How Medical Scanners Track Surgical Tools in Real-Time

This patent describes a method for medical imaging devices to accurately show the real-time 3D position of surgical tools and implants inside a patient during an operation, using a combination of initial 3D scans and live 2D X-ray images.

What it covers

The patent details a method for medical imaging systems to precisely track non-anatomical structures, like surgical instruments, inside a patient during a procedure. It starts with an initial 3D scan (Claim 1) to create a detailed anatomical model. Then, it continuously takes new 2D X-ray images (Claim 1) and uses machine learning to identify both the patient's anatomy and the surgical tools within these images (Claim 10). The system reconstructs a 3D view of the tools from these 2D images, removing visual clutter using machine learning (Claim 1). It then aligns this live 3D view of the tools with the initial detailed anatomical model, creating a "navigation volume" (Claim 1) that shows where the tools are in relation to the patient's body. For example, a surgeon could see the exact 3D path of a guide wire (Claim 14) as it moves through a blood vessel, overlaid on a pre-operative CT scan.

What it doesn't cover

• —Does not cover systems that only use a single initial 3D image without continuous 2D updates to track structures.
• —Does not cover tracking non-anatomical structures without first extracting an anatomical model from an initial 3D image.
• —Does not cover methods that calculate the non-anatomical 3D image without using machine learning to remove artifacts from partial 3D reconstructions.
• —Does not cover systems that track non-anatomical structures without registering the live anatomical 3D image with the initial 3D image to determine a coordinate transformation.
• —Does not cover methods that do not combine the anatomical model and the non-anatomical 3D image into a "navigation volume" using the determined coordinate transformation.

The clever bit

The novelty lies in combining initial high-resolution 3D anatomical data with real-time, lower-resolution 2D images to dynamically track non-anatomical objects in 3D. It specifically uses machine learning to clean up the 3D reconstructions of these objects and accurately register them within the patient's overall anatomy.

Why it matters

This technology is important for image-guided surgery and interventional procedures, where precision is critical. By providing real-time, positionally correct 3D views of instruments like catheters or guide wires (Claim 14) within a patient's anatomy, it can help surgeons navigate complex areas, reduce risks, and improve patient outcomes. It enables more accurate placement of implants and devices.

Real-world examples

• 1.Image-guided neurosurgery systems
• 2.Cardiac catheterization labs
• 3.Orthopedic surgery navigation
• 4.Interventional radiology suites
• 5.C-arm X-ray systems with navigation capabilities