Multi-dimensional perceptual quality assessment for magnetic resonance images

Area of Science:

• Medical Imaging
• Computer Vision
• Artificial Intelligence

Background:

• Assessing Magnetic Resonance Image (MRI) quality is crucial but challenging.
• Existing image quality assessment (IQA) methods, designed for natural images, lack the specificity for MRI's fine-grained characteristics.
• This gap hinders accurate diagnostic interpretation and quality control in clinical settings.

Purpose of the Study:

• To develop a specialized model for objective and accurate perceptual quality assessment of MRI.
• To provide multi-dimensional quality scores for MRI, offering deeper insights than general IQA methods.
• To establish a large-scale, expert-annotated dataset for MRI quality preference.

Main Methods:

• Creation of a large-scale, expert-annotated dataset of MRI images with perceptual quality preferences.
• Development of MPQA, a multi-dimensional perceptual quality assessment model tailored for MRI.
• Validation of MPQA against existing IQA baselines using various evaluation metrics.

Main Results:

• The proposed MPQA model achieved superior performance compared to existing methods in MRI quality assessment tasks.
• MPQA demonstrated strong accuracy, robustness, and generalizability across different MRI evaluation scenarios.
• The model provides detailed insights into factors influencing MRI image quality perception.

Conclusions:

• MPQA offers a significant advancement in MRI image quality assessment, addressing limitations of general IQA approaches.
• The model enhances the reliability of MRI interpretation by providing accurate, multi-dimensional quality scores.
• This work facilitates more objective and precise evaluation of MRI quality in clinical and research applications.