Deep learning-enhanced zero echo time MRI for femoral cortical bone quantification: A technical validation against CT

Area of Science:

• Radiology and Medical Imaging
• Orthopedics
• Biomedical Engineering

Background:

• Femoroacetabular impingement (FAI) is a hip condition linked to early osteoarthritis.
• Accurate quantitative bone measurements are crucial for FAI diagnosis and management.
• Computed tomography (CT) is a standard for bone imaging but involves radiation exposure.

Purpose of the Study:

• To compare quantitative femoral cortical bone measurements from zero echo time MRI (oZTEo) with CT.
• To evaluate the performance of deep learning (DL) reconstructions for oZTEo images.
• To assess oZTEo-DL as a potential radiation-free alternative for bone imaging in FAI patients.

Main Methods:

• 30 FAI patients underwent simultaneous MRI and CT scans.
• oZTEo MRI images were processed using DL for denoising, chemical shift correction (CSC), and background removal (BR).
• Image analysis included cortical edge location, sharpness, signal intensity, and thickness, with statistical comparisons (ANOVA, ICC).

Main Results:

• Cortical edge location and sharpness were consistent across CT, oZTEo, oZTEo-CSC, and oZTEo-CSC-BR.
• oZTEo-CSC and oZTEo-CSC-BR showed greater agreement with CT for bone assessment compared to standard oZTEo.
• DL reconstructions (oZTEo-CSC, oZTEo-CSC-BR) demonstrated higher inter-examiner repeatability (ICC ≈ 0.68) than oZTEo and CT.

Conclusions:

• oZTEo MRI with DL post-processing (CSC and BR) provides a CT-like assessment of femoral cortical bone.
• These methods offer a radiation-free alternative for bone imaging in FAI patients.
• DL-enhanced oZTEo improves measurement consistency and CT agreement, with clinically negligible differences.