Optimizing incision placement for maximum disc and endplate preparation in lumbar endo-fusion using parametric...

Area of Science:

• Spinal surgery
• Computational anatomy
• Machine learning in medicine

Background:

• Optimizing surgical approach is crucial for uniportal endoscopic lumbar fusion.
• Accurate incision placement maximizes disc and endplate preparation, enhancing surgical outcomes.
• Current methods lack patient-specific optimization for incision placement.

Purpose of the Study:

• To develop and validate computational methods for optimizing incision placement in uniportal endoscopic lumbar fusion.
• To maximize disc and endplate preparation through precise incision planning.
• To create a patient-specific preoperative planning tool.

Main Methods:

• Analysis of 605 lumbar MRI scans from 121 patients and validation with 120 scans from 40 patients.
• Development of an MRI-based anatomical delineation system and genetic algorithm (GA) optimization for incision placement.
• Machine learning, specifically artificial neural networks (ANN), used to predict optimal incision positions based on anatomical features.

Main Results:

• GA identified level-specific optimal incision distances for lumbar fusion levels (e.g., L5-S1: 95.28 ± 19.38 mm).
• ANN model demonstrated high predictive accuracy (R² = 0.935 internal, R² = 0.918 external) for optimal incision placement.
• Mean absolute error (MAE) for external validation was 6.08 mm, indicating precise prediction.

Conclusions:

• A web-based computational tool was developed to optimize incision placement for uniportal endoscopic lumbar fusion.
• This tool aids in maximizing disc and endplate preparation, potentially improving surgical success.
• The approach supports patient-specific preoperative planning for enhanced surgical outcomes.