Advancements in Material Decomposition Techniques for Spectral CT: A Comparative Analysis of Traditional and AI-Driven Methods
Year: 2025
Type: article
Abstract: Material Decomposition (MD) in Computed Tomography (CT) imaging is undergoing a significant transformation as it shifts towards the integration of Machine Learning (ML) and Artificial Intelligence (AI) frameworks. The combination of spectral (multi-energy) CT and AI-based MD techniques have the potential to enhance the management of MD-related tasks in terms of diagnostic efficiency, accuracy, and early diagnosis at high spatial resolution. Despite the challenges posed by patient exposure to ionizing radiation, image noise, and image distortions caused by artifacts, AI-based medical diagnostic techniques show great promise for the development of intelligent MD systems. In this article, we provide a comprehensive and categorized review and analysis of traditional MD and modern AI-based MD techniques for both the image and projection domains. We highlight each approach’s distinctive features, advantages, and disadvantages, as well as the intrinsic features and critical attributes of the AI-based MD techniques. Finally, we conclude by presenting key observations and laying a strong foundation for future research to design and implement AI-based MD techniques. Our aim is to provide a comprehensive understanding of the current state of the field and to encourage further exploration of this exciting and rapidly evolving area.
Source: IEEE Access
Authors: Naveed Ilyas, Farhat Naseer, Osama Sikander Khan, Hamza Ilyas, Osama Bin Yaqoob, Abderaouf Behouch, Saiqa Sagheer, Moid Sandhu, Mohsin Raza Jafri, Aamir Younas Raja
Institutions: University of California, Riverside, National University of Sciences and Technology, Khalifa University of Science and Technology, Air University, National University of Computer and Emerging Sciences, University Ferhat Abbas of Setif, Commonwealth Scientific and Industrial Research Organisation, Australian e-Health Research Centre
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,075
OpenAlex ID: W4414693468