization of blood flow throughout the cardiac cycle in both the pre- and postsurgical states. Segmentation and reconstruction of healthy and stented regions from retrospective 4D Flow-MRI images also generated 3D models with geometries which were successfully validated against their CT-derived counterparts. 0D-1D coupling efficiently captured branch flow and pressure waveforms, while 0D-3D models also enabled 3D flow visualization and quantification of clinically relevant hemodynamic parameters for in-stent thrombosis and graft limb occlusion. It was apparent that changes in 3EWM BC parameters had a pronounced effect on perfusion distribution and near-wall hemodynamics. Results show that the 3EWM parameters could be iteratively changed to simulate a range of graft limb diameters and distal vasculature conditions for a given stent-graft to determine the optimal configuration prior to surgery. To conclude, this study outlined a methodology to aid in the prediction post-surgical branch perfusion and in-stent hemodynamics in patient specific cases for the implementation of custom stent-grafts.

Keywords: 4D flow-MRI, computational fluid dynamics, vascular stent-grafts, windkessel

Procedia PDF Downloads 181

Authors: Samil Ozcan

Abstract:

Airlines companies target different customer segments in consideration of pricing, service quality, flight network, etc. and their brand positioning accords with the marketization strategies developed in the same direction. The object of this study, Turkish Airlines, has many peculiarities regarding its brand positioning as compared to its rivals in the sector. In the first place, it appeals to a global customer group because of its Star Alliance membership and its broad flight network with 315 destination points. The second group in its customer segmentation includes domestic customers. For this group, the company follows a marketing strategy that plays to local culture and accentuates the image of Turkishness as an emotional allurement. The advertisements and publicity projects designed in this regard put little emphasis on the service quality the company offers to its clients; it addresses the emotions of the consumers rather than individual benefits and relies on the historical memory of the nation and shared cultural values. This study examines the publicity work which aims at the second segment customer group focusing on Turkish Airlines’ 85th Anniversary Commercial through a symbolic meaning analysis approach. The commercial presents six stories with undertones of nationalism in its theme. Nationalism is not just the product of collective interests based on reason but a result of patriotism in the sense of loyalty to state and nation and love of ethnic belonging. While nationalism refers to concrete notions such as blood tie, common ancestor, shared history, it is not the actuality of these notions that it draws its real strength but the emotions invested in them. The myths of origin, the idea of common homeland, boundary definitions, and symbolic acculturation have instrumental importance in the development of these commonalities. The commercial offers concrete examples for an analysis of Connor’s definition of nationalism based on emotions. Turning points in the history of the Turkish Republic and the historical mission Turkish Airlines undertook in these moments are narrated in six stories in the commercial with a highly emotional theme. These emotions, in general, depend on collective memory generated by national consciousness. Collective memory is not simply remembering the past. It is constructed through the reconstruction and reinterpretation of the past in the present moment. This study inquires the motivations behind the nationalist emotions generated within the collective memory by engaging with the commercial released for the 85th anniversary of Turkish Airlines as the object of analysis. Symbols and myths can be read as key concepts that reveal the relation between 'identity and memory'. Because myths and symbols do not merely reflect on collective memory, they reconstruct it as well. In this sense, the theme of the commercial defines the image of Turkishness with virtues such as self-sacrifice, helpfulness, humanity, and courage through a process of meaning creation based on symbolic mythologizations like flag and homeland. These virtues go beyond describing the image of Turkishness and become an instrument that defines and gives meaning to Turkish identity.

Keywords: collective memory, emotions, identity, nationalism

Procedia PDF Downloads 153

Authors: Pomozova Kseniia, Gorlachev Gennadiy, Chernyaev Aleksandr, Golanov Andrey

Abstract:

In clinical practice, and especially in stereotactic radiosurgery planning, the significance of diffusion-weighted imaging (DWI) is growing. This makes the existence of software capable of quickly processing and reliably visualizing diffusion data, as well as equipped with tools for their analysis in terms of different tasks. We are developing the «MRDiffusionImaging» software on the standard C++ language. The subject part has been moved to separate class libraries and can be used on various platforms. The user interface is Windows WPF (Windows Presentation Foundation), which is a technology for managing Windows applications with access to all components of the .NET 5 or .NET Framework platform ecosystem. One of the important features is the use of a declarative markup language, XAML (eXtensible Application Markup Language), with which you can conveniently create, initialize and set properties of objects with hierarchical relationships. Graphics are generated using the DirectX environment. The MRDiffusionImaging software package has been implemented for processing diffusion magnetic resonance imaging (dMRI), which allows loading and viewing images sorted by series. An algorithm for "masking" dMRI series based on T2-weighted images was developed using a deformable surface model to exclude tissues that are not related to the area of interest from the analysis. An algorithm of distortion correction using deformable image registration based on autocorrelation of local structure has been developed. Maximum voxel dimension was 1,03 ± 0,12 mm. In an elementary brain's volume, the diffusion tensor is geometrically interpreted using an ellipsoid, which is an isosurface of the probability density of a molecule's diffusion. For the first time, non-parametric intensity distributions, neighborhood correlations, and inhomogeneities are combined in one segmentation of white matter (WM), grey matter (GM), and cerebrospinal fluid (CSF) algorithm. A tool for calculating the coefficient of average diffusion and fractional anisotropy has been created, on the basis of which it is possible to build quantitative maps for solving various clinical problems. Functionality has been created that allows clustering and segmenting images to individualize the clinical volume of radiation treatment and further assess the response (Median Dice Score = 0.963 ± 0,137). White matter tracts of the brain were visualized using two algorithms: deterministic (fiber assignment by continuous tracking) and probabilistic using the Hough transform. The proposed algorithms test candidate curves in the voxel, assigning to each one a score computed from the diffusion data, and then selects the curves with the highest scores as the potential anatomical connections. White matter fibers were visualized using a Hough transform tractography algorithm. In the context of functional radiosurgery, it is possible to reduce the irradiation volume of the internal capsule receiving 12 Gy from 0,402 cc to 0,254 cc. The «MRDiffusionImaging» will improve the efficiency and accuracy of diagnostics and stereotactic radiotherapy of intracranial pathology. We develop software with integrated, intuitive support for processing, analysis, and inclusion in the process of radiotherapy planning and evaluating its results.

Keywords: diffusion-weighted imaging, medical imaging, stereotactic radiosurgery, tractography

Procedia PDF Downloads 85

Authors: Bensaid A., Mostephaoui T., Nedjai R.

Abstract:

Nowadays, wetlands are the subject of contradictory debates opposing scientific, political and administrative meanings. Indeed, given their multiple services (drinking water, irrigation, hydrological regulation, mineral, plant and animal resources...), wetlands concentrate many socio-economic and biodiversity issues. In some regions, they can cover vast areas (>100 thousand ha) of the landscape, such as the Camargue area in the south of France, inside the Rhone delta. The high biological productivity of wetlands, the strong natural selection pressures and the diversity of aquatic environments have produced many species of plants and animals that are found nowhere else. These environments are tremendous carbon sinks and biodiversity reserves depending on their age, composition and surrounding environmental conditions, wetlands play an important role in global climate projections. Covering more than 3% of the earth's surface, wetlands have experienced since the beginning of the 1990s a tremendous revival of interest, which has resulted in the multiplication of inventories, scientific studies and management experiments. The geographical and physical characteristics of the wetlands of the central region conceal a large number of natural habitats that harbour a great biological diversity. These wetlands, one of the natural habitats, are still influenced by human activities, especially agriculture, which affects its layout and functioning. In this perspective, decision-makers need to delimit spatial objects (natural habitats) in a certain way to be able to take action. Thus, wetlands are no exception to this rule even if it seems to be a difficult exercise to delimit a type of environment as whose main characteristic is often to occupy the transition between aquatic and terrestrial environment. However, it is possible to map wetlands with databases, derived from the interpretation of photos and satellite images, such as the European database Corine Land cover, which allows quantifying and characterizing for each place the characteristic wetland types. Scientific studies have shown limitations when using high spatial resolution images (SPOT, Landsat, ASTER) for the identification and characterization of small wetlands (1 hectare). To address this limitation, it is important to note that these wetlands generally represent spatially complex features. Indeed, the use of very high spatial resolution images (>3m) is necessary to map small and large areas. However, with the recent evolution of artificial intelligence (AI) and deep learning methods for satellite image processing have shown a much better performance compared to traditional processing based only on pixel structures. Our research work is also based on spectral and textural analysis on THR images (Spot and IRC orthoimage) using two object-oriented approaches, the nearest neighbour approach (k-NN) and the Super Vector Machine approach (SVM). The k-NN approach gave good results for the delineation of wetlands (wet marshes and moors, ponds, artificial wetlands water body edges, ponds, mountain wetlands, river edges and brackish marshes) with a kappa index higher than 85%.

Keywords: land development, GIS, sand dunes, segmentation, remote sensing

Procedia PDF Downloads 72

Authors: Berna Çalışkan

Abstract:

The integrated water management is a colloborative approach to planning that brings together institutions that influence all elements of the water cycle, waterways, watershed characteristics, wetlands, ponds, lakes, floodplain areas, stream channel structure. It encourages collaboration where it will be beneficial and links between water planning and other planning processes that contribute to improving sustainable urban development and liveability. Hydraulic considerations can influence the selection of a highway corridor and the alternate routes within the corridor. widening a roadway, replacing a culvert, or repairing a bridge. Because of this, the type and amount of data needed for planning studies can vary widely depending on such elements as environmental considerations, class of the proposed highway, state of land use development, and individual site conditions. The extraction of drainage networks provide helpful preliminary drainage data from the digital elevation model (DEM). A case study was carried out using the Arc Hydro extension within ArcGIS in the study area. It provides the means for processing and presenting spatially-referenced Stream Model. Study area’s flow routing, stream levels, segmentation, drainage point processing can be obtained using DEM as the 'Input surface raster'. These processes integrate the fields of hydrologic, engineering research, and environmental modeling in a multi-disciplinary program designed to provide decision makers with a science-based understanding, and innovative tools for, the development of interdisciplinary and multi-level approach. This research helps to manage transport project planning and construction phases to analyze the surficial water flow, high-level streams, wetland sites for development of transportation infrastructure planning, implementing, maintenance, monitoring and long-term evaluations to better face the challenges and solutions associated with effective management and enhancement to deal with Low, Medium, High levels of impact. Transport projects are frequently perceived as critical to the ‘success’ of major urban, metropolitan, regional and/or national development because of their potential to affect significant socio-economic and territorial change. In this context, sustaining and development of economic and social activities depend on having sufficient Water Resources Management. The results of our research provides a workflow to build a stream network how can classify suitability map according to stream levels. Transportation projects establish, develop, incorporate and deliver effectively by selecting best location for reducing construction maintenance costs, cost-effective solutions for drainage, landslide, flood control. According to model findings, field study should be done for filling gaps and checking for errors. In future researches, this study can be extended for determining and preventing possible damage of Sensitive Areas and Vulnerable Zones supported with field investigations.

Keywords: water resources management, hydro tool, water protection, transportation

Procedia PDF Downloads 56

Authors: I. Stathopoulos, V. Syrgiamiotis, E. Karavasilis, A. Ploussi, I. Nikas, C. Hatzigiorgi, K. Platoni, E. P. Efstathopoulos

Abstract:

Introduction: Brain and central nervous system (CNS) tumors form the second most common group of cancer in children, accounting for 30% of all childhood cancers. MRI is the key imaging technique used for the visualization and management of pediatric brain tumors. Initial characterization of tumors from MRI scans is usually performed via a radiologist’s visual assessment. However, different brain tumor types do not always demonstrate clear differences in visual appearance. Using only conventional MRI to provide a definite diagnosis could potentially lead to inaccurate results, and so histopathological examination of biopsy samples is currently considered to be the gold standard for obtaining definite diagnoses. Machine learning is defined as the study of computational algorithms that can use, complex or not, mathematical relationships and patterns from empirical and scientific data to make reliable decisions. Concerning the above, machine learning techniques could provide effective and accurate ways to automate and speed up the analysis and diagnosis for medical images. Machine learning applications in radiology are or could potentially be useful in practice for medical image segmentation and registration, computer-aided detection and diagnosis systems for CT, MR or radiography images and functional MR (fMRI) images for brain activity analysis and neurological disease diagnosis. Purpose: The objective of this study is to provide an automated tool, which may assist in the imaging evaluation and classification of brain neoplasms in pediatric patients by determining the glioma type, grade and differentiating between different brain tissue types. Moreover, a future purpose is to present an alternative way of quick and accurate diagnosis in order to save time and resources in the daily medical workflow. Materials and Methods: A cohort, of 80 pediatric patients with a diagnosis of posterior fossa tumor, was used: 20 ependymomas, 20 astrocytomas, 20 medulloblastomas and 20 healthy children. The MR sequences used, for every single patient, were the following: axial T1-weighted (T1), axial T2-weighted (T2), FluidAttenuated Inversion Recovery (FLAIR), axial diffusion weighted images (DWI), axial contrast-enhanced T1-weighted (T1ce). From every sequence only a principal slice was used that manually traced by two expert radiologists. Image acquisition was carried out on a GE HDxt 1.5-T scanner. The images were preprocessed following a number of steps including noise reduction, bias-field correction, thresholding, coregistration of all sequences (T1, T2, T1ce, FLAIR, DWI), skull stripping, and histogram matching. A large number of features for investigation were chosen, which included age, tumor shape characteristics, image intensity characteristics and texture features. After selecting the features for achieving the highest accuracy using the least number of variables, four machine learning classification algorithms were used: k-Nearest Neighbour, Support-Vector Machines, C4.5 Decision Tree and Convolutional Neural Network. The machine learning schemes and the image analysis are implemented in the WEKA platform and MatLab platform respectively. Results-Conclusions: The results and the accuracy of images classification for each type of glioma by the four different algorithms are still on process.

Keywords: image classification, machine learning algorithms, pediatric MRI, pediatric oncology

Procedia PDF Downloads 149

Authors: Varinder Kaur Gambhir, Neema Gupta, Sonal Tickoo Chaudhuri

Abstract:

Bengaluru, a rapidly growing metropolis in India, with a population of 12.5 million citizens, generates 5,757 metric tonnes of solid waste per day. Despite their invaluable contribution to waste management, society and the economy, waste pickers face significant stigma, suspicion and contempt and are left with a sense of shame about their work. In this context, BBC Media Action was funded by the H&M Foundation to develop a 3-year multi-phase social media campaign to shift perceptions of waste picking and informal waste pickers amongst the Bengaluru population. Research has been used to inform project strategy and adaptation, at all stages. Formative research to inform campaign strategy used mixed methods– 14 focused group discussions followed by 406 online surveys – to explore people’s knowledge of, and attitudes towards waste pickers, and identify potential barriers and motivators to changing perceptions. Use of qualitative techniques like metaphor maps (using bank of pictures rather than direct questions to understand mindsets) helped establish the invisibility of informal waste pickers, and the quantitative research enabled audience segmentation based on attitudes towards informal waste pickers. To pretest the campaign idea, eight I-GDs (individual interaction followed by group discussions) were conducted to allow interviewees to first freely express their feelings individually, before discussing in a group. Robert Plucthik’s ‘wheel of emotions’ was used to understand audience’s emotional response to the content. A robust monitoring and evaluation is being conducted (baseline and first phase of monitoring already completed) using a rotating longitudinal panel of 1,800 social media users (exposed and unexposed to the campaign), recruited face to face and representative of the social media universe of Bengaluru city. In addition, qualitative in-depth interviews are being conducted after each phase to better understand change drivers. The research methodology and ethical protocols for impact evaluation have been independently reviewed by an Institutional Review Board. Formative research revealed that while waste on the streets is visible and is of concern to the public, informal waste pickers are virtually ‘invisible’, for most people in Bengaluru Pretesting research revealed that the creative outputs evoked emotions like acceptance and gratitude towards waste-pickers, suggesting that the content had the potential to encourage attitudinal change. After the first phase of campaign, social media analytics show that #Invaluables content reached at least 2.6 million unique people (21% of the Bengaluru population) through Facebook and Instagram. Further, impact monitoring results show significant improvements in spontaneous awareness of different segments of informal waste pickers ( such as sorters at scrap shops or dry waste collection centres -from 10% at baseline to 16% amongst exposed and no change amongst unexposed), recognition that informal waste pickers help the environment (71% at baseline to 77% among exposed and no change among unexposed) and greater discussion about informal waste pickers among those exposed (60%) as against not exposed (49%). Using the insights from this research, the planned social media intervention is designed to increase the visibility of and appreciation for the work of waste pickers in Bengaluru, supporting a more inclusive society.

Keywords: awareness, discussion, discrimination, informal waste pickers, invisibility, social media campaign, waste management

Procedia PDF Downloads 107

Authors: Parag Katira, Frederika Rentzeperis, Zuzanna Nowicka, Giada Fiandaca, Thomas Veith, Jack Farinhas, Noemi Andor

Abstract:

Resource limitations shape the outcome of competitions between genetically heterogeneous pre-malignant cells. One example of such heterogeneity is in the ploidy (DNA content) of pre-malignant cells. A whole-genome duplication (WGD) transforms a diploid cell into a tetraploid one and has been detected in 28-56% of human cancers. If a tetraploid subclone expands, it consistently does so early in tumor evolution, when cell density is still low, and competition for nutrients is comparatively weak – an observation confirmed for several tumor types. WGD+ cells need more resources to synthesize increasing amounts of DNA, RNA, and proteins. To quantify resource limitations and how they relate to ploidy, we performed a PAN cancer analysis of WGD, PET/CT, and MRI scans. Segmentation of >20 different organs from >900 PET/CT scans were performed with MOOSE. We observed a strong correlation between organ-wide population-average estimates of Oxygen and the average ploidy of cancers growing in the respective organ (Pearson R = 0.66; P= 0.001). In-vitro experiments using near-diploid and near-tetraploid lineages derived from a breast cancer cell line supported the hypothesis that DNA content influences Glucose- and Oxygen-dependent proliferation-, death- and migration rates. To model how subpopulations with variable DNA content compete in the resource-limited environment of the human brain, we developed a stochastic state-space model of the brain (S3MB). The model discretizes the brain into voxels, whereby the state of each voxel is defined by 8+ variables that are updated over time: stiffness, Oxygen, phosphate, glucose, vasculature, dead cells, migrating cells and proliferating cells of various DNA content, and treat conditions such as radiotherapy and chemotherapy. Well-established Fokker-Planck partial differential equations govern the distribution of resources and cells across voxels. We applied S3MB on sequencing and imaging data obtained from a primary GBM patient. We performed whole genome sequencing (WGS) of four surgical specimens collected during the 1ˢᵗ and 2ⁿᵈ surgeries of the GBM and used HATCHET to quantify its clonal composition and how it changes between the two surgeries. HATCHET identified two aneuploid subpopulations of ploidy 1.98 and 2.29, respectively. The low-ploidy clone was dominant at the time of the first surgery and became even more dominant upon recurrence. MRI images were available before and after each surgery and registered to MNI space. The S3MB domain was initiated from 4mm³ voxels of the MNI space. T1 post and T2 flair scan acquired after the 1ˢᵗ surgery informed tumor cell densities per voxel. Magnetic Resonance Elastography scans and PET/CT scans informed stiffness and Glucose access per voxel. We performed a parameter search to recapitulate the GBM’s tumor cell density and ploidy composition before the 2ⁿᵈ surgery. Results suggest that the high-ploidy subpopulation had a higher Glucose-dependent proliferation rate (0.70 vs. 0.49), but a lower Glucose-dependent death rate (0.47 vs. 1.42). These differences resulted in spatial differences in the distribution of the two subpopulations. Our results contribute to a better understanding of how genomics and microenvironments interact to shape cell fate decisions and could help pave the way to therapeutic strategies that mimic prognostically favorable environments.

Keywords: tumor evolution, intra-tumor heterogeneity, whole-genome doubling, mathematical modeling

Procedia PDF Downloads 73

Authors: Mian Huang, Chi Ma, Junyu Lin, William Lu

Abstract:

Introduction: Bone health attracts more attention recently and an intelligent question and answer (QA) chatbot for osteoporosis is helpful for science popularization. With Generative Pretrained Transformer (GPT) technology developing, we build an osteoporosis corpus dataset and then fine-tune LLaMA, a famous open-source GPT foundation large language model(LLM), on our self-constructed osteoporosis corpus. Evaluated by clinical orthopedic experts, our fine-tuned model outperforms vanilla LLaMA on osteoporosis QA task in Chinese. Three-dimensional quantitative computed tomography (QCT) measured bone mineral density (BMD) is considered as more accurate than DXA for BMD measurement in recent years. We develop an automatic Phantom-less QCT(PL-QCT) that is more efficient for BMD measurement since no need of an external phantom for calibration. Combined with LLM on osteoporosis, our PL-QCT provides efficient and accurate BMD measurement for our chatbot users. Material and Methods: We build an osteoporosis corpus containing about 30,000 Chinese literatures whose titles are related to osteoporosis. The whole process is done automatically, including crawling literatures in .pdf format, localizing text/figure/table region by layout segmentation algorithm and recognizing text by OCR algorithm. We train our model by continuous pre-training with Low-rank Adaptation (LoRA, rank=10) technology to adapt LLaMA-7B model to osteoporosis domain, whose basic principle is to mask the next word in the text and make the model predict that word. The loss function is defined as cross-entropy between the predicted and ground-truth word. Experiment is implemented on single NVIDIA A800 GPU for 15 days. Our automatic PL-QCT BMD measurement adopt AI-associated region-of-interest (ROI) generation algorithm for localizing vertebrae-parallel cylinder in cancellous bone. Due to no phantom for BMD calibration, we calculate ROI BMD by CT-BMD of personal muscle and fat. Results & Discussion: Clinical orthopaedic experts are invited to design 5 osteoporosis questions in Chinese, evaluating performance of vanilla LLaMA and our fine-tuned model. Our model outperforms LLaMA on over 80% of these questions, understanding ‘Expert Consensus on Osteoporosis’, ‘QCT for osteoporosis diagnosis’ and ‘Effect of age on osteoporosis’. Detailed results are shown in appendix. Future work may be done by training a larger LLM on the whole orthopaedics with more high-quality domain data, or a multi-modal GPT combining and understanding X-ray and medical text for orthopaedic computer-aided-diagnosis. However, GPT model gives unexpected outputs sometimes, such as repetitive text or seemingly normal but wrong answer (called ‘hallucination’). Even though GPT give correct answers, it cannot be considered as valid clinical diagnoses instead of clinical doctors. The PL-QCT BMD system provided by Bone’s QCT(Bone’s Technology(Shenzhen) Limited) achieves 0.1448mg/cm2(spine) and 0.0002 mg/cm2(hip) mean absolute error(MAE) and linear correlation coefficient R2=0.9970(spine) and R2=0.9991(hip)(compared to QCT-Pro(Mindways)) on 155 patients in three-center clinical trial in Guangzhou, China. Conclusion: This study builds a Chinese osteoporosis corpus and develops a fine-tuned and domain-adapted LLM as well as a PL-QCT BMD measurement system. Our fine-tuned GPT model shows better capability than LLaMA model on most testing questions on osteoporosis. Combined with our PL-QCT BMD system, we are looking forward to providing science popularization and early morning screening for potential osteoporotic patients.

Keywords: GPT, phantom-less QCT, large language model, osteoporosis

Procedia PDF Downloads 71

Authors: Eunjung Lee

Abstract:

This research investigates the intricate connection between cultural differences and online consumer behaviors by integrating Hofstede's Cultural Dimensions theory with analysis methodologies such as text mining, data mining, and topic analysis. Our aim is to provide a comprehensive understanding of how national cultural differences influence individuals' behaviors when engaging with online reviews. To ensure the relevance of our investigation, we systematically analyze and interpret the cultural nuances influencing online consumer behaviors, especially in the context of online reviews. By anchoring our research in Hofstede's Cultural Dimensions theory, we seek to offer valuable insights for marketers to tailor their strategies based on the cultural preferences of diverse global consumer bases. In our methodology, we employ advanced text mining techniques to extract insights from a diverse range of online reviews gathered globally for a specific product or service like Netflix. This approach allows us to reveal hidden cultural cues in the language used by consumers from various backgrounds. Complementing text mining, data mining techniques are applied to extract meaningful patterns from online review datasets collected from different countries, aiming to unveil underlying structures and gain a deeper understanding of the impact of cultural differences on online consumer behaviors. The study also integrates topic analysis to identify recurring subjects, sentiments, and opinions within online reviews. Marketers can leverage these insights to inform the development of culturally sensitive strategies, enhance target audience segmentation, and refine messaging approaches aligned with cultural preferences. Anchored in Hofstede's Cultural Dimensions theory, our research employs sophisticated methodologies to delve into the intricate relationship between cultural differences and online consumer behaviors. Applied to specific cultural dimensions, such as individualism vs. collectivism, masculinity vs. femininity, uncertainty avoidance, and long-term vs. short-term orientation, the study uncovers nuanced insights. For example, in exploring individualism vs. collectivism, we examine how reviewers from individualistic cultures prioritize personal experiences while those from collectivistic cultures emphasize communal opinions. Similarly, within masculinity vs. femininity, we investigate whether distinct topics align with cultural notions, such as robust features in masculine cultures and user-friendliness in feminine cultures. Examining information-seeking behaviors under uncertainty avoidance reveals how cultures differ in seeking detailed information or providing succinct reviews based on their comfort with ambiguity. Additionally, in assessing long-term vs. short-term orientation, the research explores how cultural focus on enduring benefits or immediate gratification influences reviews. These concrete examples contribute to the theoretical enhancement of Hofstede's Cultural Dimensions theory, providing a detailed understanding of cultural impacts on online consumer behaviors. As online reviews become increasingly crucial in decision-making, this research not only contributes to the academic understanding of cultural influences but also proposes practical recommendations for enhancing online review systems. Marketers can leverage these findings to design targeted and culturally relevant strategies, ultimately enhancing their global marketing effectiveness and optimizing online review systems for maximum impact.

Keywords: comparative analysis, cultural dimensions, marketing intelligence, national culture, online consumer behavior, text mining

Procedia PDF Downloads 47

Authors: Delia Gabriela Moisa, Demos Parapanos, Tim Heap

Abstract:

The tourism sector is mainly comprised of small and medium-sized tourism enterprises (SMEs), representing approximately 80% of global businesses in this field. These entities require focused attention and support to address challenges, ensuring their competitiveness and relevance in a dynamic industry characterized by continuously changing customer preferences. To address these challenges, it becomes imperative to consider not only socio-demographic factors but also delve into the intricate interplay of psychological elements influencing consumer behavior. This study investigates the impact of personality traits and travel motivations on visitor activities in Cumbria, United Kingdom, an iconic region marked by UNESCO World Heritage Sites, including The Lake District National Park and Hadrian's Wall. With a £4.1 billion tourism industry primarily driven by SMEs, Cumbria serves as an ideal setting for examining the relationship between tourist psychology and activities. Employing the Big Five personality model and the Travel Career Pattern motivation theory, this study aims to explain the relationship between psychological factors and tourist activities. The study further explores SME perspectives on personality-based market segmentation, providing strategic insights into addressing evolving tourist preferences.This pioneering mixed-methods study integrates quantitative data from 330 visitor surveys, subsequently complemented by qualitative insights from tourism SME representatives. The findings unveil that socio-demographic factors do not exhibit statistically significant variations in the activities pursued by visitors in Cumbria. However, significant correlations emerge between personality traits and motivations with preferred visitor activities. Open-minded tourists gravitate towards events and cultural activities, while Conscientious individuals favor cultural pursuits. Extraverted tourists lean towards adventurous, recreational, and wellness activities, while Agreeable personalities opt for lake cruises. Interestingly, a contrasting trend emerges as Extraversion increases, leading to a decrease in interest in cultural activities. Similarly, heightened Agreeableness corresponds to a decrease in interest in adventurous activities. Furthermore, travel motivations, including nostalgia and building relationships, drive event participation, while self-improvement and novelty-seeking lead to adventurous activities. Additionally, qualitative insights from tourism SME representatives underscore the value of targeted messaging aligned with visitor personalities for enhancing loyalty and experiences. This study contributes significantly to scholarship through its novel framework, integrating tourist psychology with activities and industry perspectives. The proposed conceptual model holds substantial practical implications for SMEs to formulate personalized offerings, optimize marketing, and strategically allocate resources tailored to tourist personalities. While the focus is on Cumbria, the methodology's universal applicability offers valuable insights for destinations globally seeking a competitive advantage. Future research addressing scale reliability and geographic specificity limitations can further advance knowledge on this critical relationship between visitor psychology, individual preferences, and industry imperatives. Moreover, by extending the investigation to other districts, future studies could draw comparisons and contrasts in the results, providing a more nuanced understanding of the factors influencing visitor psychology and preferences.

Keywords: personality trait, SME, tourist behaviour, tourist motivation, visitor activity

Procedia PDF Downloads 70