A Tool-free Neuronavigation Method based on Single-view Hand Tracking.

Area of Science:

• Neurosurgery
• Medical Imaging
• Computer Vision

Background:

• Standard neuronavigation systems are expensive, require extensive setup, and are often impractical for bedside use.
• Existing freehand methods for craniotomy placement can be error-prone and lack precision.
• There is a need for accessible, intuitive, and accurate neuronavigation solutions in diverse clinical environments.

Purpose of the Study:

• To present a novel, tool-free neuronavigation method utilizing a single RGB camera.
• To demonstrate a cost-effective and user-friendly alternative to conventional neuronavigation platforms.
• To enable accurate 3D-3D registration for intra-operative guidance.

Main Methods:

• Development of a neuronavigation system employing machine-learning-based hand pose estimation.
• The hand pose estimation acts as a proxy for optical tool tracking, eliminating the need for specialized hardware.
• Implementation of a 3D-3D registration process for pre-operative to intra-operative data alignment.

Main Results:

• The system achieves an average Target Registration Error (TRE) of 1.3cm.
• Qualitative feedback from clinical users indicated high clinical relevance.
• The framework-agnostic design allows for direct integration of future advancements in hand-tracking technology.

Conclusions:

• The proposed tool-free neuronavigation method is a clinically relevant, cost-effective, and accessible solution.
• It offers improved intuitiveness and reduced error compared to freehand techniques.
• The system's minimal setup requirements make it suitable for bedside procedures and resource-limited settings.