Artificial intelligence in MRI: Modeling, optimization, and oncological applications from a physicist’s perspective
Session
Medicine and Nursing
Description
The integration of artificial intelligence (AI) into magnetic resonance imaging (MRI) has transformed modern medical diagnostics through its ability to enhance image reconstruction and improve tumor detection and optimize workflow processes. The physicist views AI as an intersection of modeling techniques and signal processing methods and applied mathematical approaches which produce direct clinical benefits. This review evaluates the current state of MRI applications powered by AI by examining research that covers both clinical radiology workflow-based generative AI and specialized deep learning models designed for brain tumor detection. Generative models trained with hospital data have proven their ability to reduce reporting time while maintaining high diagnostic accuracy. The precision of brain tumor segmentation and classification reaches above 95% through the application of deep learning models which include convolutional neural networks (CNN), Vision Transformers (ViT) and Gated Recurrent Units (GRU). The implementation of AI technology leads to better image quality and reduced contrast agent doses and decreased patient safety risks. The implementation of AI faces ongoing difficulties in achieving generalization and maintaining interpretability while handling data heterogeneity. This review paper evaluates these advancements through physical modeling and system optimization perspectives while demonstrating how physicists connect technological advancements to medical practice implementation. The research demonstrates how AI transforms MRI operations while supporting multidisciplinary partnerships to create reliable and ethical systems with transparent explanations for medical imaging applications.
Keywords:
Artificial Intelligence (AI), Magnetic Resonance Imaging (MRI), Deep Learning, Generative Models, Interpretability
Proceedings Editor
Edmond Hajrizi
ISBN
978-9951-982-41-2
Location
UBT Lipjan, Kosovo
Start Date
25-10-2025 9:00 AM
End Date
26-10-2025 6:00 PM
DOI
10.33107/ubt-ic.2025.360
Recommended Citation
Qafleshi, Mevlan and Grantolli, Altin, "Artificial intelligence in MRI: Modeling, optimization, and oncological applications from a physicist’s perspective" (2025). UBT International Conference. 5.
https://knowledgecenter.ubt-uni.net/conference/2025UBTIC/MN/5
Artificial intelligence in MRI: Modeling, optimization, and oncological applications from a physicist’s perspective
UBT Lipjan, Kosovo
The integration of artificial intelligence (AI) into magnetic resonance imaging (MRI) has transformed modern medical diagnostics through its ability to enhance image reconstruction and improve tumor detection and optimize workflow processes. The physicist views AI as an intersection of modeling techniques and signal processing methods and applied mathematical approaches which produce direct clinical benefits. This review evaluates the current state of MRI applications powered by AI by examining research that covers both clinical radiology workflow-based generative AI and specialized deep learning models designed for brain tumor detection. Generative models trained with hospital data have proven their ability to reduce reporting time while maintaining high diagnostic accuracy. The precision of brain tumor segmentation and classification reaches above 95% through the application of deep learning models which include convolutional neural networks (CNN), Vision Transformers (ViT) and Gated Recurrent Units (GRU). The implementation of AI technology leads to better image quality and reduced contrast agent doses and decreased patient safety risks. The implementation of AI faces ongoing difficulties in achieving generalization and maintaining interpretability while handling data heterogeneity. This review paper evaluates these advancements through physical modeling and system optimization perspectives while demonstrating how physicists connect technological advancements to medical practice implementation. The research demonstrates how AI transforms MRI operations while supporting multidisciplinary partnerships to create reliable and ethical systems with transparent explanations for medical imaging applications.