Generating a Synthetic Lumbar CT from a Standard MRI Protocol.

Area of Science:

• Medical Imaging
• Radiology
• Neurosurgery

Background:

• Preoperative spine CT scans pose financial burdens and radiation risks.
• Synthetic CT (sCT) from MRI offers a potential alternative for spine surgery planning.
• sCT aims to provide accurate bone structure visualization without additional scans.

Purpose of the Study:

• To evaluate the accuracy of sCT generated from standard MRI protocols.
• To compare sCT bone structure representation with conventional CT scans.
• To assess the clinical relevance of sCT for spine surgery.

Main Methods:

• A neural network architecture was employed to convert MRI data into sCT images.
• sCT images achieved sub-millimeter resolution, surpassing standard MRI.
• Both 3D and 2D measurements were used to compare sCT with conventional CT on 500 vertebrae.

Main Results:

• 3D surface distance measurements assessed overall vertebrae and key anatomical elements.
• 2D analysis focused on critical distances within the vertebral body, pedicles, and spinous processes.
• Automated measurements were validated by orthopedic surgeons for clinical accuracy.

Conclusions:

• sCT generated from standard MRI protocols provides a precise representation of spinal bone structures.
• This technique holds promise for enhancing preoperative planning and navigation in spine surgery.
• The elimination of preoperative CT scans can lead to significant cost savings and reduced patient radiation exposure.