Enhanced bone assessment of the shoulder using zero-echo time MRI with deep-learning image reconstruction

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Zero echo-time (ZTE) MRI is valuable for shoulder imaging, but osseous detail can be limited.
• Standard-of-care (SOC) reconstruction methods may not fully optimize image quality for bone assessment.

Purpose of the Study:

• To evaluate a deep learning-based reconstruction algorithm (DLRecon) for ZTE shoulder MRI.
• To determine if DLRecon improves the delineation of osseous findings compared to SOC.

Main Methods:

• Retrospective analysis of 63 shoulder ZTE MRI exams (1.5 Tesla) from 52 patients.
• Raw data reconstructed using both SOC and a prototype DLRecon algorithm.
• Qualitative (image quality, bone delineation, confidence) and quantitative (SNR, CNR) assessments by two blinded readers.

Main Results:

• DLRecon demonstrated significantly higher image quality scores than SOC (p < 0.03), excluding artifacts.
• Superior qualitative scores for osseous pathology delineation and diagnostic confidence with DLRecon (p ≤ 0.03).
• DLRecon achieved significantly better bone CNR (95% CI [1.4-3.1]) and SNR (95% CI [15.3-21.5]) (p < 0.001).

Conclusions:

• DLRecon enhances image quality in ZTE shoulder MRI.
• Improved delineation of osseous pathologies and increased diagnostic confidence in bone assessment.
• DLRecon shows promise for routine clinical use in shoulder MRI.