Enhancing Knee MR Image Clarity through Image Domain Super-Resolution Reconstruction

Area of Science:

• Medical Imaging
• Image Processing
• Biomedical Engineering

Background:

• Clinical magnetic resonance imaging (MRI) often faces limitations in spatial resolution.
• Enhancing MRI resolution typically requires expensive hardware upgrades.
• There is a need for cost-effective methods to improve MRI scan quality.

Purpose of the Study:

• To introduce a hybrid analytical super-resolution (SR) pipeline for enhancing medical MRI resolution.
• To overcome current clinical MRI resolution limitations without additional hardware.
• To improve diagnostic accuracy and anatomical modeling capabilities.

Main Methods:

• A three-step hybrid pipeline: pre-processing (re-slicing, registration), SR reconstruction (combining orthogonal stacks), and post-processing (artefact reduction convolutional neural network - ARCNN).
• Validation on six high-resolution knee MRI datasets using various sequences.
• Quantitative and qualitative assessments of image quality and resolution enhancement.

Main Results:

• The SR pipeline demonstrated an average mean error of 1.40 ± 2.22% in voxel intensities compared to original high-resolution images.
• Qualitative analysis showed improved out-of-plane resolution while maintaining in-plane image quality.
• The method proved robust across different MRI sequences.

Conclusions:

• The hybrid SR pipeline effectively enhances medical MRI resolution without hardware costs.
• This approach offers a viable alternative for improving diagnostic accuracy and creating more accurate anatomical models.
• The method shows significant potential for clinical applications, particularly in orthopaedics.