Extraction of 3D trajectories of mandibular condyles from 2D real-time MRI

Overview

• This study aimed to investigate the feasibility of extracting 3D condylar trajectories (the movement of the jaw joint) from 2D real-time MRI scans.
• The researchers analyzed real-time MRI data from 20 healthy subjects while they opened and closed their jaws.
• They used a machine learning algorithm to segment the images and track the movement of the jaw joints.
• The study evaluated the quality and precision of the computed trajectories.

Plain English Explanation

The researchers wanted to see if they could use MRI scans to study how the jaw moves during opening and closing. This could provide a more comprehensive examination of the jaw joint compared to other methods.

They had 20 healthy people undergo real-time MRI scans while opening and closing their mouths. A computer algorithm analyzed the scans to identify and track the movement of the jaw joints (called condyles).

The researchers then evaluated the quality and accuracy of the trajectories computed from the MRI data. They looked at factors like how well the movements were reproduced, how much the head moved during the scans, and whether the left and right sides of the jaw moved symmetrically.

The segmentation (identification) of the jaw joints in the axial [https://aimodels.fyi/papers/arxiv/accurate-patient-alignment-without-unnecessary-imaging-dose] MRI slices was generally good, but the sagittal slices required some refinement. The movement reproducibility was acceptable for most cases, but head motion caused some displacement of the trajectories.

Overall, the researchers found that real-time MRI can provide reasonably accurate information about jaw joint movement, which could be useful for clinicians evaluating conditions related to the jaw and bite.

Technical Explanation

The researchers used a U-Net-based [https://aimodels.fyi/papers/arxiv/sismik-brain-mri-deep-learning-based-motion] segmentation algorithm to identify the jaw joints (condyles) in one axial and two sagittal MRI slices collected from the 20 subjects. They then projected the centers of mass of the segmented regions onto a coordinate system based on anatomical markers and temporally adjusted the data.

The quality of the computed 3D condylar trajectories was evaluated using several metrics:

1. Movement reproducibility: Assessed how consistently the jaw movements were captured across multiple trials.
2. Head motion: Quantified how much the subject's head moved during the scans, which can displace the trajectory data.
3. Slice placement symmetry: Measured the difference in the superior-inferior (up-down) position of the left and right condyles when the jaw was closed.

The segmentation of the axial slices demonstrated good-to-excellent quality, but the sagittal slices required some manual fine-tuning. The movement reproducibility was acceptable for most cases, but head motion displaced the trajectories by around 1 mm on average. The difference in the superior-inferior position of the left and right condyles in the closed jaw position was 1.7 mm on average.

Critical Analysis

While the researchers demonstrated the feasibility of extracting 3D condylar trajectories from real-time MRI, the precision of the results was limited by factors like head motion and imperfect segmentation of the sagittal slices.

The researchers acknowledge that further work is needed to improve the accuracy and robustness of the trajectory extraction, such as by incorporating additional anatomical landmarks [https://aimodels.fyi/papers/arxiv/leveraging-digital-perceptual-technologies-remote-perception-analysis] or developing more advanced segmentation techniques [https://aimodels.fyi/papers/arxiv/rapidvol-rapid-reconstruction-3d-ultrasound-volumes-from].

Additionally, the study was conducted on a relatively small sample of healthy individuals. Applying this approach to patients with jaw disorders or abnormalities may introduce additional challenges that were not addressed here.

Conclusion

Despite the limitations in precision, this study demonstrates the potential of using real-time MRI to extract 3D condylar trajectories, which could provide valuable information for clinicians evaluating jaw function and disorders. With further refinements to the methodology, this approach could become a useful tool for comprehensive assessments of the temporomandibular joint.