A Computational Framework for Automated Puncture Trajectory Planning in Hemorrhagic Stroke Surgery

Area of Science:

• Neurosurgery
• Medical Imaging
• Artificial Intelligence

Background:

• Hemorrhagic stroke treatment often involves hematoma drainage via puncture.
• Optimal puncture targeting and trajectory are critical for successful outcomes.
• Current methods lack automated planning, impacting efficiency and safety.

Purpose of the Study:

• To develop an AI-driven computational framework for automated surgical puncture path generation.
• To enhance the precision and safety of trajectory planning for hemorrhagic stroke treatment.
• To integrate advanced AI techniques for optimizing surgical approaches.

Main Methods:

• Developed the Puncture Trajectory ToolKits (PTK) software using C++/Python.
• Implemented hybrid segmentation (ResNet-50 deep learning + adaptive thresholding) for hematoma detection.
• Utilized PCA for axis extraction, Laplacian smoothing, skull quadrant theory, and safety corridor modeling.

Main Results:

• The AI framework demonstrated high accuracy in puncture trajectory planning.
• Optimized paths achieved a mean surgeon satisfaction score of 4.4/5.
• Average angle difference between automated and manual paths was 16.36°, indicating improved precision.

Conclusions:

• PTK offers a systematic pipeline for trajectory planning, outperforming conventional methods.
• The framework shows technical superiority and high clinical applicability for surgeons.
• Further integration with robotics and clinical trials are recommended for PTK.