Intraoperative evaluation of device placement in spine surgery using known-component 3D-2D image registration

Area of Science:

• Medical Imaging
• Surgical Navigation
• Image Registration

Background:

• Intraoperative radiography/fluoroscopy is crucial for assessing surgical device placement, such as spine pedicle screws.
• Qualitative interpretation of these images can be unreliable in detecting suboptimal placement or breaches of critical structures.
• There is a need for quantitative methods to enhance the quality assurance (QA) of surgical products during operations.

Purpose of the Study:

• To present a novel 3D-2D image registration method for quantitative assessment of surgical device placement.
• To leverage intraoperative radiographs with prior patient and component knowledge for improved QA.
• To enable simultaneous, single-step QA for complex spinal constructs.

Main Methods:

• Developed a known-component registration (KC-Reg) algorithm using patient-specific preoperative CT and parametric component models.
• The algorithm optimizes gradient similarity and removes the need for offline C-arm calibration.
• KC-Reg simultaneously solves for multiple component bodies, facilitating single-step QA.

Main Results:

• In a spine phantom, KC-Reg achieved mean target registration error (TRE) of 1.1 ± 0.1 mm (tip) and 0.7 ± 0.4° (angle) with calibration.
• The calibration-free formulation achieved TRE of 1.4 ± 0.6 mm.
• Simultaneous solutions showed statistically significant improvement over sequential methods; clinical application yielded TRE of 2.7 ± 2.6 mm and 1.5 ± 0.8°.

Conclusions:

• The KC-Reg algorithm provides an independent, quantitative QA of surgical products using standard intraoperative radiographic views.
• This intraoperative assessment can enhance surgical quality and safety.
• The method offers the potential for intraoperative revision of suboptimal constructs, thereby reducing revision surgery rates.