Diagnostically competitive performance of a physiology-informed generative multi-task network for contrast-free CT...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neurology

Background:

• Perfusion imaging assesses hemodynamic status and tissue perfusion.
• Computed tomography perfusion (CTP) is vital for stroke assessment but faces cost, accessibility, and contrast agent limitations.
• Contrast agents can cause allergic reactions and adverse side effects.

Purpose of the Study:

• To introduce Multitask Automated Generation of Intermodal CT perfusion maps (MAGIC), a deep learning framework.
• To develop a contrast-free method for generating CTP maps from non-contrast CT imaging.
• To enhance image fidelity by integrating physiological information.

Main Methods:

• Developed a novel deep learning framework (MAGIC) combining generative AI and physiological data.
• Mapped non-contrast CT imaging to multiple contrast-free CTP maps.
• Incorporated physiological characteristics into loss terms to improve image fidelity.
• Trained and validated the network on stroke patient CT data.

Main Results:

• Demonstrated robustness to abnormalities in brain perfusion.
• A double-blinded study with neuroradiologists validated MAGIC's visual quality and diagnostic accuracy.
• MAGIC showed favorable performance compared to traditional contrast-enhanced CTP.

Conclusions:

• MAGIC offers a promising contrast-free, cost-effective, and rapid solution for perfusion imaging.
• This technology has the potential to revolutionize healthcare, particularly in stroke management.
• The framework enhances diagnostic capabilities while mitigating risks associated with contrast agents.