Intelligent Evaluation of Global Spinal Alignment by a Decentralized Convolutional Neural Network

Area of Science:

• Orthopedics and Biomechanics
• Medical Imaging Analysis
• Artificial Intelligence in Healthcare

Background:

• Spinal misalignment, caused by degenerative changes, leads to pain and mobility issues.
• Accurate measurement of spinal alignment parameters (e.g., pelvic incidence, pelvic tilt, lumbar lordosis) is crucial for diagnosis and surgical planning.
• Manual measurement of these parameters is time-consuming and requires expert interpretation.

Purpose of the Study:

• To develop and validate an automated method for precisely measuring 12 key spinal alignment parameters.
• To leverage artificial intelligence, specifically a decentralized convolutional neural network, for enhanced accuracy and efficiency in spinal assessment.

Main Methods:

• Implementation of a decentralized convolutional neural network for spinal parameter measurement.
• Utilizing region of interest detection with dimension reduction to focus on critical anatomical areas.
• Calculation of spinal alignment parameters based on detected key points.

Main Results:

• The AI method accurately measures 12 spinal alignment parameters.
• 10 out of 12 parameters demonstrated a correlation coefficient > 0.8 compared to manual measurements.
• Pelvic tilt measurement showed a minimal absolute deviation of 1.156°.

Conclusions:

• The developed decentralized convolutional neural network offers a precise and automated solution for measuring spinal alignment parameters.
• This AI-driven approach has the potential to significantly improve the efficiency and consistency of spinal diagnosis and surgical planning.