The Current Status of AI-accelerated MRI Techniques in Clinical Use

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Artificial intelligence (AI) tools for magnetic resonance imaging (MRI) acceleration are increasingly integrated into clinical practice.
• Existing MRI acceleration techniques include compressed sensing and parallel imaging.
• AI introduces novel opportunities and complexities when combined with these established methods.

Purpose of the Study:

• To review commercially available AI tools for clinical MRI acceleration.
• To discuss the fundamental principles, advantages, and potential pitfalls of AI-accelerated MRI.
• To explore the downstream implications and logistical challenges of implementing AI in routine MRI workflows.

Main Methods:

• Review of fundamental technical aspects of AI acceleration tools, including noise reduction and super-resolution reconstruction.
• Summary of available evidence on AI-accelerated MRI performance.
• Discussion of potential errors, such as hallucinations, and their clinical relevance.

Main Results:

• AI-accelerated MRI shortens acquisition time, recovering image quality using AI algorithms.
• Potential benefits include enhanced patient comfort, reduced waiting times, and fewer motion artifacts.
• Concerns regarding "hallucinations" (e.g., false lesions) exist but appear rare at recommended clinical acceleration levels.

Conclusions:

• AI-accelerated MRI offers significant advantages but requires careful investigation of potential errors and downstream impacts.
• Challenges include managing increased data volume, potential biases in image analysis, and determining optimal acceleration levels.
• Establishing clear assessment criteria and addressing medicolegal aspects are crucial for the responsible clinical implementation of AI in MRI.