Browsing by Author "Rokne, Jon George"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Offline and Online Interactive Frameworks for MRI and CT Image Analysis in the Healthcare Domain : The Case of COVID-19, Brain Tumors and Pancreatic Tumors
    (2023-08) Sailunaz, Kashfia; Alhajj, Reda S.; Alhajj, Reda S.; Rokne, Jon George; Ozyer, Tansel; Kawash, Jalal Yusef; Agarwal, Nitin
    Medical imaging represents the organs, tissues and structures underneath the outer layers of skin and bones etc. and stores information on normal anatomical structures for abnormality detection and diagnosis. In this thesis, tools and techniques are used to automate the analysis of medical images, emphasizing the detection of brain tumor anomalies from brain MRIs, Covid infections from lung CT images and pancreatic tumor from pancreatic CT images. Image processing methods such as filtering and thresholding models, geometry models, graph models, region-based analysis, connected component analysis, machine learning models, and recent deep learning models are used. The following problems for medical images : abnormality detection, abnormal region segmentation, interactive user interface to represent the results of detection and segmentation while receiving feedbacks from healthcare professionals to improve the analysis procedure, and finally report generation, are considered in this research. Complete interactive systems containing conventional models, machine learning, and deep learning methods for different types of medical abnormalities have been proposed and developed in this thesis. The experimental results show promising outcomes that has led to the incorporation of the methods for the proposed solutions based on the observations of the performance metrics and their comparisons. Although currently separate systems have been developed for brain tumor, Covid and pancreatic cancer, the success of the developed systems show a promising potential to combine them to form a generalized system for analyzing medical imaging of different types collected from any organs to detect any type of abnormalities.
  • Thumbnail Image
    ItemOpen Access
    One Network to Rule Them All: Three Essays on Behaviors in Social Networks
    (2022-05) Eslami, Mojtaba; Tombe, Trevor; Rokne, Jon George; Oxoby, Robert; Sezer, Deniz; Kattan, Lina; Agarwal, Nitin
    This thesis is an integration of research contributions on social networks and economic behaviors. The central recurring theme is showing how network structures affect individuals’ decision-making using game-theoretic approaches. Social variables such as friendship ties, and individuals’ social positions affect others’ behaviors. These variables are used in an interdisciplinary setting drawing insights from fields such as economics, social psychology, computer science, and mathematics. My aim is to highlight the importance of the very structures of social networks whose characteristics influence individuals’ behaviors. It is hard to discern these characteristics but the social structures can be mined to elicit such structures as individuals actively interact with them. The connections among individuals force us to ponder the consequences of actions in a broader sense. Although the social network paradigm doesn’t allow singled-out strategies they provide a holistic overview of social problems in general. Social networks break the symmetry of interactions among individuals. They allow more realistic standpoints for investigating some puzzling behavioral phenomena. Network economics shows the roles of individuals are underestimated in many mainstream models. We are small, but our measure is definitely not zero. Though individuals may seem to only communicate within their local circles, their interactions reverberate far and beyond affecting distant unknown individuals. Local behaviors do not necessarily translate into intended aggregate results. This is how the network structure interacts with individuals in an indiscernible way. In the first of the three research contributions, with the help of a game defined in an open-source software ecosystem, I structurally modified the network topology in order to manipulate the flow of the peer pressure on equilibrium. The second contribution illustrates that behavioral phenomena, such as pluralistic ignorance and the friendship paradox, can be explained by the help of the network paradigm. I proposed an effective strategy to create a significant change by targeting only a few carefully selected individuals. The third contribution focuses on the behavioral aspect of communications when conducted in a social network. I explained why people don’t say what they truly believe and how they would be able to balance their extreme thoughts through conversations.
  • Thumbnail Image
    ItemOpen Access
    Social Media Emergency Analysis and Realistic Evacuation Modeling
    (2021-09-07) Sahin, Coskun; Alhajj, Reda; Alhajj, Reda; Ozyer, Tansel; Rokne, Jon George; Ruhe, Guenther; Mouhoub, Malek
    The widespread of disasters necessitates appropriate actions to be taken to avoid casualties or at least reduce them to the minimum level possible. In this work, we propose solutions to two of the ways we can help it. The first part is using the potential of social media to detect emergencies and provide further information for the public and rescue teams. It uses a multi-layer machine learning approach to find and cluster emergency-related messages. Natural language processing and information extraction techniques are adopted for location detection, casualty and severity calculation. The effectiveness of the model is shown on Twitter data in nearly real-time using its own API. The second part of the thesis investigates the area of crowd behavior modeling in order to provide a platform for simulating various emergency evacuation scenarios. Crowd behavior modeling is an important challenge especially for games, social simulation and military training software. There are various applications focusing on specific approaches to solve social, physical, emotional and cognitive dimensions of this problem. In this work, first, we built and agent-based framework that adopts OCC emotion model and Belief-Desire-Intention (BDI) approach in a heterogeneous environment containing individuals with different knowledge of the surroundings. It uses a deep Q-learning model running on top of a neural network for creating a set of partially-trained agents. Second, we simulate group decision-making process using a mathematical ferromagnetism model and analyze how it affects the overall success of the crowd achieving a goal together. Our work contributes to the literature in two different ways. Firstly, it shows that a multi-layer classification model is effective for detecting emergencies in real-time using unstructured data. Moreover, it provides evacuation simulation scenarios on the public buildings where an emergency occurs and saves valuable time for rescue teams while planning an emergency response. Secondly, it combines state-of-the-art frameworks for crowd behavior modeling with a partial learning mechanism. Our experiments show that the system is capable of simulating common group patterns during emergencies. Moreover, it provides a generic crowd framework, where partially-trained agents can find optimal solutions via interaction and collaboration.