Comparison of 3-D mandibular surfaces generated by MRI and CT

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Orthodontics

Background:

• Three-dimensional (3D) reconstruction of mandibular surfaces is crucial for diagnosing and planning orthognathic surgery.
• Computed tomography (CT) is considered the gold standard for such reconstructions, but magnetic resonance imaging (MRI) offers advantages like reduced radiation exposure.

Purpose of the Study:

• To assess the accuracy of 3D mandibular surfaces generated from MRI data.
• To quantify errors in MRI-derived surfaces by comparing them against CT-based gold standard data.

Main Methods:

• Seven patients with orthognathic deformities underwent both CT and MRI scans.
• 3D mandibular surfaces were generated using surface rendering from both imaging modalities.
• Statistical analysis included Bland-Altman plots (limits of agreement), minimal detectable change (MDC95), and intra-class correlation coefficients (ICC) to evaluate intra-individual reliability.

Main Results:

• The average total error between CT and MRI surfaces was 1.6 mm.
• The coronoid region exhibited the highest random error (MDC95).
• Other regions showed acceptable MDC95 values (e.g., <1.8 mm in transverse, <1.7 mm in anteroposterior directions).
• Intra-class correlation coefficients indicated 'almost perfect' agreement across all evaluated regions.

Conclusions:

• Random errors in 3D mandibular surface generation from MRI were quantified.
• Despite higher errors in the coronoid region, MRI-derived mandibular surfaces are suitable for evaluation in other areas.
• MRI presents a viable, lower-radiation alternative for 3D mandibular surface reconstruction in orthognathic surgery planning.