3D Segmentation for medical images (OsteoVision) = التقطيع ثلاثي الأبعاد للصور الطبية
With the increasing integration of AI across various sectors, artificial intelligence (AI) is already playing a significant role in the healthcare industry, and its use is expected to grow further. AI systems used in image processing and computer vision algorithms have shown a significant ability to perform many operations such as segmentation, classification, and detection. This project presents the application of computer vision algorithms in the field of medical imaging for diagnostic, therapeutic, and interventional purposes. This thesis explores the use of several computer vision algorithms to address different pathologies, specifically brain tumors (glioma) (see Appendix A) and knee osteoarthritis (OA), as well as tracking the progression of knee osteoarthritis using the Kellgren and Lawrence (KL) grading system, a common method for classifying the severity of OA into five grades. To achieve the desired impact, the project employs various techniques, including 3D segmentation for brain tumors, 2D segmentation for knee joints, and multinomial classification for determining the severity of knee OA injuries. The primary aims of the project are to enhance diagnostic accuracy, assist in creating treatment plans, provide an assistive tool for healthcare providers to make more informed decisions, leverage AI's capabilities to detect abnormalities that might escape the human eye, and streamline workflow. To facilitate these goals, the project incorporates a user-friendly UI, a website, and a Flutter-based mobile application, enabling healthcare providers to efficiently integrate these tools into their practice and improve patient care.
Related Books
AI in drug discovery
Constitutes the refereed proceedings of the First international workshop on ai in Drug Discovery, AIDD 2024, held as a part of the 33rd International Conference on Artificial Neural Networks, ICANN 2024, in Lugano, Switzerland, on September 19, 2024. These papers focus on various aspects of the rapidly evolving field of Artificial Intelligence (AI)-driven drug discovery in chemistry, including Big Data and advanced Machine Learning, eXplainable AI (XAI), Chemoinformatics, Use of deep learning to predict molecular properties, Modeling and prediction of chemical reaction data and Generative models.
AI in banking : Practical applications and case studies
Delves into the application of AI from theory to practice, offering detailed insights into AI project design and code implementation across eleven business scenarios in four major sectors: retail banking, e-banking, bank credit, and tech operations. it provides hands-on examples of various technologies, including automatic machine learning, integrated learning, graph computation, recommendation systems, causal inference, generative adversarial networks, supervised learning, unsupervised learning, computer vision, reinforcement learning, fuzzy control, automatic control, speech recognition, semantic understanding, bayesian networks, edge computing, and more. this book stands as a rare and practical guide to AI projects in the banking industry.
Machine learning for cyber-physical systems: selected papers from the international conference ML4CPS 2023
Contains selected papers from the international conference ML4CPS – Machine Learning for Cyber-Physical Systems, which was held in Hamburg (Germany), from 29 to 31 March 2023. Cyber-physical systems are adaptive and learning: they analyze their environment and, based on observations, learn patterns, associations and predictive models. Typical applications are condition monitoring, predictive maintenance, image processing and diagnostics. Machine learning is the key technology for these developments.
Layce (Image data poisoning) = لايس (تسميم بيانات الصور )
The ongoing growth of image generative artificial intelligence models was paved with existing drawings and art pieces by great artists both past and present, and while generative models are very useful and helpful, there is the issue of the origin of the datasets trained on, and the morality of usage regarding copyrights and artistic identity. A novel line of defense that helps artists and visual content creators actively protect their pieces emerged, dubbed Data Poisoning and it works by misleading Artificial Intelligence models that attempt to use a Poisoned Image for training, or as a reference, as the Poisoned Image will appear to the human eye identical to the original art piece, while the Artificial Intelligence model will be seeing a remarkably different image, causing generative models to generate false positive results when given a prompt poisoned by the author or when trained on data poisoned by the original owner. This study aims to study image data poisoning methods and technologies, and build an application containing multiple image models, and poisoning models as well, accompanied by a Community for artists to share art and interact with each other.
